intel-hard/intel-hard-lab
2
0 Fork 0
mirror of https://github.com/elwrcl/intel-hard-lab.git synced 2026-10-31 07:48:02 +03:00
Code Issues Packages Projects Releases Wiki Activity
Files
0bbe6b34e64751bc6f779829a9e5038cb0924844
intel-hard-lab/docs/ivb_2012/txt/ivb_ihd_os_vol1_part3.txt
T
elwrcl 0bbe6b34e6 first commit
2026-08-23 04:59:41 +03:00

10018 lines
498 KiB
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
Intel® OpenSource HD Graphics
Programmer’s Reference Manual (PRM)
Volume 1 Part 3: Graphics Core™ – Memory
Interface and Commands for the Render Engine
(Ivy Bridge)
For the 2012 Intel® Core™ Processor Family
May 2012
Revision 1.0
NOTICE:
This document contains information on products in the design phase of development, and Intel
reserves the right to add or remove product features at any time, with or without changes to this
open source documentation.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12
Creative Commons License
You are free to Share — to copy, distribute, display, and perform the work
Under the following conditions:
Attribution. You must attribute the work in the manner specified by the author or licensor (but not in any
way that suggests that they endorse you or your use of the work).
No Derivative Works. You may not alter, transform, or build upon this work.
®
INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL PRODUCTS. NO
LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL
PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL'S
TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY
WHATSOEVER AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO
SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO
FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY
PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.
A "Mission Critical Application" is any application in which failure of the Intel Product could result, directly
or indirectly, in personal injury or death. SHOULD YOU PURCHASE OR USE INTEL'S PRODUCTS FOR
ANY SUCH MISSION CRITICAL APPLICATION, YOU SHALL INDEMNIFY AND HOLD INTEL AND ITS
SUBSIDIARIES, SUBCONTRACTORS AND AFFILIATES, AND THE DIRECTORS, OFFICERS, AND
EMPLOYEES OF EACH, HARMLESS AGAINST ALL CLAIMS COSTS, DAMAGES, AND EXPENSES
AND REASONABLE ATTORNEYS' FEES ARISING OUT OF, DIRECTLY OR INDIRECTLY, ANY CLAIM
OF PRODUCT LIABILITY, PERSONAL INJURY, OR DEATH ARISING IN ANY WAY OUT OF SUCH
MISSION CRITICAL APPLICATION, WHETHER OR NOT INTEL OR ITS SUBCONTRACTOR WAS
NEGLIGENT IN THE DESIGN, MANUFACTURE, OR WARNING OF THE INTEL PRODUCT OR ANY
OF ITS PARTS.
Intel may make changes to specifications and product descriptions at any time, without notice. Designers
must not rely on the absence or characteristics of any features or instructions marked "reserved" or
"undefined". Intel reserves these for future definition and shall have no responsibility whatsoever for
conflicts or incompatibilities arising from future changes to them. The information here is subject to
change without notice. Do not finalize a design with this information.
The products described in this document may contain design defects or errors known as errata which
may cause the product to deviate from published specifications. Current characterized errata are
available on request.
Implementations of the I2C bus/protocol may require licenses from various entities, including Philips
Electronics N.V. and North American Philips Corporation.
Intel and the Intel logo are trademarks of Intel Corporation in the U.S. and other countries.
*Other names and brands may be claimed as the property of others.
Copyright © 2012, Intel Corporation. All rights reserved.
2 5/29/2012 Doc Ref #: IHD-OS-V1 Pt 3 – 05 12
Contents
1. Render Engine Command Streamer .................................................................................................... 4
1.1 Registers in Render Engine ................................................................................................................. 4
1.1.1 Introduction ..................................................................................................................................... 4
1.1.2 Outstanding Memory Requests Modulation Counters ................................................................... 6
1.1.3 Registers Used for Priority Field in Programmable Arbitration ....................................................... 9
1.1.4 Registers Used in Programmable Arbitration ............................................................................... 12
1.1.5 Virtual Memory Control ................................................................................................................. 29
1.1.6 GFX TLB In Use Virtual Address Registers ................................................................................. 34
1.1.7 GFX Pending TLB Cycles Information Registers ......................................................................... 39
1.1.8 Configuration Registers for Graphic Arbiter ................................................................................. 45
1.1.9 Context Save Registers ................................................................................................................ 47
1.1.10 Mode and Misc Ctrl Registers .................................................................................................... 54
1.1.11 RINGBUF — Ring Buffer Registers ........................................................................................... 75
1.1.12 Watchdog Timer Registers ......................................................................................................... 81
1.1.13 Interrupt Control Registers ......................................................................................................... 83
1.1.14 Logical Context Support ............................................................................................................. 89
1.1.15 Pipelines Statistics Counter Registers ..................................................................................... 101
1.1.16 Predicate Render Registers ..................................................................................................... 110
1.1.17 AUTO_DRAW Registers .......................................................................................................... 112
1.1.18 MMIO Registers for GPGPU Indirect Dispatch ........................................................................ 114
1.1.19 Memory Interface Registers ..................................................................................................... 117
1.2 Memory Interface Commands for Rendering Engine ...................................................................... 166
1.2.1 Introduction ................................................................................................................................. 166
1.2.2 Software Synchronization Commands ....................................................................................... 166
1.2.3 MI_ARB_CHECK ........................................................................................................................ 167
1.2.4 MI_ARB_ON_OFF ...................................................................................................................... 167
1.2.5 MI_BATCH_BUFFER_END ....................................................................................................... 168
1.2.6 MI_CONDITIONAL_BATCH_BUFFER_END ............................................................................. 168
1.2.7 MI_BATCH_BUFFER_START (Render) .................................................................................... 169
1.2.8 MI_CLFLUSH ............................................................................................................................. 173
1.2.9 MI_DISPLAY_FLIP ..................................................................................................................... 174
1.2.10 MI_FLUSH ................................................................................................................................ 176
1.2.11 MI_LOAD_REGISTER_IMM .................................................................................................... 178
1.2.12 MI_NOOP ................................................................................................................................. 180
1.2.13 Surface Probing ........................................................................................................................ 180
1.2.14 MI_REPORT_HEAD ................................................................................................................. 180
1.2.15 MI_SEMAPHORE_MBOX ........................................................................................................ 181
1.2.16 MI_SET_CONTEXT ................................................................................................................. 184
1.2.17 MI_STORE_DATA_IMM ........................................................................................................... 186
1.2.18 MI_STORE_DATA_INDEX ....................................................................................................... 187
1.2.19 MI_STORE_REGISTER_MEM ................................................................................................ 189
1.2.20 MI_SUSPEND_FLUSH ............................................................................................................ 190
1.2.21 MI_UPDATE_GTT .................................................................................................................... 191
1.2.22 MI_USER_INTERRUPT ........................................................................................................... 192
1.2.23 MI_WAIT_FOR_EVENT ........................................................................................................... 193
1.2.24 MI_LOAD_REGISTER_MEM ................................................................................................... 196
1.2.25 MI_URB_CLEAR ...................................................................................................................... 197
1.2.26 MI_PREDICATE ....................................................................................................................... 199
1.2.27 MI_TOPOLOGY_FILTER ......................................................................................................... 201
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 3
1. Render Engine Command Streamer
1.1 Registers in Render Engine
1.1.1 Introduction
This chapter describes the memory-mapped registers associated with the Memory Interface, including
brief descriptions of their use. The functions performed by some of these registers are discussed in more
detail in the Memory Interface Functions, Memory Interface Instructions, and Programming Environment
chapters.
The registers detailed in this chapter are used across the family of products and are extentions to
previous projects. However, slight changes may be present in some registers (i.e., for features added or
removed), or some registers may be removed entirely. These changes are clearly marked within this
chapter.
1.1.1.1 ARB_MODE – Arbiter Mode Control register
ARB_MODE - Arbiter Mode Control register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Size (in bits): 32
Trusted Type: 1
Address: 04030h
DWord Bit Description
0 31:16 Mask Bits
Default Value: 00000000000000000b
Access: RO
Format: U16
Mask bits act as write enables for the bits in the lower bits of this register
14 GAM to Bypass GTT Translation (GAM2BGTTT)
Default Value: 0b
Access: R/W
Format: MBZ
GAM to bypass GTT translation and pass logical addresses through with 0’s padded on the MSBs to
form the physical address.
13 DC GDR (DC_GDR)
Default Value: 0b
Access: R/W
DC GDR
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 4
ARB_MODE - Arbiter Mode Control register
12 HIZ GDR (HIZ_GDR)
Default Value: 0b
Access: R/W
HIZ GDR
11 STC GDR (STC_GDR)
Default Value: 0b
Access: R/W
Format: U1
STC GDR
10 BLB GDR (STC_GDR)
Default Value: 0b
Access: R/W
BLB GDR
9 GAM PD GDR (GAMPD_GDR)
Default Value: 0b
Access: R/W
GAM PD GDR
8 Color/Depth Port Share Bit (CDPS)
Default Value: 00b
Access: R/W
Format: U1
Color/Depth port share bit
This bit is used to force Color and Depth Caches to share an arbiter read request port. By default (Bit =
0) the Color Cache will NOT share the read request port with the Depth Cache.
5 Address Swizzling for Tiled Surfaces (AS4TS)
Access: R/W
Format: U1
Address Swizzling for Tiled-Surfaces. This register location is updated via GFX Driver prior to enabling
DRAM accesses. Driver needs to obtain the need for memory address swizzling via DRAM
configuration registers and set the following bits (in Display Engine and Render/Media access
streams).
Value Name
0b No address Swizzling
1b Address bit[6] needs to be swizzled for tiled surfaces
4 VMC GDR Enable (VMC_GDR_EN)
Access: R/W
When this bit is set, Data requested from the VMC client will be generated by the GDR algorithm
3 Texture Cache GDR Enable bit (TCGDREN)
Access: R/W
Format: U1
Texture Cache GDR enable bit When this bit is set, Data requested from the Texture Cache client will
be generated by the GDR algorithm (See GDR algorithm in xxx section)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 5
ARB_MODE - Arbiter Mode Control register
2 Depth Cache GDR enable bit (DCGDREN)
Access: R/W
Format: U1
When this bit is set, Data requested from the Depth Cache client will be generated by the GDR
algorithm (See GDR algorithm in xxx section)
1 Color Cache GDR enable bit(CCGDREN)
Access: R/W
Format: U1
When this bit is set, Data requested from the Color Cache client will be generated by the GDR
algorithm (See GDR algorithm in xxx section)
0 GTT Accesses GDR (GTTAGDR )
Default Value: 0b
Access: R/W
Format: U1
When this bit is enabled along with the Client’s GDR bit, PPGTT and GGTT requests for this memory
access will also be tagged as GDR to SQ.
1.1.2 Outstanding Memory Requests Modulation Counters
1.1.2.1 GFX_PEND_TLB_0 – Max Outstanding Pending TLB Requests 0
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04034h-04037h
DWord Bit Description
0 31 TEX Limit Enable bit
Format: U1
This bit is used to enable the pending TLB requests limitation function for the Texture Cache. When
set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
30 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 6
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
29:24 TEX TLB Limit Count
Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
23 ISC Limit Enable bit
Format: U1
This bit is used to enable the pending TLB requests limitation function for the Instruction Cache. When
set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
22 Reserved
Format: MBZ
21:16 ISC TLB Limit Count
Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
15 VF Limit Enable bit
Format: U1
This bit is used to enable the pending TLB requests limitation function for the Vertex Fetch. When set,
the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
14 Reserved
Format: MBZ
13:8 VF TLB Limit Count
Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
7 CS Limit Enable bit
Format: U1
This bit is used to enable the pending TLB requests limitation function for the Command Streamer.
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
6 Reserved
Format: MBZ
5:0 CS TLB Limit Count
Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 7
1.1.2.2 GFX_PEND_TLB_1 – Max Outstanding Pending TLB Requests 1
GFX_PEND_TLB_1 - Max Outstanding Pending TLB Requests 1
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04038h-0403Bh
DWord Bit Description
0 31:16 Reserved
Format: MBZ
15 RCZ Limit Enable bit
Format: U1
This bit is used to enable the pending TLB requests limitation function for the Render Depth Cache.
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
14 Reserved
Format: MBZ
13:8 RCZ TLB Limit Count
Format: U6
This bit is used to enable the pending TLB requests limitation function for the Render Color Cache.
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
7 RCC Limit Enable bit
Format: U1
This bit is used to enable the pending TLB requests limitation function for the Render Color Cache.
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
6 Reserved
Format: MBZ
5:0 RCC TLB Limit Count
Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 8
1.1.3 Registers Used for Priority Field in Programmable Arbitration
1.1.3.1 MIDARB_PRIO_HIT_REGISTER – Priority Field in Programmable
Arbitration for Hit
MIDARB_PRIO_HIT_REGISTER - Priority Field in Programmable
Arbitration for Hit
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 16
Trusted Type: 1
Address: 043A0h
DWord Bit Description
0 31:12 Reserved
11:9 Encoded Programmable Priority for MIDARB_GOTOFIELD_HIT3 Register
Encoding Priority 1 Priority 2 Priority 3
000 CS/VF/ISC MT/CTC RCC
001 CS/VF/ISC RCC MT/CTC
010 RCC CS/VF/ISC MT/CTC
011 RCC MT/CTC CS/VF/ISC
100 MT/CTC CS/VF/ISC RCC
101 MT/CTC RCC CS/VF/ISC
110 Reserved Reserved Reserved
111 Reserved Reserved Reserved
8:6 Encoded Programmable Priority for MIDARB_GOTOFIELD_HIT2 Register
5:3 Encoded Programmable Priority for MIDARB_GOTOFIELD_HIT1 Register
2:0 Encoded Programmable Priority for MIDARB_GOTOFIELD_HIT0 Register
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 9
1.1.3.2 MIDARB_PRIO_MISS_REGISTER – Priority Field in Programmable
Arbitration for Miss
MIDARB_PRIO_MISS_REGISTER - Priority Field in Programmable
Arbitration for Miss
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04204h
DWord Bit Description
0 31:20 Reserved
19:15 Encoded Programmable Priority for MIDARB_GOTOFIELD_MISS3 Register
14:10 Encoded Programmable Priority for MIDARB_GOTOFIELD_MISS2 Register
9:5 Encoded Programmable Priority for MIDARB_GOTOFIELD_MISS1 Register
4:0 Encoded Programmable Priority for MIDARB_GOTOFIELD_MISS0 Register
1.1.3.3 MIDARB_PRIO_NP_REGISTER – Priority Field in Programmable
Arbitration for Hit-NP
MIDARB_PRIO_NP_REGISTER - Priority Field in Programmable
Arbitration for Hit-NP
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043A4h
Address: 04208h
DWord Bit Description
0 31:20 Reserved
19:15 Encoded Programmable Priority for MIDARB_GOTOFIELD_MISS3 Register
Encoding Priority 1 Priority 2 Priority 3 Priority 4
00000 CS/VF/ISC MT_CTC RCC RCZ_HiZ_Stnc
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 10
MIDARB_PRIO_NP_REGISTER - Priority Field in Programmable
Arbitration for Hit-NP
00001 CS/VF/ISC RCC MT_CTC RCZ_HiZ_Stnc
00010 RCC CS/VF/ISC MT_CTC RCZ_HiZ_Stnc
00011 RCC MT_CTC CS/VF/ISC RCZ_HiZ_Stnc
00100 MT_CTC CS/VF/ISC RCC RCZ_HiZ_Stnc
00101 MT_CTC RCC CS/VF/ISC RCZ_HiZ_Stnc
01000 CS/VF/ISC MT_CTC RCZ_HiZ_Stnc RCC
01001 CS/VF/ISC RCC RCZ_HiZ_Stnc MT_CTC
01010 RCC CS/VF/ISC RCZ_HiZ_Stnc MT_CTC
01011 RCC MT_CTC RCZ_HiZ_Stnc CS/VF/ISC
01100 MT_CTC CS/VF/ISC RCZ_HiZ_Stnc RCC
01101 MT_CTC RCC RCZ_HiZ_Stnc CS/VF/ISC
10000 CS/VF/ISC RCZ_HiZ_Stnc MT_CTC RCC
10001 CS/VF/ISC RCZ_HiZ_Stnc RCC MT_CTC
10010 RCC RCZ_HiZ_Stnc CS/VF/ISC MT_CTC
10011 RCC RCZ_HiZ_Stnc MT_CTC CS/VF/ISC
10100 MT_CTC RCZ_HiZ_Stnc CS/VF/ISC RCC
10101 MT_CTC RCZ_HiZ_Stnc RCC CS/VF/ISC
11000 RCZ_HiZ_Stnc CS/VF/ISC MT_CTC RCC
11001 RCZ_HiZ_Stnc CS/VF/ISC RCC MT_CTC
11010 RCZ_HiZ_Stnc RCC CS/VF/ISC MT_CTC
11011 RCZ_HiZ_Stnc RCC MT_CTC CS/VF/ISC
11100 RCZ_HiZ_Stnc MT_CTC CS/VF/ISC RCC
11101 RCZ_HiZ_Stnc MT_CTC RCC CS/VF/ISC
Other values Reserved
14:10 Encoded Programmable Priority for MIDARB_GOTOFIELD_NP2 Register
9:5 Encoded Programmable Priority for MIDARB_GOTOFIELD_NP1 Register
4:0 Encoded Programmable Priority for MIDARB_GOTOFIELD_NP0 Register
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 11
1.1.4 Registers Used in Programmable Arbitration
1.1.4.1 MIDARB_GOTOFIELD_HIT0_REGISTER – Goto Field in Programmable
Arbitration for Hit0
MIDARB_GOTOFIELD_HIT0 - Goto Field in Programmable Arbitration
for Hit0
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 16
Trusted Type: 1
Address: 043B0h
DWord Bit Description
0 31:16 Reserved
Format: MBZ
15:14 Goto field when request vector is 111
Determines the GOTO and priority register to be used next:
Value Name Description
00b Use MIDARB_GOTOFIELD_HIT0 and MIDARB_PRIO_HIT_REGISTER[2:0]
01b Use MIDARB_GOTOFIELD_HIT1 and MIDARB_PRIO_HIT_REGISTER[5:3]
10b Use MIDARB_GOTOFIELD_HIT2 and MIDARB_PRIO_HIT_REGISTER[8:6]
11b Use MIDARB_GOTOFIELD_HIT3 and MIDARB_PRIO_HIT_REGISTER[11:9]
13:12 Goto field when request vector is 110b.
11:10 Goto field when request vector is 101b.
9:8 Goto field when request vector is 100b.
7:6 Goto field when request vector is 011b.
5:4 Goto field when request vector is 010b.
3:2 Goto field when request vector is 001b.
1:0 Goto field when request vector is 000b.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 12
1.1.4.2 MIDARB_GOTOFIELD_HIT1_REGISTER – Goto Field in Programmable
Arbitration for Hit1
MIDARB_GOTOFIELD_HIT1 - Goto Field in Programmable Arbitration
for Hit1
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 16
Trusted Type: 1
Address: 043B4h
DWord Bit Description
0 31:16 Reserved
Format: MBZ
15:14 Goto field when request vector is 111
Determines the GOTO and priority register to be used next
Value Name Description
00b Use MIDARB_GOTOFIELD_HIT0 and MIDARB_PRIO_HIT_REGISTER[2:0]
01b Use MIDARB_GOTOFIELD_HIT1 and MIDARB_PRIO_HIT_REGISTER[5:3]
10b Use MIDARB_GOTOFIELD_HIT2 and MIDARB_PRIO_HIT_REGISTER[8:6]
11b Use MIDARB_GOTOFIELD_HIT3 and MIDARB_PRIO_HIT_REGISTER[11:9]
13:12 Goto field when request vector is 110b.
11:10 Goto field when request vector is 101b.
9:8 Goto field when request vector is 100b.
7:6 Goto field when request vector is 011b.
5:4 Goto field when request vector is 010b.
3:2 Goto field when request vector is 001b.
1:0 Goto field when request vector is 000b.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 13
1.1.4.3 MIDARB_GOTOFIELD_HIT2_REGISTER – Goto Field in Programmable
Arbitration for Hit2
MIDARB_GOTOFIELD_HIT2 - Goto Field in Programmable Arbitration
for Hit2
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 16
Trusted Type: 1
Address: 043B8h
DWord Bit Description
0 31:16 Reserved
Format: MBZ
15:14 Goto field when request vector is 111.
Determines the GOTO and priority register to be used next
Value Name Description
00b Use MIDARB_GOTOFIELD_HIT0 and MIDARB_PRIO_HIT_REGISTER[2:0]
01b Use MIDARB_GOTOFIELD_HIT1 and MIDARB_PRIO_HIT_REGISTER[5:3]
10b Use MIDARB_GOTOFIELD_HIT2 and MIDARB_PRIO_HIT_REGISTER[8:6]
11b Use MIDARB_GOTOFIELD_HIT3 and MIDARB_PRIO_HIT_REGISTER[11:9]
13:12 Goto field when request vector is 110b.
11:10 Goto field when request vector is 101b.
9:8 Goto field when request vector is 100b.
7:6 Goto field when request vector is 011b.
5:4 Goto field when request vector is 010b.
3:2 Goto field when request vector is 001b.
1:0 Goto field when request vector is 000b.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 14
1.1.4.4 MIDARB_GOTOFIELD_HIT3_REGISTER – Goto Field in Programmable
Arbitration for Hit3
MIDARB_GOTOFIELD_HIT3 - Goto Field in Programmable Arbitration
for Hit3
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 16
Trusted Type: 1
Address: 043BCh
DWord Bit Description
0 31:16 Reserved
Format: MBZ
15:14 Goto field when request vector is 111.
Determines the GOTO and priority register to be used next.
Field for arbitration on next clock cycle for request entries of 111 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
Value Name Description
00b Use MIDARB_GOTOFIELD_HIT0 and MIDARB_PRIO_HIT_REGISTER[2:0]
01b Use MIDARB_GOTOFIELD_HIT1 and MIDARB_PRIO_HIT_REGISTER[5:3]
10b Use MIDARB_GOTOFIELD_HIT2 and MIDARB_PRIO_HIT_REGISTER[8:6]
11b Use MIDARB_GOTOFIELD_HIT3 and MIDARB_PRIO_HIT_REGISTER[11:9]
13:12 Goto field when request vector is 110.
Field for arbitration on next clock cycle for request entries of 110 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
11:10 Goto field when request vector is 101.
Field for arbitration on next clock cycle for request entries of 101 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
9:8 Goto field when request vector is 100.
Field for arbitration on next clock cycle for request entries of 100 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
7:6 Goto field when request vector is 011.
Field for arbitration on next clock cycle for request entries of 011 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
5:4 Goto field when request vector is 010.
Field for arbitration on next clock cycle for request entries of 010 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
3:2 Goto field when request vector is 001.
Field for arbitration on next clock cycle for request entries of 001 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
1:0 Goto field when request vector is 000.
Field for arbitration on next clock cycle for request entries of 000 corresponding to arbitration
action field entry of MIDARB_PRIO_HIT_REGISTER[11:9]
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 15
1.1.4.5 MIDARB_GOTOFIELD_NP0_REGISTER – Goto Field in Programmable
Arbitration for Hit-NP0
MIDARB_GOTOFIELD_NP0 - Goto Field in Programmable Arbitration
for Hit-NP0
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043C0h
DWord Bit Description
0 31:30 Goto field when request vector is 1111.
Determines the GOTO and priority register to be used next.
Value Name Description
00b Use MIDARB_GOTOFIELD_NP0 and MIDARB_PRIO_NP_REGISTER[4:0]
01b Use MIDARB_GOTOFIELD_NP1 and MIDARB_PRIO_NP_REGISTER[9:5]
10b Use MIDARB_GOTOFIELD_NP2 and MIDARB_PRIO_NP_REGISTER[14:10]
11b Use MIDARB_GOTOFIELD_NP3 and MIDARB_PRIO_NP_REGISTER[19:15]
29:28 Goto field when request vector is 1110b.
27:26 Goto field when request vector is 1101b.
25:24 Goto field when request vector is 1100b.
23:22 Goto field when request vector is 1011b.
21:20 Goto field when request vector is 1010b.
19:18 Goto field when request vector is 1001b.
17:16 Goto field when request vector is 1000b.
15:14 Goto field when request vector is 0111b.
13:12 Goto field when request vector is 0110b.
11:10 Goto field when request vector is 0101b.
9:8 Goto field when request vector is 0100b.
7:6 Goto field when request vector is 0011b.
5:4 Goto field when request vector is 0010b.
3:2 Goto field when request vector is 0001b.
1:0 Goto field when request vector is 0000b.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 16
1.1.4.6 MIDARB_GOTOFIELD_NP1_REGISTER – Goto Field in Programmable
Arbitration for Hit-NP1
MIDARB_GOTOFIELD_NP1 - Goto Field in Programmable Arbitration
for Hit-NP1
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043C4h
DWord Bit Description
0 31:30 Goto field when request vector is 1111.
Determines the GOTO and priority register to be used next.
Value Name Description
00b Use MIDARB_GOTOFIELD_NP0 and MIDARB_PRIO_NP_REGISTER[4:0]
01b Use MIDARB_GOTOFIELD_NP1 and MIDARB_PRIO_NP_REGISTER[9:5]
10b Use MIDARB_GOTOFIELD_NP2 and MIDARB_PRIO_NP_REGISTER[14:10]
11b Use MIDARB_GOTOFIELD_NP3 and MIDARB_PRIO_NP_REGISTER[19:15]
29:28 Goto field when request vector is 1110b.
27:26 Goto field when request vector is 1101b.
25:24 Goto field when request vector is 1100b.
23:22 Goto field when request vector is 1011b.
21:20 Goto field when request vector is 1010b.
19:18 Goto field when request vector is 1001b.
17:16 Goto field when request vector is 1000b.
15:14 Goto field when request vector is 0111b.
13:12 Goto field when request vector is 0110b.
11:10 Goto field when request vector is 0101b.
9:8 Goto field when request vector is 0100b.
7:6 Goto field when request vector is 0011b.
5:4 Goto field when request vector is 0010b.
3:2 Goto field when request vector is 0001b.
1:0 Goto field when request vector is 0000b.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 17
1.1.4.7 MIDARB_GOTOFIELD_NP2_REGISTER – Goto Field in Programmable
Arbitration for Hit-NP2
MIDARB_GOTOFIELD_NP2 - Goto Field in Programmable Arbitration
for Hit-NP2
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043C8h
DWord Bit Description
0 31:30 Goto field when request vector is 1111.
Determines the GOTO and priority register to be used next.
Value Name Description
00b Use MIDARB_GOTOFIELD_NP0 and MIDARB_PRIO_NP_REGISTER[4:0]
01b Use MIDARB_GOTOFIELD_NP1 and MIDARB_PRIO_NP_REGISTER[9:5]
10b Use MIDARB_GOTOFIELD_NP2 and MIDARB_PRIO_NP_REGISTER[14:10]
11b Use MIDARB_GOTOFIELD_NP3 and MIDARB_PRIO_NP_REGISTER[19:15]
29:28 Goto field when request vector is 1110b.
27:26 Goto field when request vector is 1101b.
25:24 Goto field when request vector is 1100b.
23:22 Goto field when request vector is 1011b.
21:20 Goto field when request vector is 1010b.
19:18 Goto field when request vector is 1001b.
17:16 Goto field when request vector is 1000b.
15:14 Goto field when request vector is 0111b.
13:12 Goto field when request vector is 0110b.
11:10 Goto field when request vector is 0101b.
9:8 Goto field when request vector is 0100b.
7:6 Goto field when request vector is 0011b.
5:4 Goto field when request vector is 0010b.
3:2 Goto field when request vector is 0001b.
1:0 Goto field when request vector is 0000b.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 18
1.1.4.8 MIDARB_GOTOFIELD_NP3_REGISTER – Goto Field in Programmable
Arbitration for Hit-NP3
MIDARB_GOTOFIELD_NP3 - Goto Field in Programmable Arbitration
for Hit-NP3
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043CCh
DWord Bit Description
0 31:30 Goto field when request vector is 1111.
Determines the GOTO and priority register to be used next.
Value Name Description
00b Use MIDARB_GOTOFIELD_NP0 and MIDARB_PRIO_NP_REGISTER[4:0]
01b Use MIDARB_GOTOFIELD_NP1 and MIDARB_PRIO_NP_REGISTER[9:5]
10b Use MIDARB_GOTOFIELD_NP2 and MIDARB_PRIO_NP_REGISTER[14:10]
11b Use MIDARB_GOTOFIELD_NP3 and MIDARB_PRIO_NP_REGISTER[19:15]
29:28 Goto field when request vector is 1110b.
27:26 Goto field when request vector is 1101b.
25:24 Goto field when request vector is 1100b.
23:22 Goto field when request vector is 1011b.
21:20 Goto field when request vector is 1010b.
19:18 Goto field when request vector is 1001b.
17:16 Goto field when request vector is 1000b.
15:14 Goto field when request vector is 0111b.
13:12 Goto field when request vector is 0110b.
11:10 Goto field when request vector is 0101b.
9:8 Goto field when request vector is 0100b.
7:6 Goto field when request vector is 0011b.
5:4 Goto field when request vector is 0010b.
3:2 Goto field when request vector is 0001b.
1:0 Goto field when request vector is 0000b.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 19
1.1.4.9 ARB_GAC_GAM_REQCNTS0 – GAC_GAM Arbitration Counters
Register 0
ARB_GAC_GAM_REQCNTS0 - GAC_GAM Arbitration Counters
Register 0
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043A8h
DWord Bit Description
0 31:22 Reserved
21:16 Number of GAC WR requests to be accumulated before applying the arbitration
15:14 Reserved
13:8 Number of GAC R requests to be accumulated before applying the arbitration
7:6 Reserved
5:0 Number of GAC RO requests to be accumulated before applying the arbitration
1.1.4.10 ARB_GAC_GAM_REQCNTS1 – GAC_GAM Arbitration Counters
Register 1
ARB_GAC_GAM_REQCNTS1 - GAC_GAM Arbitration Counters
Register 1
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043ACh
DWord Bit Description
0 31:22 Reserved
21:16 Number of GAC WR requests to be accumulated before applying the arbitration
15:14 Reserved
13:8 Number of GAC R requests to be accumulated before applying the arbitration
7:6 Reserved
5:0 Number of GAC RO requests to be accumulated before applying the arbitration
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 20
1.1.4.11 ARB_RO_GAC_GAM0 – GAC_GAM RO Arbitration Register 0
ARB_RO_GAC_GAM0 - GAC_GAM RO Arbitration Register 0
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043D0h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 1
26:24 Goto field for entry 1 when request vector is 11b
23:21 Goto field for entry 1 when request vector is 10b
20:18 Goto field for entry 1 when request vector is 01b
17:15 Goto field for entry 1 when request vector is 00b
14:13 Reserved
12 Priority for entry 01
11:9 Goto field for entry 01 when request vector is 11b
8:6 Goto field for entry 01 when request vector is 10b
5:3 Goto field for entry 01 when request vector is 01b
2:0 Goto field for entry 01 when request vector is 00b
1.1.4.12 ARB_RO_GAC_GAM1 – GAC_GAM RO Arbitration Register 1
ARB_RO_GAC_GAM1 - GAC_GAM RO Arbitration Register 1
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043D4h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 3
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 21
ARB_RO_GAC_GAM1 - GAC_GAM RO Arbitration Register 1
26:24 Goto field for entry 3 when request vector is 11b
23:21 Goto field for entry 3 when request vector is 10b
20:18 Goto field for entry 3 when request vector is 01b
17:15 Goto field for entry 3 when request vector is 00b
14:13 Reserved
12 Priority for entry 2
11:9 Goto field for entry 2 when request vector is 11b
8:6 Goto field for entry 2 when request vector is 10b
5:3 Goto field for entry 2 when request vector is 01b
2:0 Goto field for entry 2 when request vector is 00b
1.1.4.13 ARB_RO_GAC_GAM2 – GAC_GAM RO Arbitration Register 2
ARB_RO_GAC_GAM2 - GAC_GAM RO Arbitration Register 2
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043D8h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 5
26:24 Goto field for entry 5 when request vector is 11b
23:21 Goto field for entry 5 when request vector is 10b
20:18 Goto field for entry 5 when request vector is 01b
17:15 Goto field for entry 5 when request vector is 00b
14:13 Reserved
12 Priority for entry 4
11:9 Goto field for entry 4 when request vector is 11b
8:6 Goto field for entry 4 when request vector is 10b
5:3 Goto field for entry 4 when request vector is 01b
2:0 Goto field for entry 4 when request vector is 00b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 22
1.1.4.14 ARB_RO_GAC_GAM3 – GAC_GAM RO Arbitration Register 3
ARB_RO_GAC_GAM3 - GAC_GAM RO Arbitration Register 3
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043DCh
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 7
26:24 Goto field for entry 7 when request vector is 11b
23:21 Goto field for entry 7 when request vector is 10b
20:18 Goto field for entry 7 when request vector is 01b
17:15 Goto field for entry 7 when request vector is 00b
14:13 Reserved
12 Priority for entry 6
11:9 Goto field for entry 6 when request vector is 11b
8:6 Goto field for entry 6 when request vector is 10b
5:3 Goto field for entry 6 when request vector is 01b
2:0 Goto field for entry 6 when request vector is 00b
1.1.4.15 ARB_R_GAC_GAM0 – GAC_GAM R Arbitration Register 0
ARB_R_GAC_GAM0 - GAC_GAM R Arbitration Register 0
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043E0h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 1
26:24 Goto field for entry 1 when request vector is 11b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 23
ARB_R_GAC_GAM0 - GAC_GAM R Arbitration Register 0
23:21 Goto field for entry 1 when request vector is 10b
20:18 Goto field for entry 1 when request vector is 01b
17:15 Goto field for entry 1 when request vector is 00b
14:13 Reserved
12 Priority for entry 0
11:9 Goto field for entry 0 when request vector is 11b
8:6 Goto field for entry 0 when request vector is 10b
5:3 Goto field for entry 0 when request vector is 01b
2:0 Goto field for entry 0 when request vector is 00b
1.1.4.16 ARB_R_GAC_GAM1 – GAC_GAM R Arbitration Register 1
ARB_R_GAC_GAM1 - GAC_GAM R Arbitration Register 1
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043E4h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 3
26:24 Goto field for entry 3 when request vector is 11b
23:21 Goto field for entry 3 when request vector is 10b
20:18 Goto field for entry 3 when request vector is 01b
17:15 Goto field for entry 3 when request vector is 00b
14:13 Reserved
12 Priority for entry 2
11:9 Goto field for entry 2 when request vector is 11b
8:6 Goto field for entry 2 when request vector is 10b
5:3 Goto field for entry 2 when request vector is 01b
2:0 Goto field for entry 2 when request vector is 00b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 24
1.1.4.17 ARB_R_GAC_GAM2 – GAC_GAM R Arbitration Register 2
ARB_R_GAC_GAM2 - GAC_GAM R Arbitration Register 2
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043E8h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 5
26:24 Goto field for entry 5 when request vector is 11b
23:21 Goto field for entry 5 when request vector is 10b
20:18 Goto field for entry 5 when request vector is 01b
17:15 Goto field for entry 5 when request vector is 00b
14:13 Reserved
12 Priority for entry 4
11:9 Goto field for entry 4 when request vector is 11b
8:6 Goto field for entry 4 when request vector is 10b
5:3 Goto field for entry 4 when request vector is 01b
2:0 Goto field for entry 4 when request vector is 00b
1.1.4.18 ARB_R_GAC_GAM3 – GAC_GAM R Arbitration Register 3
ARB_R_GAC_GAM3 - GAC_GAM R Arbitration Register 3
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043ECh
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 7
26:24 Goto field for entry 7 when request vector is 11b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 25
ARB_R_GAC_GAM3 - GAC_GAM R Arbitration Register 3
23:21 Goto field for entry 7 when request vector is 10b
20:18 Goto field for entry 7 when request vector is 01b
17:15 Goto field for entry 7 when request vector is 00b
14:13 Reserved
12 Priority for entry 6
11:9 Goto field for entry 6 when request vector is 11b
8:6 Goto field for entry 6 when request vector is 10b
5:3 Goto field for entry 6 when request vector is 01b
2:0 Goto field for entry 6 when request vector is 00b
1.1.4.19 ARB_WR_GAC_GAM0 – GAC_GAM WR Arbitration Register 0
ARB_WR_GAC_GAM0 - GAC_GAM WR Arbitration Register 0
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043F0h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 1
26:24 Goto field for entry 1 when request vector is 11b
23:21 Goto field for entry 1 when request vector is 10b
20:18 Goto field for entry 1 when request vector is 01b
17:15 Goto field for entry 1 when request vector is 00b
14:13 Reserved
12 Priority for entry 0
11:9 Goto field for entry 0 when request vector is 11b
8:6 Goto field for entry 0 when request vector is 10b
5:3 Goto field for entry 0 when request vector is 01b
2:0 Goto field for entry 0 when request vector is 00b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 26
1.1.4.20 ARB_WR_GAC_GAM1 – GAC_GAM WR Arbitration Register 1
ARB_WR_GAC_GAM1 - GAC_GAM WR Arbitration Register 1
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043F4h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 3
26:24 Goto field for entry 3 when request vector is 11b
23:21 Goto field for entry 3 when request vector is 10b
20:18 Goto field for entry 3 when request vector is 01b
17:15 Goto field for entry 3 when request vector is 00b
14:13 Reserved
12 Priority for entry 2
11:9 Goto field for entry 2 when request vector is 11b
8:6 Goto field for entry 2 when request vector is 10b
5:3 Goto field for entry 2 when request vector is 01b
2:0 Goto field for entry 2 when request vector is 00b
1.1.4.21 ARB_WR_GAC_GAM2 – GAC_GAM WR Arbitration Register 2
ARB_WR_GAC_GAM2 - GAC_GAM WR Arbitration Register 2
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043F8h
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 5
26:24 Goto field for entry 5 when request vector is 11b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 27
ARB_WR_GAC_GAM2 - GAC_GAM WR Arbitration Register 2
23:21 Goto field for entry 5 when request vector is 10b
20:18 Goto field for entry 5 when request vector is 01b
17:15 Goto field for entry 5 when request vector is 00b
14:13 Reserved
12 Priority for entry 4
11:9 Goto field for entry 4 when request vector is 11b
8:6 Goto field for entry 4 when request vector is 10b
5:3 Goto field for entry 4 when request vector is 01b
2:0 Goto field for entry 4 when request vector is 00b
1.1.4.22 ARB_WR_GAC_GAM3 – GAC_GAM WR Arbitration Register 3
ARB_WR_GAC_GAM3 - GAC_GAM WR Arbitration Register 3
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 043FCh
DWord Bit Description
0 31:28 Reserved
27 Priority for entry 7
26:24 Goto field for entry 7 when request vector is 11b
23:21 Goto field for entry 7 when request vector is 10b
20:18 Goto field for entry 7 when request vector is 01b
17:15 Goto field for entry 7 when request vector is 00b
14:13 Reserved
12 Priority for entry 6
11:9 Goto field for entry 6 when request vector is 11b
8:6 Goto field for entry 6 when request vector is 10b
5:3 Goto field for entry 6 when request vector is 01b
2:0 Goto field for entry 6 when request vector is 00b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 28
1.1.5 Virtual Memory Control
1.1.5.1 HWS_PGA — Hardware Status Page Address Register
Programming Note: If this register is written, a workload must subsequently be dispatched to the render
command streamer.
HWS_PGA - Hardware Status Page Address Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04080h
This register is used to program the 4 KB-aligned System Memory address of the Hardware Status Page used to
report hardware status into (typically cacheable) System Memory.
Programming Notes Project
If this register is written, a workload must subsequently be dispatched to the Render command streamer.
DWord Bit Description
0 31:12 Address
Format: GraphicsAddress[31:12]
This field is used by SW to specify Bits 31:12 of the 4 KB-aligned System Memory address of the 4 KB
page known as the Hardware Status Page. The Global GTT is used to map this page from the graphics
virtual address to physical address.
Programming Notes
If the Per-Process Virtual Address Space and Exec List Enable bit is set, HW requires that the status
page is programmed to allow for the context switch status to be reported.
11:0 Reserved
Format: MBZ
The following table defines the layout of the Hardware Status Page:
Hardware Status Page Layout
Source: RenderCS
Default 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
Value: 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 29
Hardware Status Page Layout
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 30
Hardware Status Page Layout
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 31
Hardware Status Page Layout
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000, 0x00000000,
0x00000000, 0x00000000
DWord Bit Description
0 31:0 Interrupt Status Register Storage
Project: All
The content of the ISR register is written to this location whenever an “unmasked” bit of the ISR (as
determined by the HWSTAM register) changes state.
1..3 31:0 Reserved
Project: All
Must not be used.
4 31:0 Ring Head Pointer Storage
Project: All
The contents of the Ring Buffer Head Pointer register (register DWord 1) are written to this location
either as result of an MI_REPORT_HEAD instruction or as the result of an “automatic report” (see
RINGBUF registers).
5..15 31:0 Reserved
Project: All
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 32
Hardware Status Page Layout
Must not be used.
16..27 31:0 Context Status DWords
Project: All
28..30 31:0 Reserved
Project: All
Must not be used.
31 31:0 Last Written Status Offset
Project: All
32..1023 31:0 General Purpose
Project: All
These locations can be used for general purpose via the MI_STORE_DATA_INDEX or
MI_STORE_DATA_IMM instructions.
1.1.5.2 PP_DCLV – PPGTT Directory Cacheline Valid Register
PP_DCLV - PPGTT Directory Cacheline Valid Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Size (in bits): 64
Address: 02220h
Description Project
Access: R/W
This register controls update of the on-chip PPGTT Directory Cache during a context restore. Bits that are
set will trigger the load of the corresponding 16 directory entry group. This register is restored with context
(prior to restoring the on-chip directory cache itself). This register is also restored when switching to a
context whose LRCA matches the current CCID if the Force PD Restore bit is set in the context descriptor.
The context image of this register must be updated and maintained by SW; SW should not normally need to
read this register.
This register can also effectively be used to limit the size of a process's virtual address space. Any access
by a process that requires a PD entry in a set that is not enabled in this register will cause a fatal error, and
no fetch of the PD entry will be attempted.
Programming Notes Project
Page Directory Base Register is a Global Context Register (power context) and not maintained per context in
ring buffer mode of submission. One should explicitly load PP_DCLV followed by PP_DIR_BASE register
through Load Register Immediate commands in Ring Buffer before submitting a context. One should
program these registers after ensuring the pipe is completely flushed with TLB’s invalidated.
DWord Bit Description
0 63:32 Reserved
Project: All
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 33
PP_DCLV - PPGTT Directory Cacheline Valid Register
Format: MBZ
31:0 PPGTT Directory Cache Restore [1..32] 16 entries
Project: All
Format: BitMask[Enable]
If set, the [1st..32nd] 16 entries of the directory cache are considered valid and will be brought in on
context restore. If clear, these entries are considered invalid and fetch of these entries will not be
attempted.
1.1.6 GFX TLB In Use Virtual Address Registers
1.1.6.1 MTTLB_VA — MT Virtual Page Address Registers
MTTLB_VA - MT Virtual Page Address Registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04800h-048FCh
DWord Bit Description
0 31:12 Address
Format: GraphicsAddress[31:12]
Page virtual address.
11:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 34
1.1.6.2 MTTLB_VLD — Valid Bit Vector 0 for MTTLB
MTTLB_VLD0 - Valid Bit Vector 0 for MTTLB
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04780h-04783h
This register contains the valid bits for entries 0-31 of MTTLB (Texture and constant cache TLB).
DWord Bit Description
0 31:0 Valid bits per entry
1.1.6.3 MTTLB_VLD — Valid Bit Vector 1 for MTTLB
MTTLB_VLD1 - Valid Bit Vector 1 for MTTLB
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04784h-04787h
This register contains the valid bits for entries 0-31 of MTTLB (Texture and constant cache TLBVertex Fetch,
Instruction Cache, and Command Streamer TLB).
DWord Bit Description
0 31:0 Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 35
1.1.6.4 VICTLB_VA — VIC Virtual page Address Registers
VICTLB_VA - VIC Virtual page Address Registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04900h-049FCh
These registers are directly mapped to the current Virtual Addresses in the VICTLB (Vertex Fetch, Instruction Cache,
and Command Streamer TLB.)
DWord Bit Description
0 31:12 Address
Format: GraphicsAddress[31:12]
Page virtual address.
11:0 Reserved
Format: MBZ
1.1.6.5 VICTLB_VLD — Valid Bit Vector 0 for MTVICTLB
VICTLB_VLD0 - Valid Bit Vector 0 for MTVICTLB
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04788h-0478Bh
This register contains the valid bits for entries 0-31 of VICTLB (Vertex Fetch, Instruction Cache, and Command
Streamer TLB).
DWord Bit Description
0 31:0 Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 36
1.1.6.6 VICTLB_VLD — Valid Bit Vector 1 for MTVICTLB
MTVICTLB_VLD1 - Valid Bit Vector 1 for MTVICTLB
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 0478Ch-0478Fh
This register contains the valid bits for entries 0-31 of VICTLB (Vertex Fetch, Instruction Cache, and Command
Streamer TLB).
DWord Bit Description
0 31:0 Valid bits per entry
1.1.6.7 RCCTLB_VA — Virtual page Address Registers
RCCTLB_VA - RCC Virtual page Address Registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04A00h-04AFCh
These registers are directly mapped to the current Virtual Addresses in the RCCTLB (Render Cache for Color TLB).
DWord Bit Description
0 31:12 Address
Project: All
Format: GraphicsAddress[31:12]
Page virtual address.
11:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 37
1.1.6.8 RCCTLB_VLD — Valid Bit Vector 0 for RCCTLB
RCCTLB_VLD0 - Valid Bit Vector 0 for RCCTLB
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04790h-04793h
This register contains the valid bits for entries 0-31 of RCCTLB (Render Cache for Color TLB).
DWord Bit Description
0 31:0 Valid bits per entry
1.1.6.9 RCZTLB_VA — RCZ Virtual Page Address Registers
RCZTLB_VA - RCZ Virtual Page Address Registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04B00h-04BFCh
These registers are directly mapped to the current Virtual Addresses in the RCZTLB (Render Cache for Z (Depth), Hi
Z, and Stencil TLB).
DWord Bit Description
0 31:12 Address
Format: GraphicsAddress[31:12]
Page virtual address.
11:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 38
1.1.6.10 RCZTLB_VLD0 — Valid Bit Vector 0 for RCZTLB
RCZTLB_VLD0 - Valid Bit Vector 0 for RCZTLB
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 04798h-0479Bh
This register contains the valid bits for entries 0-31 of RCZTLB (Render Cache for Z (Depth), Hi Z, and Stencil TLB).
DWord Bit Description
0 31:0 Valid bits per entry
1.1.6.11 RCZTLB_VLD1 — Valid Bit Vector 1 for RCZTLB
RCZTLB_VLD1 - Valid Bit Vector 1 for RCZTLB
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Trusted Type: 1
Address: 0479Ch-0479Fh
This register contains the valid bits for entries 0-31 of RCZTLB (Render Cache for Z (Depth), Hi Z, and Stencil TLB).
DWord Bit Description
0 31:0 Valid bits per entry
1.1.7 GFX Pending TLB Cycles Information Registers
The following registers contain information about cycles that did not complete their TLB translation.
Information is organized as 64 entries, where each entry has a valid and ready bit, collapsed into
separate registers.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 39
1.1.7.1 TLBPEND_VLD0 - Valid Bit Vector 0 for TLBPEND Registers
TLBPEND_VLD0 - Valid Bit Vector 0 for TLBPEND registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04700h-04703h
This register contains the valid bits for entries 0-31 of TLBPEND structure (Cycles pending TLB translation).
DWord Bit Description
0 31:0 Valid bits per entry
1.1.7.2 TLBPEND_VLD1 - Valid Bit Vector 1 for TLBPEND Registers
TLBPEND_VLD1 - Valid Bit Vector 1 for TLBPEND registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04704h-04707h
This register contains the valid bits for entries 32-63 of TLBPEND structure (Cycles pending TLB translation).
DWord Bit Description
0 31:0 Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 40
1.1.7.3 TLBPEND_RDY0 - Ready Bit Vector 0 for TLBPEND Registers
TLBPEND_RDY0 - Ready Bit Vector 0 for TLBPEND registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04708h-0470Bh
This register contains the ready bits for entries 0-31 of TLBPEND structure (Cycles pending TLB translation).
DWord Bit Description
0 31:0 Ready bits per entry
1.1.7.4 TLBPEND_RDY1 - Ready Bit Vector 1 for TLBPEND Registers
TLBPEND_RDY1 - Ready Bit Vector 1 for TLBPEND registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 0470Ch-0470Fh
This register contains the ready bits for entries 32-63 of TLBPEND structure (Cycles pending TLB translation).
DWord Bit Description
0 31:0 Ready bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 41
1.1.7.5 TLBPEND_SEC0 — Section 0 of TLBPEND Entry
TLBPEND_SEC0 - Section 0 of TLBPEND Entry
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04400h-044FCh
This register is directly mapped to the TLBPEND Array in the Graphic Arbiter.
DWord Bit Description
0 31 vtstatus
This bit will be used in conjunction with the ready bit to determine the stage of the translation. See table
below.
30:28 GTT bits
Bits 3:1 of the GTT entry used to translate the Virtual Address. 000 if translation is pending.
27:0 Current address
The value of this field depends on the stage of the TLB translation for this entry: VA – bits 27:20 = 00,
bits 19:0 = Bits 31:12 of the Virtual Address of the cycle.
VTDMODE Valid Ready Vtstatus
Meaning
DC 0 DC
DC Entry is invalid
0 1 0 0 Entry was a TLB miss. Waiting for TLB translation.
0 1 0 1
Entry was a Hit not present. Waiting for TLB translation from a previous miss.
0 1 1 0 Not possible
0 1 1 1 TLB translation complete. Entry ready
1 1 0 0 Entry was a TLB miss. Waiting for TLB translation.
1 1 0 1
Entry was a Hit not present. Waiting for TLB translation from a previous miss.
1 1 1 0 GPA translation complete. Entry ready for VTD translation.
1 1 1 1 TLB translation complete. Entry ready
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 42
1.1.7.6 TLBPEND_SEC1 — Section 1 of TLBPEND entry
TLBPEND_SEC1 - Section 1 of TLBPEND Entry
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04500h-045FCh
This register is directly mapped to the current Virtual Addresses in the MTTLB (Texture and constant cache
TLBRender Cache for Z (Depth), Hi Z, and Stencil TLB).
DWord Bit Description
0 31:28 Current address
Bits 9:6 of the Virtual Address of the cycle.
27:24 Cacheability Control Bits
Bits 3:1 of the GTT entry used to translate the Virtual Address. 000 if translation is pending.
3
2 Graphics Data Type (GFDT). This field contains the GFDT bit for this surface when writes occur.
GFDT can also be set by the GTT. The effective GFDT is the logical OR of this field with the GFDT
from the GTT entry. This field is ignored for reads.
1:0 Cacheability Control. This field controls cacheability in the mid-level cache (MLC) and last-level
cache (LLC).
00: use cacheability control bits from GTT entry
01: data is not cached in LLC or MLC
10: data is cached in LLC but not MLC
11: data is cached in both LLC and MLC
23 ZLR bit
Flag to indicate this is a zero length read, a read used to calculate a physical address for a write.
22:4 TAG
Cycle identification TAG.
3:0 SRC ID
Encoding of unit generating this cycle .
Value Name
0000b CS_RD_SRCID
0001b VF_RD_SRCID
0010b ISC_SRCID
0011b MT_SRCID
0100b RCC_SRCID
0101b HZARB_SRCID
0110b RCZ_SRCID
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 43
TLBPEND_SEC1 - Section 1 of TLBPEND Entry
0111b CTC_SRCID
1000b CS_WR_SRCID
1001b MBC_SRCID
1010b Reserved
1011b CS_RD_PWRCTX
1100b RC_R4WRCMP
1101b RESRVD2_SRCID
1110b RESRVD1_SRCID
1111b RESRVD0_SRCID
1.1.7.7 TLBPEND_SEC2 — Section 2 of TLBPEND entry
TLBPEND_SEC2 - Section 2 of TLBPEND Entry
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04600h-046FCh
This register is directly mapped to the current Virtual Addresses in the MTTLB (Texture and constant cache TLB).
DWord Bit Description
0 31:14 Reserved
13 Big Page Attribute
This entry is using a 32K page.
12:8 Current Address
Format: GraphicsAddress[14:10]
Bits 14:10 of the Virtual Address of the cycle.
7:0 PAT Entry
Location of Physical Address in Physical Address Table.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 44
1.1.8 Configuration Registers for Graphic Arbiter
1.1.8.1 ZSHR — Depth/Early Depth TLB Partitioning Register
ZSHR - Depth/Early Depth TLB Partitioning Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000020
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04050h
This register is used to determine the number of TLB entries from the total of 64 available to be used by the Depth
partition of the TLB. The rest of the entries are used for the Early Depth/Stencil TLB.
DWord Bit Description
0 31:6 Reserved
Format: MBZ
5:0 Number of TLB Entries Out of 64 used for Depth TLB
Default Value: 32
The rest are be used for Early Depth/Stencil TLB. Default value is 32.
1.1.8.2 Color/Depth Write FIFO Watermarks
CZWMRK - Color/Depth Write FIFO Watermarks
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 04060h
This register is directly mapped to the current Virtual Addresses in the MTTLB (Texture and constant cache TLB).
DWord Bit Description
0 31:24 Reserved
Format: MBZ
23:18 Color Wr Burst Size
This is the maximum size of the requests burst, from the last High Watermark trip, before reevaluating
the High Watermark again.
17:16 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 45
CZWMRK - Color/Depth Write FIFO Watermarks
15:12 Color Wr FIFO High Watermark
This is the number of accumulated Color writes that will trigger a Burst of Z Writes.
11:6 Z Wr Burst Size
This is the maximum size of the requests burst, from the last High Watermark trip, before reevaluating
the High Watermark again.
5:4 Reserved
Format: MBZ
3:0 Z Wr FIFO High Watermark
This is the number of accumulated Depth writes that will trigger a Burst of Z Writes.
1.1.8.3 PP_PFD[0:31] – PPGTT Page Fault Data Registers
PP_PFD[0:31] - PPGTT Page Fault Data Registers
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 04580h
The GTT Page Fault Log entries can be read from these registers.
4580h-4583h: Fault Entry 0
...
45FCh-45FFh: Fault Entry 31
DWord Bit Description
0 31:12 Fault Entry Page Address
Format: GraphicsAddress[31:12]
This RO field contains the faulting page address for this Fault Log entry. This field will contain a valid
fault address only if the bit in the GTT Page Fault Indication Register corresponding with the address
offset of this entry is set.
11:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 46
1.1.9 Context Save Registers
1.1.9.1 SVG_CTX — SVG Context Save Register
SVG_CTX - SVG Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: WO
Size (in bits): 32
Address: 06FFCh
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
15:1 Reserved
Format: MBZ
0 Context Save Start
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will kick off a context save. Once the save is
complete the bit will be cleared.
1.1.9.2 SVL_CTX— SVL Context Save Register
SVL_CTX - SVL Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 07FFCh
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
15:1 Reserved
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 47
SVL_CTX - SVL Context Save Register
Format: MBZ
0 Context Save Start
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will initiate a context save. Once the save is
complete the bit will be cleared.
1.1.9.3 WM_CTX— WM Context Save Register
WM_CTX - WM Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 05FFCh
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
15:1 Reserved
Format: MBZ
0 Context Save Start
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will kick off a context save. Once the save is
complete the bit will be cleared.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 48
1.1.9.4 SC_CTX— SC Context Save Register
SC_CTX - SC Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 0E1FCh
Address: 0F1FCh
Name: SC_CTX_SLICE1
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
15:5 Reserved
Format: MBZ
4 Context Save Start(MMIO and NP State)
Format: Enable
When a 1 is written to this bit with the mask bit set, it will kick off a context save. Once the save is
complete the bit will be cleared.
3:1 Reserved
Format: MBZ
0 Context Save Start(MMIO Only)
Format: Enable
When a 1 is written to this bit with the mask bit set, it will kick off a context save. Once the save is
complete the bit will be cleared.
1.1.9.5 DM_CTX — DM Context Save Register
DM_CTX - DM Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 49
DM_CTX - DM Context Save Register
Address: 0E0FCh
Address: 0F0FCh
Name: DM_CTX_SLICE1
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0.
15:1 Reserved
Format: MBZ
0 Context Save Start
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will initiate a context save. Once the save is
complete the bit will be cleared.
1.1.9.6 SARB_CTX— SARB Context Save Register
SARB_CTX - SARB Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 0B1FCh
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
15:1 Reserved
Format: MBZ
0 Context Save Start
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will kick off a context save. Once the save is
complete the bit will be cleared.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 50
1.1.9.7 VSC_CTX— VSC Context Save Register
VSC_CTX - VSC Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 051FCh
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
15:1 Reserved
Format: MBZ
0 Context Save Start
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will kick off a context save. Once the save is
complete, the bit will be cleared.
1.1.9.8 GPM_CTX— GPM Context Save Register
GPM_CTX - GPM Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 080FCh
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
15:1 Reserved
Format: MBZ
0 Context Save Start
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 51
GPM_CTX - GPM Context Save Register
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will initiate a context save. Once the save is
complete the bit will be cleared.
1.1.9.9 SOL_CTX— SOL Context Save Register
SOL_CTX - SOL Context Save Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 052FCh
This register is used to send messages to enable context saving. This register may not be written from CPU.
DWord Bit Description
0 31:6 Context Save Address/Offset
Default Value: 0h
Format: Address
This field specifies where the SOL context is to be saved.
If Power Context Save Mode is disabled, then the value of this field is the virtual address of the
location for SOL context to be saved.
If Power Context Save Mode is enabled, then the value of this field is the offset into the power context
image for SOL context to be saved.
5:2 Reserved
Format: MBZ
1 Power Context Save Mode
Default Value: 0h
Format: Enable
If set, then the save from SOL is for Power Context Save. If clear, then the save is for Ring Context
Save.
0 Context Save Start
Default Value: 0h
Format: Enable
When a 1 is written to this bit with the mask bit set, it will kick off a context save. Once the save is
complete, the bit will be cleared.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 52
1.1.9.10 1.1.9.11 RING_BUFFER_HEAD_PREEMPT_REG
RING_BUFFER_HEAD_PREEMPT_REG -
RING_BUFFER_HEAD_PREEMPT_REG
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 0214Ch
Name: RCS RING_BUFFER_HEAD_PREEMPT_REG
ShortName: RCS_RING_BUFFER_HEAD_PREEMPT_REG
This register contains the Head pointer offset in the ring when the last PREEMPTABLE command was executed
and caused the head pointer to move due to the UHPTR register being valid. If the PREEMPTABLE command is
executed as part of the batch buffer then the value of the register will be the offset in the ring of the command past
the batch buffer start that contained the preemptable command.
This is a global register and context save restored as part of power context image.
Programming Notes
Programming Restriction:
This register should NEVER be programmed by driver. This is for HW internal use only.
DWord Bit Description
0 31:21 Reserved
Format: MBZ
20:2 Preempted Head Offset
Format: U19
This field contains the Head pointer offset in the ring when the last MI_ARB_CHECK command was
executed and caused the head pointer to move due to the UHPTR register being valid.
1:0 Ring/Batch Indicator
Format: Enabled
Value Name Description
0h Ring Preemptable command was executed in ring and caused head pointer to be updated.
1h Batch Preemptable command was executed in batch and caused head pointer to be updated.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 53
1.1.9.11 BB_ADDR_DIFF—Batch Buffer Address Difference Register
BB_ADDR_DIFF - Batch Address Difference Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02154h
This register contains the difference between the start of the last batch and where the last initiated Batch Buffer is
currently fetching commands.
Programming Notes
Programming Restriction:
This register should NEVER be programmed by driver, this is for HW internal use only.
DWord Bit Description
0 31:2 Batch Buffer Address Difference
Format: GraphicsAddress[31:2]
This field specifies the DWord-aligned difference between the starting address of the batch buffer and
where the last initiated Batch Buffer is currently fetching commands.
1:0 Reserved
Format: MBZ
1.1.10 Mode and Misc Ctrl Registers
1.1.10.1 GT4 Mode Control Register
B/D/F/Type:MBCunit
Address Offset:9038-903Bh
Default Value:0h
Access: RW; RO;
Size:32 bits
Bit Access Default Value RST/PWR Description
1:0 R/W 00b Core
GT4 Usage mode:
00: Non-GT4
01: GT4 is used in Alternate Frame rendering Mode (AFR)
10: Basic Split Frame rendering Mode (SFR)
11: Complex Split Frame rendering Mode (SFR w/ CBR)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 54
Basic Split Frame Rendering is like CBR for all units except Windower. Windower should not be doing
any checker boarding in basic SFR. The split programming should be done scissor range programming.
Complex Split Frame Rendering (aka CBR) is already defined in many DCNs
1.1.10.2 MI_MODE — Render Mode Register for Software Interface
MI_MODE - Render Mode Register for Software Interface
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 0209Ch
The MI_MODE register contains information that controls software interface aspects of the Memory Interface
function.
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0
14 Async Flip Performance mode
Format: U1
Value Name Description
0h Performance mode enabled [Default] The stall of the flip event is in the windower
1h Performance mode disabled The stall of the flip event is in the command stream
Programming Notes
This bit should be set to '1' on all projects disabling Async Flip Performance mode.
When Async Flip Performance mode is enabled stall is in the Windower allowing the commands
following the MI_WAIT_FOR_EVENT to be parsed by command streamer, this breaks the usage
model of controlling the display message generation in display engine using
MI_LOAD_REGISTER_IMMEDIATE commands from ring buffer.
13 Flush Performance mode
Format: U1
Value Name Description
0h run fast restore [Default] No NonPipelined SV flush.
1h run slow legacy restore With NonPipelined SV flush.
11 Invalidate UHPTR enable
Format: Enable
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 55
MI_MODE - Render Mode Register for Software Interface
If bit set H/W clears the valid bit of UHPTR (2134h, bit 0) when current active head pointer is equal to
UHPTR.
10 Reserved
Format: MBZ
9 Rings Idle
Format: U1
Read Only Status bit
Value Name Description
0h Not Idle [Default] Parser not Idle or Ring Arbiter not Idle.
1h Idle Parser Idle and Ring Arbiter Idle.
Programming Notes
Writes to this bit are not allowed.
8 Stop Rings
Format: U1
Value Name Description
0h [Default] Normal Operation.
1h Parser is turned off and Ring arbitration is turned off.
Programming Notes
Software must set this bit to force the Rings and Command Parser to Idle. Software must read a 1 in
the Ring Idle bit after setting this bit to ensure that the hardware is idle.
Software must clear this bit for Rings to resume normal operation.
6 Vertex Shader Timer Dispatch Enable
Format: Enable
Value Name Description
0h Disable Disable the timer for dispatch of single vertices from the vertex shader. Vertex
[Default] shader will try to collect 2 vertices before a dispatch
1h Enable Enable the timer for dispatch of single vertices. Dispatch a single vertex shader
thread after the timer expires.
5 Reserved
Format: MBZ
4 Reserved
Format: MBZ
3:1 Reserved
Format: MBZ
0 Mask IIR disable
Format: Disable
Mask IIR disable. Nominally the Interrupt controller masks interrupts in the IIR register if an interrupt
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 56
MI_MODE - Render Mode Register for Software Interface
acknowledge from the 3gio interface is pending. Setting this bit to a 1 allows interrupts to be visible to
the interrupt controller while an interrupt acknowledge is pending.
1.1.10.3 FF_Mode - Thread Mode Register
FF_Mode - Thread Mode Register
Register Space: MMIO: 0/2/0
Source: RenderCS
0x28A01010
Access: R/W
Size (in bits): 32
Address: 020A0h
This register is used to program the FF shader Mode.
DWord Bit Description
0 31 Reserved
Format: MBZ
30 Reserved
Format: MBZ
29:26 DS Hit Max Value
Format: U4
Description Project
If the number of hits reaches the DS Hit Max Value and there is a pending miss to be
dispatched, the DS will dispatch the pending miss vertex as a single dispatch.
Programming the value beyond the range will have undefined behavior.
Value Name Project
10 [Default]
[1,11]
25:20 VS Hit Max Value
Format: U6
Description Project
If the number of hits reaches the VS Hit Max Value and there is a pending miss to be
dispatched, the VS will dispatch the pending miss vertex as a single dispatch.
Programming the value beyond the range will have undefined behavior.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 57
FF_Mode - Thread Mode Register
Value Name Project
10 [Default]
[1,58]
19 DS Reference Count Full Force Miss Enable
Project:
Format: Enable
Value Name Description
0b [Default] On a hit to the DS cache and the associated handle’s reference count is full then stall
until a derefernce.
1b On a hit to the DS cache and the associated handle’s reference count is full then force
the cycle as a miss and allocate a new handle.
18:17 TS Thread Dispatch Mode
Format: U2
Value Name Description
0h Load Balanced Thread Dispatch will load balance the half slices of the threads. Note: this will
[Default] cause possible corruption if input handles are reused due to instancing or
topologies that reuse vertices(i.e. strips and fans)
1h Half Slice 0 All threads will be dispatched to Half Slice 0.
2h Half Slice 1 All threads will be dispatched to Half Slice 1.
3h Reserved
16 TS Thread Dispatch Override Enable
Format: Enable
Value Name Description
0h Disable [Default] Hardware will decide which half slice the thread will dispatch.
1h Enable The value in the TS Thread Dispatch Mode will be used for dispatch.
15 VS Reference Count Full Force Miss Enable
Format: U1
Value Name Description
[0,1]
0b [Default] On a hit to the VS cache and the associated handle's reference count is full then stall
until a derefernce.
1b On a hit to the VS cache and the associated handle's reference count is full then force
the cycle as a miss and allocate a new handle.
14:13 VS Thread Dispatch Mode
Format: U2
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 58
FF_Mode - Thread Mode Register
Value Name Description
0h Load Balanced Thread Dispatch will load balance the half slices of the threads. Note: this will
[Default] cause possible corruption if input handles are reused due to instancing or
topologies that reuse vertices (i.e. strips and fans)
1h Half Slice 0 All threads will be dispatched to Half Slice 0.
2h Half Slice 1 All threads will be dispatched to Half Slice 1.
3h Reserved
12 VS Thread Dispatch Override Enable
Format: Enable
Value Name Description
0h Disable Hardware will decide which half slice the thread will dispatch.
1h Enable [Default] The value in the VS Thread Dispatch Mode will be used for dispatch.
11:7 Reserved
Format: MBZ
6:5 DS Thread Dispatch Mode
Format: U2
Value Name Description
0h Load Balanced Thread Dispatch will load balance the half slices of the threads. Note: this will
[Default] cause possible corruption if input handles are reused due to instancing or
topologies that reuse vertices (i.e., strips and fans).
1h Half Slice 0 All threads will be dispatched to Half Slice 0.
2h Half Slice 1 All threads will be dispatched to Half Slice 1.
3h Reserved
4 DS Thread Dispatch Override Enable
Format: Enable
Value Name Description
0h Disable Hardware will decide which half slice the thread will dispatch.
1h Enable [Default] The value in the DS Thread Dispatch Mode will be used for dispatch.
3:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 59
1.1.10.4 GFX_MODE – Graphics Mode Register
GFX_MODE - Graphics Mode Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000800
Size (in bits): 32
Trusted Type: 1
Address: 0229Ch
Description Project
This register contains a control bit for the new 2-level PPGTT functions.
DefaultValue = 00002800h
DWord Bit Description
0 31:16 Mask Bits
Format: Mask[15:0]
Must be set to modify corresponding bit in Bits 15:0. (All implemented bits)
14 Reserved
Project:
Format: MBZ
13 Flush TLB invalidation Mode
Format: U1
This field controls the invalidation if the TLB cache inside the hardware. When enabled this bit limits the
invalidation of the TLB only to batch buffer boundaries, to pipe_control commands which have the TLB
invalidation bit set and sync flushes. If disabled, the TLB caches are flushed for every full flush of the
pipeline.
12 Reserved
Project: All
Format: MBZ
11 Replay Mode
Format: U1 Context Switch Granularity
This field controls the granularity of the replay mechanism when coming back into a previously
preempted context.
Value Name Description Project
0h mid-triangle Super span Level. Pipeline is not flushed. This implies commands
preemption parsed are executed speculatively and may not complete before a
context switch.
1h mid-cmdbuffer Drawcall Level. Pipeline is flushed before switching to the next
preemption [Default] context. Commands parsed are commited to completing before a
context switch
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 60
GFX_MODE - Graphics Mode Register
Programming Notes
A fixed function pipe flush is required before modifying this fieldUnless pre-emption at a mid-triangle is
required the bit must be set.
10 Reserved
Project: All
Format: MBZ
9 Per-Process GTT Enable
Format: Enabled
Per-Process GTT Enable
Value Name Description
0h PPGTT When clear, the Global GTT will be used to translate memory access from
Disable designated commands and for commands that select the PPGTT as their
[Default] translation space.
1h PPGTT When set, the PPGTT will be used to translate memory access from designated
Enable commands and for commands that select the PPGTT as their translation space.
The PD Offset and PD Cacheline Valid registers must be set in all pipes (blitter,
MFX, render) before any workload is submitted to hardware. This mode enables
support for big pages (32k)
8 Reserved
Format: MBZ
7 Reserved
Format: MBZ
6:1 Reserved
Format: MBZ
0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 61
1.1.10.5 GT_MODE – GT Mode Register
GT_MODE - GT Mode Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000200
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 07008h
This Register is used to control the 6EU and 12EU configuration for GT.
Write 0x01FF01FF to this register enables the 6EU mode.
RegisterType = MMIO_SVL
DWord Bit Description
0 31:16 Mask Bits
Format: Mask[15:0]
Must be set to modify corresponding bit in Bits 15:0. (All implemented bits)
15 Reserved
Format: MBZ
14:11 Reserved
Format: MBZ
10 Reserved
Format: MBZ
9 WIZ Hashing Mode High Bit
Format: U1
This field adds additional hashing modes in combination with the WIZ Hashing Mode field. The Value
column in the table below refers to this field (high bit) and the WIZ Hashing Mode field (low bit).
This field is don’t care if the Hashing Disable bit is set.
Value Name Description
0h 8x8 Checkerboard hashing
1h [Default] 8x4 Checkerboard hashing
2h 16x4 Checkerboard hashing
3h Reserved
7 WIZ Hashing Mode
Project:
Format: U1
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 62
GT_MODE - GT Mode Register
Description Project
This field configures the Hashing mode in Windower. This field is don’t care if the
Hashing Disable bit is set.
The WIZ Hashing Mode High Bit field is combined with this field to enable additional
modes.
2 Reserved
Format: MBZ
0 Reserved
1.1.10.6 Cache_Mode_0— Cache Mode Register 0
Cache_Mode_0 - Cache Mode Register 0
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
0x00000004
Access: R/W
Size (in bits): 32
Address: 07000h
This register is used to control the operation of the Render and Sampler L2 Caches. All reserved bits are
implemented as read/write.
Before changing the value of this register, GFX pipeline must be idle i.e. full flush is required.
This Register is saved and restored as part of Context.
RegisterType = MMIO_SVL
DWord Bit Description
0 31:16 Masks
Format: Mask[15:0]
A 1 in a bit in this field allows the modification of the corresponding bit in Bits 15:0.
15 Sampler L2 Disable
Format: Disable
Value Name Description
0h [Default] Sampler L2 Cache Enabled.
1h Sampler L2 Cache Disabled all accesses are treated as misses.
14:12 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 63
Cache_Mode_0 - Cache Mode Register 0
11 Reserved
Format: MBZ
10 Reserved
Format: MBZ
9 Sampler L2 TLB Prefetch Enable
Value Name Description
0h [Default] TLB Prefetch Disabled
1h TLB Prefetch Enabled
7:6 Sampler L2 Request Arbitration
Format: U2
Value Name Description
00b Round Robin
01b Fetch are Highest Priority
10b Constants are Highest Priority
11b Reserved
5 STC Eviction Policy
Format: Disable
If this bit is set, RCCunit will have LRA as replacement policy. The default value i.e. ( when this bit is
reset ) indicates that non-LRA eviction policy. This bit must be reset. LRA replacement policy is not
supported.
Note: If this bit is set to "1", bit 7 of 0x7010h must also be set to "1"
4 RCC Eviction Policy
Format: Disable
If this bit is set, RCCunit will have LRA as replacement policy. The default value i.e. ( when this bit is
reset ) indicates that non-LRA eviction policy. This bit must be reset. LRA replacement policy is not
supported.
Note: If this bit is set to "1", bit 7 of 0x7010h must also be set to "1"
3 Hierarchical Z Disable
Mask: MMIO(0x2120)#19
Format: U1
Value Name
0h Enable
1h Disable
2 Hierarchical Z RAW Stall Optimization Disable
Format: U1
The Hierarchical Z RAW Stall Optimization allows non-overlapping polygons in the same 8x4
pixel/sample area to be processed without stalling waiting for the earlier ones to write to Hierarchical Z
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 64
Cache_Mode_0 - Cache Mode Register 0
buffer.
Value Name Description
0h Enable Enables the hierarchical Z RAW Stall Optimization.
1h Disable [Default] Disables the hierarchical Z RAW Stall Optimization.
1 Disable clock gating in the pixel backend
Format: Disable
0 Render Cache Operational Flush Enable
Format: Enable
Value Name Description
0h Disable Operational Flush Disabled (recommended for performance when not rendering
[Default] to the front buffer)
1h Enable Operational Flush Enabled (required when rendering to the front buffer)
Errata Description Project
Erratum This bit must be set to ‘0’ (Disable)
1.1.10.7 Cache_Mode_1— Cache Mode Register 1
Cache_Mode_1 - Cache Mode Register 1
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000180
Access: Read/32 bit Write Only
Size (in bits): 32
Address: 07004h
Description Project
RegisterType: MMIO_SVL
Before changing the value of this register, GFX pipeline must be idle; i.e., full flush is required. This Register
is saved and restored as part of Context.
DWord Bit Description
0 31:16 Mask Bits for 15:0
Format: Mask[15:0]
Must be set to modify corresponding data bit. Reads to this field returns zero.
15 Reserved
Project: All
Format: MBZ
14 Reserved
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 65
Cache_Mode_1 - Cache Mode Register 1
Format: MBZ
12 HIZ Eviction Policy
Project: All
Format: U1
If this bit is set, Hizunit will have LRA as replacement policy. The default value i.e. (when this bit is
reset) indicates the non-LRA eviction policy. For performance reasons, this bit must be reset.
Value Name Description Project
0h [Default] Non-LRA eviction Policy All
1h LRA eviction Policy All
Programming Notes Project
If this bit is set to "1", bit 3 of 0x7010h must also be set to "1"
11 Reserved
Format: MBZ
8:7 Sampler Cache Set XOR selection
Project: All
Format: U2
These bits have an impact only when the Sampler cache is configured in 16 way set associative mode.
If the cache is being used for immediate data or for blitter data these bits have no effect.
Value Name Description Project
00b None No XOR. All
01b Scheme 1 All
New_set_mask[3:0] = Tiled_address[16:13].
New_set[3:0] less than or = New_set_mask[3:0] ^Old_set[3:0].
Rationale: These bits can distinguish among 16 different equivalent
classes of virtual pages. These bits also represent the lsb for tile rows
ranging from a pitch of 1 tile to 16 tiles.
10b Scheme 2 All
New_set_mask[3] = Tiled_address[17] ^ Tiled_address[16].
New_set_mask[2] = Tiled_address[16] ^ Tiled_address[15].
New_set_mask[1] = Tiled_address[15] ^ Tiled_address[14].
New_set_mask[0] = Tiled_address[14] ^ Tiled_address[13].
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 66
Cache_Mode_1 - Cache Mode Register 1
New_set[3:0] less than or = New_set_mask[3:0] ^ Old_set[3:0].
Rationale: More bits on each XOR can give better statistical uniformity on
sets and since two lsbs are taken for each tile row size, it reduces the
chance of aliasing on sets.
11b Scheme 3
[Default] New_set_mask[3] = Tiled_address[22] ^ Tiled_address[21] ^
Tiled_address[20] ^ Tiled_address[19].
New_set_mask[2] = Tiled_address[18] ^ Tiled_address[17] ^
Tiled_address[16].
New_set_mask[1] = Tiled_address[15] ^ Tiled_address[14].
New_set_mask[0] = Tiled_address[13].
New_set[3:0] less than or = New_set_mask[3:0] ^ Old_set[3:0].
Rationale: More bits on each XOR can give better statistical uniformity on
sets and since each XOR has different bits, it reduces the chance of
aliasing on sets even more.
6 Pixel Backend sub-span collection Optimization Disable
Format: Disable
Value Name Description
0h [Default] Enables two contiguous quads to be collected as 4X2 access for RCZ interface. This
allows for less bank collision and less RAM power on RCZ.
1h Disables this optimization and therefore only one valid sub-span is sent to RCZ on the
4X2 interface.
Programming Notes Project
This bit must be set.
5 MCS Cache Disable
Format: Disable
For Programming restrictions please refer to the 3D Pipeline.
Value Name Description
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 67
Cache_Mode_1 - Cache Mode Register 1
0h [Default] MCS cache enabled. It allows RTs with MCS buffer enabled to be rendered using
either MSAA compression for MSRT OR with color clear feature for non MSRT.
1h MCS cache is disabled. Hence no MSAA compression for MSRT and no color clear for
non-MSRT.
4 Reserved
Format: MBZ
3 Depth Read Hit Write-Only Optimization Disable
Format: Disable
Description Project
This bit must always be reset to "0".
Value Name Description Project
0h [Default] Read Hit Write-only optimization is enabled in the Depth cache (RCZ).
1h Read Hit Write-only optimization is disabled in the Depth cache (RCZ).
2:1 Reserved
1.1.10.8 GAFS_MODE — Mode Register for GAFS
GAFS_MODE - Mode Register for GAFS
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 0212Ch
DWord Bit Description
0 31:16 Mask Bits
Format: Mask[15:0]
Masks: These bits serve as write enables for bits 15:0. If this register is written with any of these bits
clear the corresponding bit in the field 15:0 will not be modified. Reading these bits always returns 0s.
15:10 Reserved
Format: MBZ
8:2 Reserved
Format: MBZ
0 Selection of Arbitration for GAFS
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 68
GAFS_MODE - Mode Register for GAFS
Format: MBZ
GAFS data return policy from FFROB is a round-robin. This bit freezes the round robin to FF pipeline
order.
1.1.10.9 INSTPM—Instruction Parser Mode Register
INSTPM - Instruction Parser Mode Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 020C0h
The INSTPM register is used to control the operation of the Instruction Parser. Certain classes of instructions can be
disabled (ignored) – often useful for detecting performance bottlenecks. Also, Synchronizing Flush operations can be
initiated – useful for ensuring the completion (vs. only parsing) of rendering instructions.
Programming Notes
 If an instruction type is disabled, the parser will read those instructions but not process them.
Error checking will be performed even if the instruction is ignored.
All Reserved bits are implemented.
This Register is saved and restored as part of Context.
DWor
d Bit Description
0 31:1 Mask Bits
6 Format: Mask[15:0]
Masks: These bits serve as write enables for bits 15:0. If this register is written with any of these bits clear
the corresponding bit in the field 15:0 will not be modified. Reading these bits always returns 0s.
14:1 Reserved
3
Format: MBZ
11 CLFLUSH Toggle
Format: U1
This bit changes polarity each time the MI_CLFLUSH command completes.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 69
INSTPM - Instruction Parser Mode Register
10 Reserved
Format: MBZ
9 TLB Invalidate
Format: U1
If set, this bit allows the command stream engine to invalidate the 3D render TLBs. This bit is valid only
with the Sync flush enable. Note: GFX soft resets do not invalidate TLBs, it is up to GFX driver to explicitly
invalidate TLBs post reset.
8 Memory Sync Enable
Format: U1
If set, this bit allows the command stream engine to write out the data from the local caches to memory.
This bit is valid only with the Sync flush enable
7 Force Sync Command Ordering
Format: Enable
Description Project
By default, driver/OS synchronization commands (MI_STORE_DATA_IMM, for instance) can
execute out of order with respect to 3D state and 3D primitive commands. When set, this bit forces
ordering of these commands. See section 3.2.2 for a list of these commands.
[This bit should be programmed to 1.
Value Name
0b [Default]
1b
6 CONSTANT_BUFFER Address Offset Disable
Format: Disable
When this bit is clear, the 3DSTATE_CONSTANT_* Buffers’ Starting Address is used as a
DynamicStateOffset. That is, it serves as an offset from the Dynamic State Base Address. Accesses will be
subject to Dynamic State bounds checking.
When this bit is set, the 3DSTATE_CONSTANT_* Buffers’ Starting Address is used as a true
GraphicsAddress (not an offset). No bounds checking will be performed during access.
5 Sync Flush Enable
Format: U1
This field is used to request a Sync Flush operation. The device will automatically clear this bit before
completing the operation. See Sync Flush (Programming Environment).
Programming Notes Project
 The command parser must be stopped prior to issuing this command by setting
the Stop Rings bit in register MI_MODE. Only after observing Rings Idle set in
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 70
INSTPM - Instruction Parser Mode Register
MI_MODE can a Sync Flush be issued by setting this bit. Once this bit becomes
clear again, indicating flush complete, the command parser is re-enabled by
clearing Stop Rings.
Workaround :
Write 0x2050[31:0] = 0x00010001 (disable sequence)
Write 0x2700[31:0] = 0x00000000 (Wake up CS but don't do anything)
Poll 0x22AC[3:0] = 0 (Guarantees render pipe is awake)
VT-d request(Sync Flush) (Normal VT-d cycles(Replace with Sync Flush Steps)
Write 0x2050[31:0] = 0x00010000 ( Enable sequence (to enter RC6) )
3 Media Instruction Disable
Format: U1
This bit instructs the Renderer instruction parser to parse and error-check Media instructions, but not
execute them. Format = Disable
2 3D Rendering Instruction Disable
Format: U1
This bit instructs the Renderer instruction parser to parse and error-check 3D Rendering instructions, but
not execute them. This bit must always be set by software if 3D State Instruction Disable is set. Setting
this bit without setting 3D State Instruction Disable is allowed.
Format = Disable
1 3D State Instruction Disable
Format: Disable
0 Texture Palette Load Instruction Disable
Format: U1
This bit instructs the Renderer instruction parser to parse and error-check Texture Palette Load
instructions, but not execute them. Format = Disable
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 71
1.1.10.10 EXCC—Execute Condition Code Register
EXCC - Execute Condition Code Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W,RO
Size (in bits): 32
Trusted Type: 1
Address: 02028h
This register contains user defined and hardware generated conditions that are used by MI_WAIT_FOR_EVENT
commands. An MI_WAIT_FOR_EVENT instruction excludes the executing ring from arbitration if the selected event
evaluates to a 1, while instruction is discarded if the condition evaluates to a 0. Once excluded a ring is enabled into
arbitration when the selected condition evaluates to a 0.
This register also contains control for the invalidation of indirect state pointers on context restore.
DWord Bit Description
0 31:16 Mask Bits
Format: Mask[15:0]
These bits serves as a write enable for bits 15:0. If this register is written with any of these bits clear
the corresponding bit in the field 15:0 will not be modified. Reading these bits always returns 0s.
15:12 Reserved
Format: MBZ
11 Pending Indirect State Dirty Bit
This field keeps track of whether or not an indirect state pointer command has been parsed in the
current context. Clears either on a context save or explicitly through a flush command.
10:7 Pending Indirect State Counter
This field keeps track of the maximum number of indirect state pointers pending in the system.
When the register is saved/restored, it saves either a value of 1 or 0. This field is Read-Only.
6:5 Reserved
Format: MBZ
4:0 User Defined Condition Codes
The software may signal a Stream Semaphore by setting the Mask bit and Signal Bit together to
match the bit field specified in a WAIT_FOR_EVENT (Semaphore).
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 72
1.1.10.11 NOPID — NOP Identification Register
NOPID - NOP Identification Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Size (in bits): 32
Trusted Type: 1
Address: 02094h
Description Project
Access: RW
The NOPID register contains the Noop Identification value specified by the last MI_NOOP instruction that
enabled this register to be updated.
DWord Bit Description
0 31:22 Reserved
Format: MBZ
1.1.10.12 FBC RT BASE ADDRESS REGISTER
FBC_RT_BASE_ADDR_REGISTER -
FBC_RT_BASE_ADDR_REGISTER
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: Read/32 bit Write Only
Size (in bits): 32
Address: 07020h
This Register is saved and restored as part of Context.
DWord Bit Description
0 31:12 FBC RT Base Address
Format: PPGraphicsAddress[31:12]
4KB aligned Base Address as mapped in the PPGTT or in the GGTT for the render target.
Must be set to modify corresponding data bit. Reads to this field returns zero. This base address must
be the one that is either front buffer or the back-buffer (a flip target). It can be only programmed once
per context. It must be programmed before any draw call binding that render target base address.
11:2 Reserved
Format: MBZ
1 FBC Front Buffer Target
Format: Enable
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 73
FBC_RT_BASE_ADDR_REGISTER -
FBC_RT_BASE_ADDR_REGISTER
Value Name Description Project
0h [Default] FBC is targeting the Back Buffer for compression. This buffer can be cached in
the MLC/LLC, so a GFDT flush is required before FBC can begin compression.
1h FBC is targeting the Font Buffer for compression. This buffer cannot be cached
in the MLC/LLC. FBC compression can begin after any RC flush.
0 PPGTT Render Target Base Address Valid for FBC
Access: None
Exists If: Always
Format: Enable
Format: GraphicsAddress[31:0]U32
Value Name Description
0h [Default] Base address in this register [31:12] is not valid and therefore FBC will not get
any modifications from rendering.
1h Base address in this register [31:12] is valid and HW needs to compare the
current render target base address with this base address to provide
modifications to FBC.
Programming Notes
Workaround : Do not enable Render Command Streamer tracking for FBC.
Instead insert a LRI to address 0x50380 with data 0x00000004 after the
PIPE_CONTROL that follows each render submission.
1.1.10.13 RVSYNC – Render/Video Semaphore Sync Register
RVSYNC - Render/Video Semaphore Sync Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 02040h
This register is written by VCS, read by CS.
DWord Bit Description
0 31:0 Semaphore Data
Semaphore data for synchronization between render engine and blitter engine.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 74
1.1.10.14 RBSYNC – Render/Blitter Semaphore Sync Register
RBSYNC - Render/Blitter Semaphore Sync Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 02044h
This register is written by BCS, read by CS.
DWord Bit Description
0 31:0 Semaphore Data
Semaphore data for synchronization between render engine and blitter engine.
1.1.11 RINGBUF — Ring Buffer Registers
See the “Device Programming Environment” chapter for detailed information on these registers
1.1.11.1 RING_BUFFER_TAIL
RING_BUFFER_TAIL - Ring Buffer Tail
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: R/W
Address: 02030h
Name: RCS Ring Buffer Tail
ShortName: RCS_RING_BUFFER_TAIL
Address: 12030h
Name: VCS Ring Buffer Tail
ShortName: VCS_RING_BUFFER_TAIL
Address: 22030h
Name: BCS Ring Buffer Tail
ShortName: BCS_RING_BUFFER_TAIL
These registers are used to define and operate the “ring buffer” mechanism which can be used to pass instructions
to the command interface. The buffer itself is located in a linear memory region. The ring buffer is defined by a 4
Dword register set that includes starting address, length, head offset, tail offset, and control information.
Refer to the Programming Interface chapter for a detailed description of the parameters specified in this ring buffer
register set, restrictions on the placement of ring buffer memory, arbitration rules, and in how the ring buffer can be
used to pass instructions.
Ring Buffer Tail Offsets must be properly programmed before ring is enabled. A Ring Buffer can be enabled when
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 75
RING_BUFFER_TAIL - Ring Buffer Tail
empty.
DWord Bit Description
0 31:21 Reserved
Format: MBZ
20:3 Tail Offset
Format: GraphicsAddress[20:3]
This field is written by software to specify where the valid instructions placed in the ring buffer end. The
value written points to the QWord past the last valid QWord of instructions. In other words, it can be
defined as the next QWord that software will write instructions into.
Software must write subsequent instructions to QWords following the Tail Offset, possibly wrapping
around to the top of the buffer (i.e., software can’t skip around within the buffer).
Note that all DWords prior to the location indicated by the Tail Offset must contain valid instruction
data – which may require instruction padding by software. See Head Offset for more information.
2:0 Reserved
Format: MBZ
1.1.11.2 RING_BUFFER_HEAD
RING_BUFFER_HEAD - Ring Buffer Head
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: R/W
Address: 02034h
Name: RCS Ring Buffer Head
ShortName: RCS_RING_BUFFER_HEAD
Address: 12034h
Name: VCS Ring Buffer Head
ShortName: VCS_RING_BUFFER_HEAD
Address: 22034h
Name: BCS Ring Buffer Head
ShortName: BCS_RING_BUFFER_HEAD
This register is used to define and operate the ring buffer mechanism which can be used to pass instructions to the
command interface. The buffer itself is located in a physical memory region. The ring buffer is defined by a 4 Dword
register set that includes starting address, length, head offset, tail offset, and control information. Refer to the
Programming Interface chapter for a detailed description of the parameters specified in this ring buffer register set,
restrictions on the placement of ring buffer memory, arbitration rules, and in how the ring buffer can be used to pass
instructions.
Ring Buffer Head Offsets must be properly programmed before ring is enabled. A Ring Buffer can be
enabled when empty.
DWord Bit Description
0 31:21 Wrap Count
Format: U11 count of ring buffer wraps
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 76
RING_BUFFER_HEAD - Ring Buffer Head
This field is incremented by 1 whenever the Head Offset wraps from the end of the buffer back to the
start (i.e., whenever it wraps back to 0). Appending this field to the Head Offset field effectively creates
a virtual 4GB Head “Pointer” which can be used as a tag associated with instructions placed in a ring
buffer. The Wrap Count itself will wrap to 0 upon overflow.
20:2 Head Offset
Format: GraphicsAddress[20:2] DWord Offset
This field indicates the offset of the next instruction DWord to be parsed. Software will initialize this field
to select the first DWord to be parsed once the RB is enabled. (Writing the Head Offset while the RB is
enabled is UNDEFINED). Subsequently, the device will increment this offset as it executes instructions
– until it reaches the QWord specified by the Tail Offset. At this point the ring buffer is considered
“empty”.
Programming Notes
A RB can be enabled empty or containing some number of valid instructions.
1 Reserved
Format: MBZ
0 Wait for Condition Indicator
Source: RenderCS
This is a read only value used to indicate whether or not the command streamer is currently waiting for
a conditional code to be cleared from 0x2028
0 Reserved
Source: BlitterCS, VideoCS, VideoCS2, VideoEnhancementCS
Format: MBZ
1.1.11.3 RING_BUFFER_START
RING_BUFFER_START - Ring Buffer Start
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: R/W
Address: 02038h
Name: RCS Ring Buffer Start
ShortName: RCS_RING_BUFFER_START
Address: 12038h
Name: VCS Ring Buffer Start
ShortName: VCS_RING_BUFFER_START
Address: 22038h
Name: BCS Ring Buffer Start
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 77
RING_BUFFER_START - Ring Buffer Start
ShortName: BCS_RING_BUFFER_START
These registers are used to define and operate the “ring buffer” mechanism which can be used to pass instructions
to the command interface. The buffer itself is located in a physical memory region. The ring buffer is defined by a 4
Dword register set that includes starting address, length, head offset, tail offset, and control information. Refer to the
Programming Interface chapter for a detailed description of the parameters specified in this ring buffer register set,
restrictions on the placement of ring buffer memory, arbitration rules, and in how the ring buffer can be used to pass
instructions.
DWord Bit Description
0 31:12 Starting Address
Format: GraphicsAddress[31:12]RingBuffer
This field specifies Bits 31:12 of the 4KB-aligned starting Graphics Address of the ring buffer.
Address bits 31 down to 29 must be zero.
All ring buffer pages must map to Main Memory (uncached) pages. Ring Buffer addresses are
always translated through the global GTT.
11:0 Reserved
Format: MBZ
1.1.11.4 RING_BUFFER_CONTROL
RING_BUFFER_CTL - Ring Buffer Control
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: R/W
Address: 0203Ch
Name: RCS Ring Buffer Control
ShortName: RCS_RING_BUFFER_CTL
Address: 1203Ch
Name: VCS Ring Buffer Control
ShortName: VCS_RING_BUFFER_CTL
Address: 2203Ch
Name: BCS Ring Buffer Control
ShortName: BCS_RING_BUFFER_CTL
These registers are used to define and operate the ring buffer mechanism which can be used to pass instructions to
the command interface. The buffer itself is located in a physical memory region. The ring buffer is defined by a 4
Dword register set that includes starting address, length, head offset, tail offset, and control information. Refer to the
Programming Interface chapter for a detailed description of the parameters specified in this ring buffer register set,
restrictions on the placement of ring buffer memory, arbitration rules, and in how the ring buffer can be used to pass
instructions.
Ring Buffer Head and Tail Offsets must be properly programmed before it is enabled. A Ring Buffer can be
enabled when empty.
DWord Bit Description
0 31:21 Reserved
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 78
RING_BUFFER_CTL - Ring Buffer Control
Format: MBZ
20:12 Buffer Length
Format: U9-1 in 4 KB pages – 1
This field is written by SW to specify the length of the ring buffer in 4 KB Pages.Range = [0 = 1 page =
4 KB, 1FFh = 512 pages = 2 MB]
Value Name Description
0 1 page = 4 KB
1FFh 512 pages = 2 MB
11 RBWait
Indicates that this ring has executed a WAIT_FOR_EVENT instruction and is currently waiting.
Software can write a “1” to clear this bit, write of “0” has no effect. When the RB is waiting for an event
and this bit is cleared, the wait will be terminated and the RB will be returned to arbitration.
10 Semaphore Wait
Description Project
Indicates that this ring has executed a MI_SEMAPHORE_MBOX instruction with register
compare and is currently waiting.
9 Reserved
Format: MBZ
8 Reserved
Source: RenderCS, BlitterCS
Format: MBZ
8 Disable Register Accesses
Source: VideoCS, VideoCS2, VideoEnhancementCS
Value Name Description
0 R/W Ring is allowed to access (read or write) MMIO space.
1 Read Only Ring is not allowed to write MMIO space. Ring is allowed to read registers.
7:3 Reserved
Format: MBZ
2:1 Automatic Report Head Pointer
Source: BlitterCS, VideoCS, VideoCS2, VideoEnhancementCS
Description Project
This field is written by software to control the automatic “reporting” (write) of this ring buffer’s
“Head Pointer” register (register DWord 1) to the corresponding location within the Hardware
Status Page. Automatic reporting can either be disabled or enabled at 4KB, 64KB or 128KB
boundaries within the ring buffer.
The head pointer will be reported to the head pointer location in the Per-Process Hardware
Status Page when it passes each 4KB page boundary. When the above-mentioned bit is set,
reporting will behave just as on the prior devices (as documented above), and option 2 is not
legal.
Value Name Description
0 MI_AUTOREPORT_OFF Automatic reporting disabled
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 79
RING_BUFFER_CTL - Ring Buffer Control
1 MI_AUTOREPORT_64KB Report every 16 pages (64KB)
2 MI_AUTOREPORT_4KB Report every page (4KB)This mode must not be enabled in Ring
Buffer mode of scheduling to minimize the auto reports.
3 MI_AUTOREPORT_128KB Report every 32 pages (128KB)
2:1 Reserved
Source: RenderCS
Format: MBZ
0 Ring Buffer Enable
Format: Enable
This field is used to enable or disable this ring buffer. It can be enabled or disabled regardless of
whether there are valid instructions pending. If disabled and the ring head equals ring tail, all state
currently loaded in hardware is considered invalid.
Programming Notes Project
SW should follow the below programming notes while enabling render engine's ring buffer for
the first time, this would be coming out of boot, standby, hibernate or reset.
SW should set the Force Wakeup bit to prevent GT from entering C6.
SW should dispatch workload (dummy) to initialize render engine with default state such that
any context switches that occur subsequently (Power Save) will save and restore coherent
device state. Indirect pointers used in 3D states should point to valid graphics surface existing
in memory. PP_DCLV followed by PP_DIR_BASE register should be programmed as part of
initialization workload if PPGTT is enabled in GFX_MODE register.
Once the render engine is programmed with valid state and the configuration, Force Wakeup
bit should be reset to enable C6 entry.
1.1.11.5 UHPTR — Pending Head Pointer Register
UHPTR - Pending Head Pointer Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: R/W
Address: 02134h
Name: RCS Pending Head Pointer Register
ShortName: RCS_UHPTR
Address: 12134h
Name: VCS Pending Head Pointer Register
ShortName: VCS_UHPTR
Address: 22134h
Name: BCS Pending Head Pointer Register
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 80
UHPTR - Pending Head Pointer Register
ShortName: BCS_UHPTR
Programming Notes
Once SW uses UHPTR to preempt the existing workload, should explicitly program MI_SET_CONTEXT to save the
preempted context status before submitting the new workload. In case SW doesn’t want to save the state of the
preempted context, it should at the minimum program RS_PREEMPT_STATUS to 0x0 so that the register status
doesn’t interfere with the new workloads.
DWord Bit Description
0 31:3 Head Pointer Address
Format: GraphicsAddress[31:3]
This register represents the GFX address offset where execution should continue in the ring buffer
following execution of an MI_ARB_CHECK command.
2:1 Reserved
Format: MBZ
0 Head Pointer Valid
This bit is set by the software to request a pre-emption.
It is reset by hardware when an MI_ARB_CHECK command is parsed by the command streamer.
The hardware uses the head pointer programmed in this register at the time the reset is generated.
Value Name Description
0 InValid No valid updated head pointer register, resume execution at the current location in the
ring buffer
1 Valid Indicates that there is an updated head pointer programmed in this register
1.1.12 Watchdog Timer Registers
These 2 registers together implement a watchdog timer. Writing ones to the control register enables the
counter, and writing zeroes disables the counter. The 2nd register is programmed with a threshold value
which, when reached, signals an interrupt then resets the counter to 0. Program the threshold value
before enabling the counter or extremely frequent interrupts may result.
Note that the counter itself is not observable. It increments with the main render clock.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 81
1.1.12.1 PR_CTR_CTL—Render Watchdog Counter Control
PR_CTR_CTL - Render Watchdog Counter Control
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02178h
DWord Bit Description
0 31 Count Select
Format: select
0 – Use the timestamp to increment the watchdog count (every 640ns)1 – Use the fixed function clock
(csclk) to increment the watchdog count
30:0 Counter Logic Op
This field specifies the action to be taken by the clock counter to generate interrupts. Writing 0 into this
register causes a core render clock counter to be kicked off. Writing 1 into this register causes a core
render clock counter to be stopped and reset to 0.
1.1.12.2 PR_CTR_THRSH—Render Watchdog Counter Threshold
PR_CTR_THRSH - Render Watchdog Counter Threshold
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00150000
Access: R/W
Size (in bits): 32
Address: 0217Ch
DWord Bit Description
0 31:0 Counter logic Threshold
Default Value: 00150000h
Format: U32
This field specifies the threshold that the hardware checks against for the value of the render clock
counter before generating an interrupt. The counter in hardware generates an interrupt when the
threshold is reached, rolls over and starts counting again. The interrupt generated is the “Media Hang
Notify” interrupt since this watchdog timer is intended primarily to remedy VLD hangs on the main
pipeline.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 82
1.1.12.3 PR_CTR—Render Watchdog Counter
PR_CTR - Render Watchdog Counter
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 02190h
DWord Bit Description
0 31:0 Counter Value
Format: U32
This register reflects the render watchdog counter value itself. It cannot be written to.
1.1.13 Interrupt Control Registers
The Interrupt Control Registers described in this section all share the same bit definition. The bit definition
is as follows:
Bit Definition for Interrupt Control Registers
Source: RenderCS
Default Value: 0x00000000
DWord Bit Description
0 31:12 Reserved
Format: MBZ
Reserved for other command streamers - can not be allocated by main command streamer.
11:10 Reserved
Format: MBZ
These bits may be assigned to interrupts on future products/steppings.
9 Performance Monitoring Buffer Half-Full Interrupt
For internal trigger (timerbased) based reporting, if the report buffer crosses half full limit, this interrupt
is generated.
8 Context Switch Interrupt
7 Page Fault
Project: All
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 83
Bit Definition for Interrupt Control Registers
Description Project
This bit is set whenever there is a pending GGTT/PPGTT (page or directory) fault in Render
command streamer.
6 Timeout Counter Expired
Set when the render pipe timeout counter (0x02190) has reached the timeout thresh-hold value
(0x0217c).
5 Reserved
Format: MBZ
4 PIPE_CONTROL Notify Interrupt
The Pipe Control packet (Fences) specified in 3D pipeline document may optionally generate an
Interrupt. The Store QW associated with a fence is completed ahead of the interrupt.
3 Render Command Parser Master Error
When this status bit is set, it indicates that the hardware has detected an error. It is set by the device
upon an error condition and cleared by a CPU write of a one to the appropriate bit contained in the
Error ID register followed by a write of a one to this bit in the IIR. Further information on the source of
the error comes from the "Error Status Register" which along with the "Error Mask Register" determine
which error conditions will cause the error status bit to be set and the interrupt to occur.
Page Table Error: Indicates a page table error.
Instruction Parser Error: The Render Instruction Parser encounters an error while parsing an
instruction.
2 Sync Status
This bit is set in the Hardware Status Page DW offset 0 when the Instruction Parser completes a flush
with the sync enable bit active in the INSTPM register. The toggle event will happen after the render
engine is flushed. The HW Status DWord write resulting from this toggle will cause the CPU’s view of
graphics memory to be coherent as well (flush and invalidate the render cache). It is the driver's
responsibility to clear this bit before the next sync flush with HWSP write enabled.
0 Render Command Parser User Interrupt
This status bit is set when an MI_USER_INTERRUPT instruction is executed on the Render Command
Parser. Note that instruction execution is not halted and proceeds normally. A mechanism such as an
MI_STORE_DATA instruction is required to associate a particular meaning to a user interrupt.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 84
The following table specifies the settings of interrupt bits stored upon a “Hardware Status Write” due to
ISR changes:
ISR bit Reporting via Hardware
Status Write (when unmasked
Bit Interrupt Bit via HWSTAM)
Set when event occurs, cleared
9 Performance Monitoring Buffer Half-Full Interrupt
when event cleared
Not supported to be unmasked
8 Context Switch Interrupt: Set when a context switch has just occurred.
7 Page Fault: This bit is set whenever there is a pending PPGTT (page or Set when event occurs, cleared
directory) fault. when event cleared
Not supported to be unmasked
6 Media Decode Pipeline Counter Exceeded Notify Interrupt: The
counter threshold for the execution of the media pipeline is exceeded.
Driver needs to attempt hang recovery.
5 L3 Parity interrupt
PIPE_CONTROL packet - Notify Enable 0
4
Master Error Set when error occurs, cleared
3
when error cleared
Sync Status Toggled every SyncFlush Event
2
1 Reserved
User Interrupt 0
0
1.1.13.1 HWSTAM — Hardware Status Mask Register
HWSTAM - Hardware Status Mask Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0xFFFFFFFF
Access: R/W,RO
Size (in bits): 32
Trusted Type: 1
Address: 02098h
The HWSTAM register has the same format as the Interrupt Control Registers. The bits in this register are mask bits
that prevent the corresponding bits in the Interrupt Status Register from generating a Hardware Status Write (PCI
write cycle). Any unmasked interrupt bit (HWSTAM bit set to 0) will allow the Interrupt Status Register to be written to
the ISR location (within the memory page specified by the Hardware Status Page Address Register) when that
Interrupt Status Register bit changes state.
Programming Notes
 To write the interrupt to the HWSP, the corresponding IMR bit must also be clear (enabled).
 At most 1 bit can be unmasked at any given time.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 85
DWord Bit Description
0 31:0 Hardware Status Mask Register
Default Value: FFFFFFFFh
Format: Array of Masks
Refer to the Interrupt Control Register section for bit definitions. Reserved bits are RO.
1.1.13.2 IMR—Interrupt Mask Register
IMR - Interrupt Mask Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0xFFFFFFFF
Access: R/W,RO
Size (in bits): 32
Address: 020A8h
The IMR register is used by software to control which Interrupt Status Register bits are masked or unmasked.
Unmasked bits will be reported in the IIR, possibly triggering a CPU interrupt, and will persist in the IIR until cleared
by software. Masked bits will not be reported in the IIR and therefore cannot generate CPU interrupts.
DWord Bit Description
0 31:0 Interrupt Mask Bits
Format: InterruptMask[32] Refer to the Interrupt Control Register section for bit definitions.
This field contains a bit mask which selects which interrupt bits (from the ISR) are reported in the IIR.
Reserved bits in the Interrupt Control Register are RO.
Value Name Description
FFFF FFFFh [Default]
0h Not Masked Will be reported in the IIR
1h Masked Will not be reported in the IIR
1.1.13.3 Hardware-Detected Error Bit Definitions (for EIR, EMR, ESR)
This section defines the Hardware-Detected Error bit definitions and ordering that is common to the EIR,
EMR and ESR registers. The EMR selects which error conditions (bits) in the ESR are reported in the
EIR. Any bit set in the EIR will cause the Master Error bit in the ISR to be set. EIR bits will remain set until
the appropriate bit(s) in the EIR is cleared by writing the appropriate EIR bits with ‘1’ (except for the
unrecoverable bits described below).
The following table describes the Hardware-Detected Error bits:
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 86
Hardware-Detected Error Bit Definitions
Source: RenderCS
Default Value: 0x00000000
DWord Bit Description
0 31:3 Reserved
Format: MBZ
2 Reserved
Format: MBZ
1 Reserved
Format: MBZ
0 Instruction Error
This bit is set when the Renderer Instruction Parser detects an error while parsing an instruction.
Instruction errors include:
Client ID value (Bits 31:29 of the Header) is not supported (only MI, 2D and 3D are supported).
Defeatured MI Instruction Opcodes:
Value Name Description
1 Instruction Error detected
Programming Notes
This error indications cannot be cleared except by reset (i.e., it is a fatal error).
1.1.13.3.1 EIR — Error Identity Register
EIR - Error Identity Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W,RO
Size (in bits): 32
Address: 020B0h
The EIR register contains the persistent values of Hardware-Detected Error Condition bits. Any bit set in this register
will cause the Master Error bit in the ISR to be set. The EIR register is also used by software to clear detected errors
(by writing a 1 to the appropriate bit(s)), except for the unrecoverable bits described.)
DWord Bit Description
0 31:16 Reserved
Format: MBZ
15:0 Error Identity Bits
Format: Array of Error condition bits See the table titled Hardware-Detected Error Bits.
This register contains the persistent values of ESR error status bits that are unmasked via the EMR
register. (See Table Table 3-3. Hardware-Detected Error Bits). The logical OR of all (defined) bits in
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 87
EIR - Error Identity Register
this register is reported in the Master Error bit of the Interrupt Status Register. In order to clear an error
condition, software must first clear the error by writing a 1 to the appropriate bit(s) in this field. If
required, software should then proceed to clear the Master Error bit of the IIR. Reserved bits are RO.
Value Name
1h Error occurred
Programming Notes
Writing a 1 to a set bit will cause that error condition to be cleared. However, neither the Page Table
Error bit (Bit 4) nor the Instruction Error bit (Bit 0) can be cleared except by reset (i.e., it is a fatal error).
1.1.13.3.2 EMR—Error Mask Register
EMR - Error Mask Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x000000FF
Access: R/W,RO
Size (in bits): 32
Address: 020B4h
The EMR register is used by software to control which Error Status Register bits are masked or unmasked.
Unmasked bits will be reported in the EIR, thus setting the Master Error ISR bit and possibly triggering a CPU
interrupt, and will persist in the EIR until cleared by software. Masked bits will not be reported in the EIR and
therefore cannot generate Master Error conditions or CPU interrupts. Reserved bits are RO.
DWord Bit Description
0 31:8 Reserved
Format: Must Be One
Programming Notes
These bits are not implemented in HW and must be set to ’1’
7:0 Error Mask Bits
Format: Array of error condition mask bits See the table titled Hardware-Detected Error Bits.
This register contains a bit mask that selects which error condition bits (from the ESR) are
reported in the EIR.
Value Name Description
FFh [Default]
0h Not Masked Will be reported in the EIR
1h Masked Will not be reported in the EIR
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 88
1.1.13.3.3 ESR—Error Status Register
ESR - Error Status Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 020B8h
The ESR register contains the current values of all Hardware-Detected Error condition bits (these are all by definition
persistent). The EMR register selects which of these error conditions are reported in the persistent EIR (i.e., set bits
must be cleared by software) and thereby causing a Master Error interrupt condition to be reported in the ISR.
DWord Bit Description
0 31:16 Reserved
Format: MBZ
15:0 Error Status Bits
Format: Array of error condition bits See the table titled Hardware-Detected Error Bits.
This register contains the non-persistent values of all hardware-detected error condition bits.
Value Name
1h Error Condition Detected
1.1.14 Logical Context Support
1.1.14.1 BB_ADDR — Batch Buffer Head Pointer
BB_ADDR - Batch Buffer Head Pointer Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 02140h
Name: RCS Batch Buffer Head Pointer Register
ShortName: RCS_BB_ADDR
Address: 12140h
Name: VCS Batch Buffer Head Pointer Register
ShortName: VCS_BB_ADDR
Address: 1A140h
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 89
BB_ADDR - Batch Buffer Head Pointer Register
Name: VECS Batch Buffer Head Pointer Register
ShortName: VECS_BB_ADDR
Address: 22140h
Name: BCS Batch Buffer Head Pointer Register
ShortName: BCS_BB_ADDR
This register contains the current DWord Graphics Memory Address of the last-initiated batch buffer.
Programming Notes
Programming Restriction: This register should NEVER be programmed by driver. This is for HW internal use only.
DWord Bit Description
0 31:3 Batch Buffer Head Pointer
Source: BlitterCS, VideoCS, VideoCS2, VideoEnhancementCS
Format: GraphicsAddress[31:3]
This field specifies the DWord-aligned Graphics Memory Address where the last initiated Batch Buffer is
currently fetching commands. If no batch buffer is currently active, the Valid bit will be 0 and this field will
be meaningless.
31:2 Batch Buffer Head Pointer
Source: RenderCS
Format: GraphicsAddress[31:2]
This field specifies the DWord-aligned Graphics Memory Address where the last initiated Batch Buffer is
currently fetching commands. If no batch buffer is currently active, the Valid bit will be 0 and this field will
be meaningless.
2 Reserved
Source: BlitterCS, VideoCS, VideoCS2, VideoEnhancementCS
Format: MBZ
1 Reserved
Format: MBZ
0 Valid
Format: U1
Value Name Description
0h Invalid [Default] Batch buffer Invalid
1h Valid Batch buffer Valid
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 90
1.1.14.2 BB_STATE – Batch Buffer State Register
BB_STATE - Batch Buffer State Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RO
Size (in bits): 32
Address: 02110h
Name: RCS Batch Buffer State Register
ShortName: RCS_BB_STATE
This register contains the attributes of the current batch buffer initiated from the Ring Buffer. These include the
security indicator.
This register should not be written by software. These fields should only get written by a context restore. Software
should always set these fields via the MI_BATCH_BUFFER_START command when initiating a batch buffer.
This register is saved and restored with context.
DWord Bit Description
0 31:9 Reserved
Format: MBZ
8 Reserved
Format: MBZ
7 Reserved
Format: MBZ
5 Address Space Indicator
Value Name Description
0h GGTT [Default] This batch buffer is located in GGTT memory
1h PPGTT This batch buffer is located in PPGTT memory.
3:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 91
1.1.14.3 CCID—Current Context Register
CCID - Current Context Register
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02180h
This register contains the current logical rendering context address associated with the ring buffer in ring buffer mode
of scheduling. This register contents are not valid in Exec-List mode of scheduling.
Programming Notes
The CCID register must not be written directly (via MMIO) unless the Command Streamer is completely idle (i.e., the
Ring Buffer is empty and the pipeline is idle). Note that, under normal conditions, the CCID register should only be
updated from the command stream using the MI_SET_CONTEXT command.
DWord Bit Description
0 31:12 Logical Render Context Address (LRCA)
Default Value: 0h
Format: GraphicsAddress[31:11]
This field contains the 4 KB-aligned Graphics Memory Address of the current Logical Rendering
Context. Bit 11 MBZ.
11:10 Reserved
Format: MBZ
8 Reserved
Format: Must Be One
7:4 Reserved
Format: MBZ
3 Extended State Save Enable
Format: Enable
If set, the extended state identified in the Logical Context Data section of the Memory Data Formats
chapter, is saved as part of switching away from this logical context.
2 Extended State Restore Enable
Format: Enable
If set, the extended state identified in the Logical Context Data section of the Memory Data Formats
chapter, was loaded (or restored) as part of switching to this logical context.
1 Reserved
Format: MBZ
0 Valid
Format: U1
Value Name Description
0h Invalid The other fields of this register are invalid. A switch away from the context will
[Default] not invoke a context save operation.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 92
CCID - Current Context Register
1h Valid The other fields of this register are valid, and a switch from the context will
invoke the normal context save/restore operations.
1.1.14.4 CXT_SIZE—Context Sizes
CXT_SIZE - Context Sizes
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x48A7B8CD
Access: R/W
Size (in bits): 32
Trusted Type: 1
Address: 021A8h
The actual size of a logical rendering context is the amount of data stored/restored during a context switch and is
measured in 64B cache lines. This register will be power context save/restored. Note that this register will default to
the correct value, so software should not have to modify it.
DWord Bit Description
0 31:25 Power Context Size
Default Value: 24h
This field indicates the Power context data that need to be save restored.
24:22 Ring Context Size
Default Value: 2h
This field indicates the Ring context data that need to be save restored.
21:16 Render Context Size
Default Value: 27h
This field indicates the render context data that need to be save restored when extended mode is not
enabled for a context.
15:9 Extended Context Size
Default Value: 5Ch
This field indicates the render context data that need to be save restored when extended mode is
enabled for a context. Note: Render context is subset of this context.
8:6 GT1 Mode
Default Value: 3h
This field indicates the amount of data that need not be save/restored from render context in GT1
mode. Note: This is the amount of data not save/restored from TDL and SC in GT1 mode.
5:0 VF State Context Size
Default Value: Dh
This field indicates the amount of VF unit data context save/restored in cachelines.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 93
1.1.14.5 MTCH_CID_RST – Matched Context ID Reset Register
MTCH_CID_RST - Matched Context ID Reset Register
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 0222Ch
This register is used to generate a Context ID specific reset (Render Only). To initiate a reset, the register is written
with the pending bit set. Hardware compares the current context ID with the register and on match generates a
Render Only reset. After reset is complete, HW clears the pending bit and can be programmed to generate an
interrupt. The match bit is set. If the current context ID does not match this register, the pending bit is reset and an
interrupt is generated. The match bit is reset.The match indicates the result of the last comparison, and its valid only
when pending bit is zero.Please see MCIDRST interrupt bit assignment in the Interrupt Control Registers.
DWord Bit Description
0 31:12 Match Context ID
Format: U20
Contains the context ID to be compared with the currently running context ID.
11:2 Reserved
Format: MBZ
1 Match
Format: U20
This bit indicates the result of the match operation; 1 means the Current Context ID matches the
Match Context ID field.
0 Pending
Format: U20
This bit indicates that a matched context ID reset is pending. The bit should be set when the register
is written (in order to have a pending MTCH_CID_RST request), and will be reset by hardware to
indicate that the operation is completed (Either with a match or mismatch)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 94
1.1.14.6 SYNC_FLIP_STATUS – Wait for Event and Display Flip Flags Register
SYNC_FLIP_STATUS - Wait For Event and Display Flip Flags
Register
Register Space: MMIO: 0/2/0
Project: All
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 022D0h
Name: RCS Wait For Event and Display Flip Flags Register
ShortName: RCS_SYNC_FLIP_STATUS
This register is the saved value of what wait for events are still valid. This register is part of context save and restore
for RC6 feature.
Programming Notes
Programming Restriction:
This register should NEVER be programmed by SW, this is for HW internal use only.
DWord Bit Description
0 31 Reserved
Format: MBZ
30 Display Plane A Asyncronous Display Flip Pending
Format: Enable
This field enables a wait for the duration of a Display Plane A “Flip Pending” condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
29 Display Plane A Syncronous Flip Display Pending
Format: Enable
This field enables a wait for the duration of a Display Plane A “Flip Pending” condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
28 Display Sprite A Syncronous Flip Display Pending
Format: Enable
This field enables a wait for the duration of a Display Sprite A “Flip Pending” condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition in the Device
Programming Interface chapter of MI Functions.
27 Reserved
Format: MBZ
26 Display Plane B Asyncronous Display Flip Pending
Format: Enable
This field enables a wait for the duration of a Display Plane B “Flip Pending” condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 95
SYNC_FLIP_STATUS - Wait For Event and Display Flip Flags
Register
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
25 Display Plane B Syncronous Flip Display Pending
Format: Enable
This field enables a wait for the duration of a Display Plane B Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
24 Display Sprite B Syncronous Flip Display Pending
Format: Enable
This field enables a wait for the duration of a Display Sprite B Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition in the Device
Programming Interface chapter of MI Functions.
23 Display Plane A Asyncronous Performance Flip Pending Wait Enable
Source: RenderCS
Format: Enable
This field enables a wait for the duration of a Display Plane A Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
22 Display Plane A Asyncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane A Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
21 Display Plane A Syncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane A Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
20 Display Sprite A Syncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Sprite A Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition in the Device
Programming Interface chapter of MI Functions.
19 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 96
SYNC_FLIP_STATUS - Wait For Event and Display Flip Flags
Register
18 Display Pipe A Scan Line Wait Enable
Format: Enable
This field enables a wait while a Display Pipe A Scan Line condition exists. This condition is defined as
the the start of the scan line specified in the Pipe A Display Scan Line Count Range Compare Register.
See Scan Line Event in the Device Programming Interface chapter of MI Functions.
17 Display Pipe A Vertical Blank Wait Enable
Format: Enable
This field enables a wait until the next Display Pipe A Vertical Blank event occurs. This event is defined as
the start of the next Display Pipe A vertical blank period. Note that this can cause a wait for up to an entire
refresh period. See Vertical Blank Event (See Programming Interface).
16 Display Pipe A H Blank Wait Enable
Format: Enable
This field enables a wait until the start of next Display Pipe A Horizontal Blank event occurs. This event is
defined as the start of the next Display A Horizontal blank period. Note that this can cause a wait for up to
a line. See Horizontal Blank Event in the Device Programming Interface chapter of MI Functions.
15 Display Plane B Asyncronous Performance Flip Pending Wait Enable
Source: RenderCS
Format: Enable
This field enables a wait for the duration of a Display Plane B Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
14 Display Plane B Asyncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane B Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
13 Display Plane B Syncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane B Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the Device
Programming Interface chapter of MI Functions.
12 Display Sprite B Syncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Sprite B Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address has
now been loaded into the active front buffer registers). See Display Flip Pending Condition in the Device
Programming Interface chapter of MI Functions.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 97
SYNC_FLIP_STATUS - Wait For Event and Display Flip Flags
Register
11 Reserved
Format: MBZ
10 Display Pipe B Scan Line Wait Enable
Format: Enable
This field enables a wait while a Display Pipe B Scan Line condition exists. This condition is defined as
the the start of the scan line specified in the Pipe B Display Scan Line Count Range Compare Register.
See Scan Line Event in the Device Programming Interface chapter of MI Functions.
9 Display Pipe B Vertical Blank Wait Enable
Format: Enable
This field enables a wait until the next Display Pipe B Vertical Blank event occurs. This event is defined as
the start of the next Display Pipe B vertical blank period. Note that this can cause a wait for up to an entire
refresh period. See Vertical Blank Event (See Programming Interface).
8 Display Pipe B H Blank Wait Enable
Format: Enable
This field enables a wait until the start of next Display Pipe B Horizontal Blank event occurs. This event is
defined as the start of the next Display B Horizontal blank period. Note that this can cause a wait for up to
a line. See Horizontal Blank Event in the Device Programming Interface chapter of MI Functions.
7:5 Reserved
Format: MBZ
4:0 Condition Code Wait Select
This field enables a wait for the duration that the corresponding condition code is active. These enable
select one of 15 condition codes in the EXCC register, that cause the parser to wait until that condition-
code in the EXCC is cleared.
Value Name Description
0h Not Enabled Condition Code Wait not enabled
1h-5h Enabled Condition Code select enabled; selects one of 5 codes, 0 – 4
6h-15h Reserved
Programming Notes
Note that not all condition codes are implemented. The parser operation is UNDEFINED if an
unimplemented condition code is selected by this field. The description of the EXCC register (Memory
Interface Registers) lists the codes that are implemented.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 98
1.1.14.7 SYNC_FLIP_STATUS_1 – Wait for Event and Display Flip Flags Register
1
SYNC_FLIP_STATUS_1 - Wait For Event and Display Flip Flags
Register 1
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 022D4h
Name: RCS Wait For Event and Display Flip Flags Register 1
ShortName: RCS_SYNC_FLIP_STATUS_1
This register is the saved value of what wait for events are still valid. This register is part of context save and restore
for RC6 feature.
DWord Bit Description
0 31:27 Reserved
Format: MBZ
26:15 Reserved
Format: MBZ
11 SyncFlush Status
Format: Enable
This field toggles on completion of sync flush. This bit toggle generates Interrupt and also reports
interrupt status to HWSP on sync flush done.
10 Display Plane C Asyncronous Display Flip Pending
Format: Enable
This field enables a wait for the duration of a Display Plane C Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the
Device Programming Interface chapter of MI Functions.
9 Display Plane C Syncronous Flip Display Pending
Format: Enable
This field enables a wait for the duration of a Display Plane C Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the
Device Programming Interface chapter of MI Functions.
8 Display Sprite C Syncronous Flip Display Pending
Format: Enable
This field enables a wait for the duration of a Display Sprite C Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers). See Display Flip Pending Condition in the
Device Programming Interface chapter of MI Functions.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 99
SYNC_FLIP_STATUS_1 - Wait For Event and Display Flip Flags
Register 1
7 Display Plane C Asyncronous Performance Flip Pending Wait Enable
Source: RenderCS
Format: Enable
This field enables a wait for the duration of a Display Plane C Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the
Device Programming Interface chapter of MI Functions.
6 Display Plane C Asyncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane C “Flip Pending” condition. If a flip request
is pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the
Device Programming Interface chapter of MI Functions.
5 Display Plane C Syncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane C Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers). See Display Flip Pending Condition (in the
Device Programming Interface chapter of MI Functions.
4 Display Sprite C Syncronous Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Sprite C Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers). See Display Flip Pending Condition in the
Device Programming Interface chapter of MI Functions.
3 Reserved
Format: MBZ
2 Display Pipe C Scan Line Wait Enable
Format: Enable
This field enables a wait while a Display Pipe C Scan Line condition exists. This condition is defined as
the the start of the scan line specified in the Pipe C Display Scan Line Count Range Compare Register.
See Scan Line Event in the Device Programming Interface chapter of MI Functions.
1 Display Pipe C Vertical Blank Wait Enable
Format: Enable
This field enables a wait until the next Display Pipe C Vertical Blank event occurs. This event is defined
as the start of the next Display Pipe C vertical blank period. Note that this can cause a wait for up to an
entire refresh period. See Vertical Blank Event (See Programming Interface).
0 Display Pipe C H Blank Wait Enable
Format: Enable
This field enables a wait until the start of next Display Pipe C Horizontal Blank event occurs. This event
is defined as the start of the next Display C Horizontal blank period. Note that this can cause a wait for
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 100
SYNC_FLIP_STATUS_1 - Wait For Event and Display Flip Flags
Register 1
up to a line. See Horizontal Blank Event in the Device Programming Interface chapter of MI Functions.
1.1.15 Pipelines Statistics Counter Registers
These registers keep continuous count of statistics regarding the 3D pipeline. They are saved and
restored with context but should not be changed by software except to reset them to 0 at context creation
time. These registers may be read at any time; however, to obtain a meaningful result, a pipeline flush
just prior to reading the registers is necessary in order to synchronize the counts with the primitive stream.
1.1.15.1 IA_VERTICES_COUNT — Reported Vertices Counter
IA_VERTICES_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02310h
This register stores the count of vertices processed by VF. This register is part of the context save and restore.
DWord Bit Description
0 63:0 IA Vertices Count Report
Total number of vertices fetched by the VF stage. This count is updated for every input vertex as long as
Statistics Enable is set in VF_STATE (see the Vertex Fetch Chapter in the 3D Volume.)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 101
1.1.15.2 IA_PRIMITIVES_COUNT — Reported Vertex Fetch Output Primitives
Counter
IA_PRIMITIVES_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02318h
This register stores the count of primitives generated by VF. This register is part of the context save and restore.
DWord Bit Description
0 63:0 IA Primitives Count Report
Total number of primitives output by the Vertex Fetch (IA) stage. This count is updated for every
primitive output by the VF stage, as long as Statistics Enable is set in VF_STATE (see the Vertex Fetch
Chapter in the 3D Volume.)
1.1.15.3 VS_INVOCATION_COUNT— Reported Vertex Shader Invocation
Counter
VS_INVOCATION_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02320h
This register stores the value of the vertex count shaded by VS. This register is part of the context save and restore.
DWord Bit Description
0 63:0 VS Invocation Count Report
Number of vertices that are dispatched as threads by the VS stage. Updated only when Statistics
Enable is set in VS_STATE (see the Vertex Shader Chapter in the 3D Volume.)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 102
1.1.15.4 HS_INVOCATION_COUNT— Reported Hull Shader Invocation Counter
HS_INVOCATION_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02300h
This register stores the number of patch objects processed by the HS unit. E.g., A PATCHLIST_2 topology with 6
vertices would cause this counter to increment by 3 (there are 3 2-vertex patch objects in that topology).This register
is part of the context save and restore.
DWord Bit Description
0 63:0 HS Invocation Count
Number of patch objects processed by the HS stage. Updated only when HS Enable and HS
Statistics Enable are set in 3DSTATE_HS
1.1.15.5 DS_INVOCATION_COUNT— Reported Domain Shader Invocation
Counter
DS_INVOCATION_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02308h
This register stores the number of domain points shaded by the DS threads. Domain points which hit in the DS cache
will not cause this register to increment. Note that the spawning of a DS thread which shades two domain points will
cause this counter to increment by two.This register is part of the context save and restore.
DWord Bit Description
0 63:0 DS Invocation Count
Number of domain points shaded by the DS threads. Updated only when DS Function Enable and
Statistics Enable are set in 3DSTATE_DS
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 103
1.1.15.6 GS_INVOCATION_COUNT — Reported Geometry Shader Thread
Invocation Counter
GS_INVOCATION_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02328h
This register stores the number of objects that are part of geometry shader threads. This register is part of the
context save and restore.
DWord Bit Description
0 63:0 GS Invocation Count
Number of objects that are dispatched as a geometry shader threads invoked by the GS stage. Updated
only when Statistics Enable is set in GS_STATE (see the Geometry Shader Chapter in the 3D
Volume.)
1.1.15.7 GS_PRIMITIVES_COUNT — Reported Geometry Shader Output
Primitives Counter
GS_PRIMITIVES_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02330h
This register reflects the total number of primitives that have been output by the Geometry Shader stage. This
register is part of the context save and restore.
DWord Bit Description
0 63:0 GS Primitives Count
Total number of primitives output by the geometry stage. Updated only when Statistics Enable is set in
GS_STATE (see the Geometry Shader Chapter in the 3D Volume.)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 104
1.1.15.8 CL_INVOCATION_COUNT— Reported Clipper Thread Invocation
Counter
CL_INVOCATION_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02338h
This register stores the count of objects entering the Clipper stage. This register is part of the context save and
restore.
DWord Bit Description
0 63:0 CL Invocation Count Report
Number of objects entering the clipper stage. Updated only when Statistics Enable is set in
CLIP_STATE (see the Clipper Chapter in the 3D Volume.)
1.1.15.9 CL_PRIMITIVES_COUNT— Reported Clipper Output Primitives Counter
CL_PRIMITIVES_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02340h
This register reflects the total number of primitives that have been output by the clipper. This register is part of the
context save and restore.
DWord Bit Description
0 63:0 Clipped Primitives Output Count
Total number of primitives output by the clipper stage. This count is updated for every primitive output by
the clipper stage, as long as Statistics Enable is set in SF_STATE (see the Clipper and SF Chapters in
the 3D Volume.)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 105
1.1.15.10 PS_INVOCATION_COUNT— Reported Pixels Shaded Counter
PS_INVOCATION_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02348h
DWord Bit Description
0 63:0 PS Invocation Count
Reflects a count of the total number of pixels (including unlit “helper pixels” within a subspan that need to
go through the PS shader to provide 2x2 gradients) that are dispatched to pixel shader invocations while
Statistics Enable is set in the Windower. See the Windower chapter of the 3D volume for details. This
count will generally be much greater than the actual count of PS threads since a single thread may
process up to 32 pixels.
1.1.15.11 PS_DEPTH_COUNT — Reported Pixels Passing Depth Test Counter
PS_DEPTH_COUNT
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02350h
This register stores the value of the count of pixels that have passed the depth test. This register is part of the
context save and restore. Note that the value of this register can be obtained in a pipeline-synchronous fashion
without a pipeline flush by using the 3DCONTROL command. See 3D Overview in the 3D volume.
DWord Bit Description
0 63:0 Depth Count
This register reflects the total number of pixels that have passed the depth test (i.e., will be visible). All
pixels are counted when Statistics Enable is set in the Windower State. See the Windower chapter of the
3D volume for details. Pixels that pass the depth test but fail the stencil test will not be counted.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 106
1.1.15.12 TIMESTAMP — Reported Timestamp Count
TIMESTAMP - Reported Timestamp Count
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: RO. This register is not set by the context restore.
Size (in bits): 64
Address: 02358h
This register provides an elapsed real-time value that can be used as a timestamp for GPU events over short periods
of time. Note that the value of this register can be obtained in a 3D pipeline-synchronous fashion without a pipeline
flush by using the PIPE_CONTROL command. See 3D Geometry Pipeline in the “3D and Media” volume.This
register (effectively) counts at a constant frequency by adjusting the increment amount according to the actual
reference clock frequency. SW therefore does not need to know the reference clock frequency.This register is not
reset by a graphics reset. It will maintain its value unless a full chipset reset is performed.
Note: This timestamp register reflects the value of the PCU TSC. The PCU TSC counts 10ns increments; this
timestamp reflects bits 38:3 of the TSC (i.e. 80ns granularity, rolling over every 1.5 hours).
DWord Bit Description
0 63:36 Reserved
Format: MBZ
35:0 Timestamp Value
Format: U32
This register toggles every 80 ns. The upper 28 bits are zero.
1.1.15.13 SO_NUM_PRIMS_WRITTEN[0:3]— Stream Output Num Primitives
Written Counters
SO_NUM_PRIMS_WRITTEN[0:3] - Stream Output Num Primitives
Written Counter
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Address: 05200h-0521Fh
There is one 64-bit register for each of the 4 supported streams:5200h-5207h SO_NUM_PRIMS_WRITTEN0 (for
Stream Out Stream #0)5208h-520Fh SO_NUM_PRIMS_WRITTEN1 (for Stream Out Stream #1)5210h-5217h
SO_NUM_PRIMS_WRITTEN2 (for Stream Out Stream #2)5218h-521Fh SO_NUM_PRIMS_WRITTEN3 (for Stream
Out Stream #3)These registers are used to count the number of primitives (aka objects: points, lines, triangles) which
the SO stage has successfully written to a particular “stream’s” Streamed Vertex Output buffers, subject to buffer
overflow detection. (See the Stream Output section of the 3D pipeline volume).These registers are part of the context
save and restore.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 107
DWord Bit Description
0 63:0 Num Prims Written Count
Format: U64
This count is incremented (by one) every time a GS thread outputs a DataPort Streamed Vertex Buffer
Write message with the Increment Num Prims Written bit set in the message header (see the Geometry
Shader and Data Port chapters in the 3D Volume.)
1.1.15.14 SO_PRIM_STORAGE_NEEDED[0:3] —Stream Output Primitive Storage
Needed Counters
SO_PRIM_STORAGE_NEEDED[0:3] - Stream Output Primitive
Storage Needed Counters
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: RO. This register is set by the context restore.
Size (in bits): 64
Address: 05240h-0525Fh
There is one 64-bit register for each of the 4 supported streams:
5240h-5247h SO_PRIM_STORAGE_NEEDED0 (for Stream Out Stream #0)
5248h-524Fh SO_PRIM_STORAGE_NEEDED1 (for Stream Out Stream #1)
5250h-5257h SO_PRIM_STORAGE_NEEDED2 (for Stream Out Stream #2)
5258h-525Fh SO_PRIM_STORAGE_NEEDED3 (for Stream Out Stream #3)
These registers are used to count the number of primitives (aka objects: points, lines, triangles) which the SO stage
has or would have written to a particular “stream’s” Streamed Vertex Output buffers if all buffers had been large
enough to accommodate the writes. (See the Stream Output section of the 3D pipeline volume).
These registers are part of the context save and restore.
DWord Bit Description
0 63:0 Prim Storage Needed Count
Format: U64
This count is incremented (by one) by the SOL stage for each object (point, line, triangle) it writes or
attempts to write to the corresponding stream’s output buffers. The count is not affected by the actual
number of buffers bound to the stream.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 108
1.1.15.15 SO_WRITE_OFFSET[0:3] —Stream Output Write Offsets
SO_WRITE_OFFSET[0:3] - Stream Output Write Offsets
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: RW. This register is set by the context restore.
Size (in bits): 32
Address: 05280h-0528Fh
There is one R/W 32-bit register for each of the 4 supported stream output buffer slots:
5280h-5283h SO_WRITE_OFFSET0 (for Stream Out Buffer #0)
5284h-5287h SO_ WRITE_OFFSET1 (for Stream Out Buffer #1)
5288h-528Bh SO_ WRITE_OFFSET2 (for Stream Out Buffer #2)
528Ch-528Fh SO_ WRITE_OFFSET3 (for Stream Out Buffer #3)
These registers are used to set and track a DWord-granular Write Offset for each of the 4 Stream Output Buffer
slots. Software can directly write them via MI_LOAD_REGxxx commands. The SOL stage will increment them as
part of stream output processing. Software can cause them to be written to memory via MI_STORE_REGxxx
commands. (See the Stream Output section of the 3D pipeline volume).
These registers are part of the context save and restore.
Programming Notes
 Software must ensure that no HW stream output operations can be in process or otherwise pending at the
point that the MI_LOAD/STORE commands are processed. This will likely require a pipeline flush.
 The SOL stage will effectively advance the write offset by the buffer’s Surface Pitch after each vertex is written
(assuming no overflow is detected in any targetted SO buffer). Under “normal” conditions one would expect
software to initialize the WriteOffset to some (possibly zero) multiple of Surface Pitch in order to align vertex
writes to the buffer’s Base Address, though it is not required to do so.
DWord Bit Description
0 31:2 Write Offset
Format: U30
This field contains a DWord offset from the corresponding SO buffer’s Base Address value. The SOL
stage uses this value as a write offset when performing writes to the buffer. The SOL stage will
increment this value as a part of performing stream output to the buffer. Note that the SOL stage uses
the buffer’s Surface Pitch to advance the Write Offset, without regard to the buffer’s Base Address (see
Programming Notes above).
1:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 109
1.1.16 Predicate Render Registers
1.1.16.1 MI_PREDICATE_SRC0 - Predicate Rendering Temporary Register0
MI_PREDICATE_SRC0 - Predicate Rendering Temporary Register0
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Address: 02400h-02407h
DWord Bit Description
0 63:0 MI_PREDICATE_SRC0
This register is a temporary register for Predicate Rendering. See Predicate Rendering section for more
details.
1.1.16.2 1.1.16.2 MI_PREDICATE_SRC1– Predicate Rendering Temporary
Register1
MI_PREDICATE_SRC1 - Predicate Rendering Temporary Register1
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Address: 02408h-0240Fh
DWord Bit Description
0 63:0 MI_PREDICATE_SRC1
This register is a temporary register for Predicate Rendering. See Predicate Rendering section for more
details.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 110
1.1.16.3 MI_PREDICATE_DATA– Predicate Rendering Data Storage
MI_PREDICATE_DATA - Predicate Rendering Data Storage
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Address: 02410h-02417h
DWord Bit Description
0 63:0 MI_PREDICATE_DATA
This register is used either as computed value based off the MI_PREDICATE_SRC0 and
MI_PREDICATE_SRC1 or a temporary register. See Predicate Rendering section for more details.
1.1.16.4 MI_PREDICATE_RESULT – Predicate Rendering Data Result
MI_PREDICATE_RESULT - Predicate Rendering Data Result
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02418h
DWord Bit Description
0 31:1 Reserved
Format: MBZ
0 MI_PREDICATE_RESULT
This bit is the result of the last MI_PREDICATE.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 111
1.1.17 AUTO_DRAW Registers
1.1.17.1 3DPRIM_END_OFFSET – Auto Draw End Offset
3DPRIM_END_OFFSET - Auto Draw End Offset
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02420h-02423h
DWord Bit Description
0 31:0 End Offset
Format: U32
This register is used to store the end offset value used by the Vertex Fetch to determine when to stop
processing the 3D_PRIMITIVE command. This register is valid when the End Offset Enable is set in the
3D_PRIMITIVE command.
1.1.17.2 3DPRIM_START_VERTEX – Load Indirect Start Vertex
3DPRIM_START_VERTEX - Load Indirect Start Vertex
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02430h-02433h
DWord Bit Description
0 31:0 Start Vertex
Format: U32
This register is used to store the Start Vertex of the 3D_PRIMITIVE command when Load Indirect
Enable is set.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 112
1.1.17.3 3DPRIM_VERTEX_COUNT – Load Indirect Vertex Count
3DPRIM_VERTEX_COUNT - Load Indirect Vertex Count
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02434h-02437h
DWord Bit Description
0 31:0 Vertex Count
Format: U32
This register is used to store the Vertex Count of the 3D_PRIMITIVE command when Load Indirect
Enable is set.
1.1.17.4 3DPRIM_INSTANCE_COUNT – Load Indirect Instance Count
3DPRIM_INSTANCE_COUNT - Load Indirect Instance Count
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02438h-0243Bh
DWord Bit Description
0 31:0 Instance Count
This register is used to store the Instance Count of the 3D_PRIMITIVE command when Load Indirect
Enable is set.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 113
1.1.17.5 3DPRIM_START_INSTANCE – Load Indirect Start Instance
3DPRIM_START_INSTANCE - Load Indirect Start Instance
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 0243Ch-0243Fh
DWord Bit Description
0 31:0 Start Vertex
Format: U32
This register is used to store the Start Instance of the 3D_PRIMITIVE command when Load Indirect
Enable is set.
1.1.17.6 3DPRIM_BASE_VERTEX – Load Indirect Base Vertex
3DPRIM_BASE_VERTEX - Load Indirect Base Vertex
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02440h-02443h
DWord Bit Description
0 31:0 Base Vertex
Format: S31
This register is used to store the Base Vertex of the 3D_PRIMITIVE command when Load Indirect
Enable is set.
1.1.18 MMIO Registers for GPGPU Indirect Dispatch
This register is normally written with the MI_LOAD_REGISTER_MEMORY command rather than from the
CPU.
These registers should not be written with 0 for these projects. To avoid this, the
MI_LOAD_REGISTER_MEMORY command which writes them from an address in memory which was
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 114
written by a previous GPGPU_WALKER command will need to be checked with the following command
sequence. The commands in red are the additional commands to implement the workaround:
MI_LOAD_REGISTER_MEMORY Xaddress, GPGPU_DISPATCHDIMX
MI_CONDITIONAL_BATCH_BUFFER_END Xaddress, 0 // Compare X dimension to 0, end batch buffer if
0
MI_LOAD_REGISTER_MEMORY GPGPU_DISPATCHDIMY
MI_CONDITIONAL_BATCH_BUFFER_END Yaddress, 0 // Compare Y dimension to 0, end batch buffer if
0
MI_LOAD_REGISTER_MEMORY GPGPU_DISPATCHDIMZ
MI_CONDITIONAL_BATCH_BUFFER_END Zaddress, 0 // Compare Z dimension to 0, end batch buffer if
0
GPGPU_WALKER // Walker with indirect dispatch
This way, if any dimension is 0 we would not execute the GPGPU_WALKER. This has the limitation that
the indirect GPGPU_WALKER has to be the last WALKER of the batch buffer.
1.1.18.1 GPGPU_DISPATCHDIM(X/Y/Z) - GPGPU Dispatch Dimension (X/Y/Z)
These registers are normally written with the MI_LOAD_REGISTER_MEMORY command rather than
from the CPU.
GPGPU_DISPATCHDIMX - GPGPU Dispatch Dimension X
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02500h
DWord Bit Description
0 31:0 Dispatch Dimension X
Format: U32
The number of thread groups to be dispatched in the X dimension (max x + 1).
Value Name Project
1,FFFFFFFFh
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 115
GPGPU_DISPATCHDIMY - GPGPU Dispatch Dimension Y
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02504h
DWord Bit Description
0 31:0 Dispatch Dimension Y
Format: U32
The number of thread groups to be dispatched in the Y dimension (max y + 1
Value Name Project
1,FFFFFFFFh
GPGPU_DISPATCHDIMZ - GPGPU Dispatch Dimension Z
Register Space: MMIO: 0/2/0
Source: RenderCS
Default Value: 0x00000000
Access: R/W
Size (in bits): 32
Address: 02508h
DWord Bit Description
0 31:0 Dispatch Dimension Z
Format: U32
The number of thread groups to be dispatched in the Zdimension (max Z + 1)
Value Name Project
1,FFFFFFFFh
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 116
1.1.18.2 TS_GPGPU_THREADS_DISPATCHED – Count Active Channels
Dispatched
TS_GPGPU_THREADS_DISPATCHED - Count Active Channels
Dispatched
Register Space: MMIO: 0/2/0
Project: All
Source: RenderCS
Default Value: 0x00000000, 0x00000000
Access: R/W
Size (in bits): 64
Trusted Type: 1
Address: 02290h
This register is used to count the number of active channels that TS sends for dispatch. For each dispatch the active
bits in the execution mask are summed and added to this register. This register is reset when a write occurs to 2290h
DWord Bit Description
0 63:0 GPGPU_THREADS_DISPATCHED
Format: U64
This count is increased by the number of active bits in the execution mask each time the TS sends a
GPGPU dispatch.
1.1.19 Memory Interface Registers
1.1.19.1 PWRCTX_REST_DONE - Power Context Restore Done
PWRCTX_REST_DONE - Power Context Restore Done
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04000h-04003h
Power Context Restore Done
DWord Bit Description
0 31:16 Mask Bits
Default Value: 0000h
Access: RO
Mask Bits
15:1 Extra Mask Bits
Default Value: 000000000000000b
Access: R/W
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 117
PWRCTX_REST_DONE - Power Context Restore Done
Extra Mask Bits
0 Restore Done
Default Value: 0b
Access: R/W
GAM - CS will write to indicate ‘restore done’
It is a config message register between CS & GAM
1.1.19.2 WR_WATERMARK - Write Watermark
WR_WATERMARK - Write Watermark
Register Space: MMIO: 0/2/0
Default Value: 0x000FFEA4
Address: 04028h-0402Bh
Write Watermark
DWord Bit Description
0 31:20 Counter Extra Bits
Default Value: 000000000000b
Access: R/W
Counter Extra Bits
19 Watermark Timeout Enable
Default Value: 1b
Access: R/W
Watermark timeout enable.
18:8 Watermark Timeout
Default Value: 11111111110b
Access: R/W
Number of clocks that the write pipe queue is allowed to keep a ready write cycle, without reads or
writes to the queue. Once this value is met, and if the feature is enabled, the watermark will be
considered reach, and all pending write requests will be issued.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 118
WR_WATERMARK - Write Watermark
7 Watermark En
Default Value: 1b
Access: R/W
Enable write request grouping
6:0 High Watermark
Default Value: 0100100b
Access: R/W
This is the number of write requests to be collected before initiating a write burst. Once a burst is
initiated, it will continue until all the available writes are requested.
1.1.19.3 GFX_PRIO_CTRL - GFX Arbiter Client Priority Control
GFX_PRIO_CTRL - GFX Arbiter Client Priority Control
Register Space: MMIO: 0/2/0
Default Value: 0x00011D10
Address: 0402Ch-0402Fh
GFX Arbiter Client Priority Control
DWord Bit Description
0 31:17 Extra 402C Register
Default Value: 000000000000000b
Access: R/W
Extra 402C Register
16:12 Read Rstrm Max Reject
Default Value: 10001b
Access: R/W
Read Rstrm Max Reject
11:9 gapc_gam_c_priority
Default Value: 110b
Access: R/W
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 119
GFX_PRIO_CTRL - GFX Arbiter Client Priority Control
gapc_gam_c_priority - Lowest Bit [9] is not used
8:6 gapc_gam_z_priority
Default Value: 100b
Access: R/W
gapc_gam_z_priority - Lowest Bit [6] is not used
5:3 gapc_gam_l3_priority
Default Value: 010b
Access: R/W
gapc_gam_l3_priority - Lowest Bit [3] is not used
2:0 gafm_gam_priority
Default Value: 000b
Access: R/W
Client Priority control bitss
gafm_gam_priority - Lowest Bit [0] is not used
1.1.19.4 1.1.19.4 GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests
0
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04034h-04037h
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
DWord Bit Description
0 31 TEX Limit Enable Bit
Default Value: 0b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 120
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
Access: R/W
TEX Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the Texture Cache
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
30 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
29:24 TEX TLB Limit Count
Default Value: 000000b
Access: R/W
TEX TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read
23 DC Limit Enable bit
Default Value: 0b
Access: R/W
DC Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the Instruction Cache.
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
22 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
21:16 DC TLB Limit Count
Default Value: 000000b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 121
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
Access: R/W
DC TLB Limit Count Project:All Format:U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
15 VF Limit Enable bit
Default Value: 0b
Access: R/W
VF Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the Vertex Fetch
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
14 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
13:8 VF TLB Limit Count
Default Value: 000000b
Access: R/W
VF TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
7 VMC Limit Enable bit
Default Value: 0b
Access: R/W
VMC Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the Video Motion
Compensation . When set, the number of internal pending read requests which require a TLB read will
not exceed the programmed counter value.
6 Reserved Bit
Default Value: 0b
Access: RO
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 122
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
Reserved Project: All Format: MBZ
5:0 VMC TLB Limit Count
Default Value: 000000b
Access: R/W
VMC TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
B/D/F/Type:0/0/0/GAMunit_Config
Address Offset:4034-4037h
Default Value:00000000h
Access: RO; RW;
Size:32 bits
GFX_PEND_TLB_0 - Max Outstanding Pending TLB Requests 0
Default
Bit Access Value RST/PWR Description
31 RW 0b Core
TEX Limit Enable Bit (TEXLEN):
TEX Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the
Texture Cache
When set, the number of internal pending read requests which require a TLB
read will not exceed the programmed counter value.
30 RO 0b Core
Reserved Bit (RSVD):
Reserved Project: All Format: MBZ
29:24 RW 000000b Core
TEX TLB Limit Count (TEXTLBLCNT):
TEX TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require
a TLB read
23 RW 0b Core
DC Limit Enable bit (DCLEN):
DC Limit Enable bit Project: All Format: U1
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 123
Default
Bit Access Value RST/PWR Description
This bit is used to enable the pending TLB requests limitation function for the
Instruction Cache.
When set, the number of internal pending read requests which require a TLB
read will not exceed the programmed counter value.
22 RO 0b Core
Reserved Bit (RSVD):
Reserved Project: All Format: MBZ
21:16 RW 000000b Core
DC TLB Limit Count (DCTLBLCNT):
DC TLB Limit Count Project:All Format:U6
This is the MAX number of Allowed internal pending read requests which require
a TLB read.
15 RW 0b Core
VF Limit Enable bit (VFLEN):
VF Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the
Vertex Fetch
When set, the number of internal pending read requests which require a TLB
read will not exceed the programmed counter value.
14 RO 0b Core
Reserved Bit (RSVD):
Reserved Project: All Format: MBZ
13:8 RW 000000b Core
VF TLB Limit Count (VFTLBLCNT):
VF TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require
a TLB read.
7 RW 0b Core
VMC Limit Enable bit (VMCLEN):
VMC Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the
Video Motion Compensation. When set, the number of internal pending read
requests which require a TLB read will not exceed the programmed counter
value.
6 RO 0b Core
Reserved Bit (RSVD):
Reserved Project: All Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 124
Default
Bit Access Value RST/PWR Description
5:0 RW 000000b Core
VMC TLB Limit Count (VMCTLBLCNT):
VMC TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require
a TLB read.
1.1.19.5 GFX_PEND_TLB_1 - Max Outstanding Pending TLB Requests 1
GFX_PEND_TLB_1 - Max Outstanding Pending TLB Requests 1
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04038h-0403Bh
GFX_PEND_TLB_1 - Max Outstanding pending TLB requests 1
DWord Bit Description
0 31 SOL Limit Enable bit
Default Value: 0b
Access: R/W
SOL Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the SOL. When set, the
number of internal pending read requests which require a TLB read will not exceed the programmed
counter value.
30 Reserved Bits
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
29:24 SOL TLB Limit Count
Default Value: 000000b
Access: R/W
SOL TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
23 L3 Limit Enable bit
Default Value: 0b
Access: R/W
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 125
GFX_PEND_TLB_1 - Max Outstanding Pending TLB Requests 1
L3 Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the L3. When set, the
number of internal pending read requests which require a TLB read will not exceed the programmed
counter value.
22 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
21:16 L3 TLB Limit Count
Default Value: 000000b
Access: R/W
L3 TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
15 RCZ Limit Enable bit
Default Value: 0b
Access: R/W
RCZ Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the Render Depth Cache
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
14 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
Programming Notes
""
13:8 RCZ TLB Limit Count
Default Value: 000000b
Access: R/W
RCZ TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 126
GFX_PEND_TLB_1 - Max Outstanding Pending TLB Requests 1
7 RCC Limit Enable bit
Default Value: 0b
Access: R/W
RCC Limit Enable bit Project: All Format: U1
This bit is used to enable the pending TLB requests limitation function for the Render Color Cache.
When set, the number of internal pending read requests which require a TLB read will not exceed the
programmed counter value.
6 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
5:0 RCC TLB Limit Count
Default Value: 000000b
Access: R/W
RCC TLB Limit Count Project: All Format: U6
This is the MAX number of Allowed internal pending read requests which require a TLB read.
1.1.19.6 L3_LRA_0 - L3 LRA 0
L3_LRA_0 - L3 LRA 0
Register Space: MMIO: 0/2/0
Default Value: 0x3F201F00
Address: 0403Ch-0403Fh
L3 LRA 0
DWord Bit Description
0 31:24 L3 LRA1 Max
Default Value: 00111111b
Access: R/W
L3 LRA1 Max Project: All Format: U6
Maximum value of programmable LRA1
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 127
L3_LRA_0 - L3 LRA 0
23:16 L3 LRA1 Min
Default Value: 00100000b
Access: R/W
L3 LRA1 Min Project: All Format: U6
Minimum value of programmable LRA1
15:8 L3 LRA0 Max
Default Value: 00011111b
Access: R/W
L3 LRA0 Max Project: All Format: U6
Maximum value of programmable LRA0
7:0 L3 LRA0 Min
Default Value: 00000000b
Access: R/W
L3 LRA0 Min Project: All Format: U6
Minimum value of programmable LRA1
1.1.19.7 L3_LRA_1 - L3 LRA 1
L3_LRA_1 - L3 LRA 1
Register Space: MMIO: 0/2/0
Default Value: 0x0900FF40
Address: 04040h-04043h
L3 LRA 1
DWord Bit Description
0 31:30 Reserved Bits
Default Value: 00b
Access: RO
Reserved Bits
29:28 DC
Default Value: 00b
Access: R/W
Which LRA should DC use
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 128
L3_LRA_1 - L3 LRA 1
27:26 TEXTURE
Default Value: 10b
Access: R/W
Which LRA should TEXTURE use
25:24 L3
Default Value: 01b
Access: R/W
Which LRA should L3 use
23:16 Reserved Bits
Default Value: 00000000b
Access: RO
Reserved Bits
15:8 L3 LRA2 Max
Default Value: 11111111b
Access: R/W
L3 LRA2 Max Project: All Format: U6
Maximum value of programmable LRA2
7:0 L3 LRA2 Min
Default Value: 01000000b
Access: R/W
L3 LRA2 Min Project: All Format: U6
Minimum value of programmable LRA2
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 129
1.1.19.8 CVS_TLB_LRA_0 - CVS TLB LRA 0
CVS_TLB_LRA_0 - CVS TLB LRA 0
Register Space: MMIO: 0/2/0
Default Value: 0x1F080700
Address: 04044h-04047h
CVS TLB LRA 0
DWord Bit Description
0 31:29 Reserved Bits
Default Value: 000b
Access: RO
Reserved Project: All Format: MBZ
28:24 CVS LRA1 Max
Default Value: 11111b
Access: R/W
CVS LRA1 Max Project: All Format: MBZ
Maximum value of programmable LRA1
23:21 Reserved Bits
Default Value: 000b
Access: RO
Reserved Project: All Format: MBZ
20:16 CVS LRA1 Min
Default Value: 01000b
Access: R/W
CVS LRA1 Min Project: All Format: U6
Minimum value of programmable LRA1
15:13 Reserved Bits
Default Value: 000b
Access: RO
Reserved Project: All Format: MBZ
12:8 CVS LRA0 Max
Default Value: 00111b
Access: R/W
CVS LRA0 Max Project: All Format: MBZ
Maximum value of programmable LRA0
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 130
CVS_TLB_LRA_0 - CVS TLB LRA 0
7:5 Reserved Bits
Default Value: 000b
Access: RO
Reserved Project: All Format: MBZ
4:0 CVS LRA0 Min
Default Value: 00000b
Access: R/W
CVS LRA0 Min Project: All Format: U6
Minimum value of programmable LRA0
1.1.19.9 _TLB_LRA_1 - CVS TLB LRA 1
CVS_TLB_LRA_1 - CVS TLB LRA 1
Register Space: MMIO: 0/2/0
Default Value: 0x00001F18
Address: 04048h-0404Bh
CVS TLB LRA 1
DWord Bit Description
0 31:13 Reserved Bits
Default Value: 0000000000000000000b
Access: RO
Reserved Project: All Format: MBZ
12:8 CVS LRA2 Max
Default Value: 11111b
Access: R/W
CVS LRA2 Max Project: All Format: MBZ
Maximum value of programmable LRA2
7:5 Reserved Bits
Default Value: 000b
Access: RO
Reserved Project: All Format: MBZ
4:0 CVS LRA2 Min
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 131
CVS_TLB_LRA_1 - CVS TLB LRA 1
Default Value: 11000b
Access: R/W
CVS LRA2 Min Project: All Format: U6
Minimum value of programmable LRA2
1.1.19.10 CVS_TLB_LRA_2 - CVS TLB LRA 2
CVS_TLB_LRA_2 - CVS TLB LRA 2
Register Space: MMIO: 0/2/0
Default Value: 0x00000005
Address: 0404Ch-0404Fh
CVS TLB LRA 2
DWord Bit Description
0 31:6 Reserved Bits
Default Value: 00000000000000000000000000b
Access: RO
Reserved Project: All Format: MBZ
5:4 CS LRA
Default Value: 00b
Access: R/W
CS LRA Project: All Format: U6
Which LRA should CS use
3:2 VF LRA
Default Value: 01b
Access: R/W
VF LRA Project: All Format: U1
Which LRA should VF use
1:0 SO LRA
Default Value: 01b
Access: R/W
SO LRA Project: All Format: MBZ
Which LRA should SO use
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 132
1.1.19.11 ZTLB_LRA_0 - ZTLB LRA 0
ZTLB_LRA_0 - ZTLB LRA 0
Register Space: MMIO: 0/2/0
Default Value: 0x1F107F00
Address: 04050h-04053h
ZTLB TLB LRA 0
DWord Bit Description
0 31 Reserved Bits
Default Value: 0b
Access: RO
Reserved Bits
30:24 ZTLB LRA1 Max
Default Value: 0011111b
Access: R/W
ZTLB LRA1 Max Project: All Format: U6
Maximum value of programmable LRA1
23 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: U1
22:16 ZTLB LRA1 Min
Default Value: 0010000b
Access: R/W
ZTLB LRA1 Min Project: All Format: MBZ
Minimum value of programmable LRA1
15 Reserved Bits
Default Value: 0b
Access: RO
Reserved Bits
14:8 ZTLB LRA0 Max
Default Value: 1111111b
Access: R/W
ZTLB LRA0 Max Project: All Format: U1
Maximum value of programmable LRA0
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 133
ZTLB_LRA_0 - ZTLB LRA 0
7 Reserved Bit
Default Value: 0b
Access: RO
Reserved Project: All Format: U1
6:0 ZTLB LRA0 Min
Default Value: 0000000b
Access: R/W
ZTLB LRA0 Min Project: All Format: U6
Minimum value of programmable LRA0
1.1.19.12 ZTLB_LRA_1 - ZTLB LRA 1
ZTLB_LRA_1 - ZTLB LRA 1
Register Space: MMIO: 0/2/0
Default Value: 0x00002F20
Address: 04054h-04057h
ZTLB TLB LRA 1
DWord Bit Description
0 31:22 Reserved Bits
Default Value: 0000000000b
Access: RO
Reserved Project: All Format: MBZ
21:20 STC LRA
Default Value: 00b
Access: R/W
STC LRA Project: All Format: U6
Which LRA should STC use
19:18 HIZ LRA
Default Value: 00b
Access: R/W
HIZ LRA Project: All Format: U1
Which LRA should HIZ use
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 134
ZTLB_LRA_1 - ZTLB LRA 1
17:16 RCZ LRA
Default Value: 00b
Access: R/W
RCZ LRA Project: All Format: MBZ
Which LRA should RCZ use
15 Reserved Bits
Default Value: 0b
Access: RO
Reserved Bits
14:8 ZTLB LRA2 Max
Default Value: 0101111b
Access: R/W
ZTLB LRA2 Max Project: All Format: U1
Maximum value of programmable LRA2
7 Reserved Bits
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
6:0 ZTLB LRA2 Min
Default Value: 0100000b
Access: R/W
ZTLB LRA2 Min Project: All Format: U6
Minimum value of programmable LRA2
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 135
1.1.19.13 RCC_LRA_0 - RCC LRA 0
RCC_LRA_0 - RCC LRA 0
Register Space: MMIO: 0/2/0
Default Value: 0x3F100F00
Address: 04058h-0405Bh
RCC LRA 0
DWord Bit Description
0 31:30 Reserved Bit
Default Value: 00b
Access: RO
Reserved Project: All Format: U1
29:24 RCC LRA1 Max
Default Value: 111111b
Access: R/W
RCC LRA1 Max Project: All Format: U6
Maximum value of programmable LRA1
23:22 Reserved Bit
Default Value: 00b
Access: RO
Reserved Project: All Format: U1
21:16 RCC LRA1 Min
Default Value: 010000b
Access: R/W
RCC LRA1 Min Project: All Format: MBZ
Minimum value of programmable LRA1
15:14 Reserved Bit
Default Value: 00b
Access: RO
Reserved Project: All Format: U1
13:8 RCC LRA0 Max
Default Value: 001111b
Access: R/W
RCC LRA0 Max Project: All Format: U1
Maximum value of programmable LRA0
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 136
RCC_LRA_0 - RCC LRA 0
7:6 Reserved Bit
Default Value: 00b
Access: RO
Reserved Project: All Format: U1
5:0 RCC LRA0 Min
Default Value: 000000b
Access: R/W
RCC LRA0 Min Project: All Format: U6
Minimum value of programmable LRA0
1.1.19.14 RCC_LRA_1 - RCC LRA 1
RCC_LRA_1 - RCC LRA 1
Register Space: MMIO: 0/2/0
Default Value: 0x00010000
Address: 0405Ch-0405Fh
RCC LRA 1
DWord Bit Description
0 31:20 Reserved Bits
Default Value: 000000000000b
Access: RO
Reserved Project: All Format: MBZ
19:18 MSC LRA
Default Value: 00b
Access: R/W
MSC LRA Project: All Format: U1
Which LRA should MSC use
17:16 RCC LRA
Default Value: 01b
Access: R/W
RCC LRA Project: All Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 137
RCC_LRA_1 - RCC LRA 1
Which LRA should RCC use
15:0 Reserved Bits
Default Value: 0000000000000000b
Access: RO
Reserved Project: All Format: MBZ
1.1.19.15 CASC_LRA_0 - CASC LRA 0
CASC_LRA_0 - CASC LRA 0
Register Space: MMIO: 0/2/0
Default Value: 0x1F100F00
Address: 04060h-04063h
CASC LRA 0
DWord Bit Description
0 31:24 CASC LRA1 Max
Default Value: 00011111b
Access: R/W
CASC LRA1 Max Project: All Format: U6
Maximum value of programmable LRA1
Maximum Allow Value: 159
23:16 CASC LRA1 Min
Default Value: 00010000b
Access: R/W
CASC LRA1 Min Project: All Format: U6
Minimum value of programmable LRA1
15:8 CASC LRA0 Max
Default Value: 00001111b
Access: R/W
CASC LRA0 Max Project: All Format: U6
Maximum value of programmable LRA0
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 138
CASC_LRA_0 - CASC LRA 0
Maximum Allow Value: 159
7:0 CASC LRA0 Min
Default Value: 00000000b
Access: R/W
CASC LRA0 Min Project: All Format: U6
Minimum value of programmable LRA1
1.1.19.16 CASC_LRA_1 - CASC LRA 1
CASC_LRA_1 - CASC LRA 1
Register Space: MMIO: 0/2/0
Default Value: 0x3F302F20
Address: 04064h-04067h
CASC LRA 1
DWord Bit Description
0 31:24 CASC LRA3 Max
Default Value: 00111111b
Access: R/W
CASC LRA3 Max Project: All Format: U6
Maximum value of programmable LRA3
23:16 CASC LRA3 Min
Default Value: 00110000b
Access: R/W
CASC LRA3 Min Project: All Format: U6
Minimum value of programmable LRA3
15:8 CASC LRA2 Max
Default Value: 00101111b
Access: R/W
CASC LRA2 Max Project: All Format: U6
Maximum value of programmable LRA2
7:0 CASC LRA2 Min
Default Value: 00100000b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 139
CASC_LRA_1 - CASC LRA 1
Access: R/W
CASC LRA2 Min Project: All Format: U6
Minimum value of programmable LRA2
1.1.19.17 CASC_LRA_2 - CASC LRA 2
CASC_LRA_2 - CASC LRA 2
Register Space: MMIO: 0/2/0
Default Value: 0x00009F40
Address: 04068h-0406Bh
CASC LRA 2
DWord Bit Description
0 31:16 Reserved Bits
Default Value: 0000h
Access: RO
Reserved Project: All Format: MBZ
15:8 CASC LRA4 Max
Default Value: 10011111b
Access: R/W
CASC LRA4 Max Project: All Format: U6
Maximum value of programmable LRA4
Maximum Allow Value: 159
7:0 CASC LRA4 Min
Default Value: 01000000b
Access: R/W
CASC LRA4 Min Project: All Format: U6
Minimum value of programmable LRA4
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 140
1.1.19.18 CASC_LRA_3 - CASC LRA 3
CASC_LRA_3 - CASC LRA 3
Register Space: MMIO: 0/2/0
Default Value: 0x000014E4
Address: 0406Ch-0406Fh
CASC LRA 3
DWord Bit Description
0 31:18 Reserved Bits
Default Value: 00000000000000b
Access: RO
Reserved Project: All Format: MBZ
17:15 BCS LRA
Default Value: 000b
Access: R/W
BCS LRA Project: All Format: U6
Which LRA should use
14:12 BLB LRA
Default Value: 001b
Access: R/W
BLB LRA Project: All Format: U6
Which LRA should use
11:9 VCS LRA
Default Value: 010b
Access: R/W
VCS LRA Project: All Format: U6
Which LRA should use
8:6 VMX LRA
Default Value: 011b
Access: R/W
VMX LRA Project: All Format: U6
Which LRA should use
5:3 VMC LRA
Default Value: 100b
Access: R/W
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 141
CASC_LRA_3 - CASC LRA 3
VMC LRA Project: All Format: U6
Which LRA should use
2:0 VCR LRA
Default Value: 100b
Access: R/W
VCR LRA Project: All Format: U6
Which LRA should use
1.1.19.19 MEDIA_MAX_REQ_COUNT - MAX Requests Allowed - CASC
MEDIA_MAX_REQ_COUNT - MAX Requests Allowed - CASC
Register Space: MMIO: 0/2/0
Default Value: 0x10202020
Address: 04070h-04073h
Programmable Request Count - CASC
DWord Bit Description
0 31:24 GFX Max Request Limit Count
Default Value: 00010000b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each engine . Requests are counted, regardless of kind of cycle (Miss/Hit/Present)
Minimum count value must be = 1
23:16 MFX/BLT Max Request Limit Count
Default Value: 00100000b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each engine . Requests are counted, regardless of kind of cycle (Miss/Hit/Present)
Minimum count value must be = 1
15:14 Reserved Bits
Default Value: 00b
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 142
MEDIA_MAX_REQ_COUNT - MAX Requests Allowed - CASC
Access: RO
Reserved Bits
13:8 VLF Max Request Limit Count
Default Value: 100000b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each client. Requests are counted, regardless of kind of cycle (Miss/Hit/Present )
Minimum count value must be = 1
7:6 Reserved Bits
Default Value: 00b
Access: RO
Reserved Project: All Format: MBZ
5:0 CASC Max Request Limit Count
Default Value: 100000b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each client. Requests are counted, regardless of kind of cycle (Miss/Hit/Present )
Minimum count value must be = 1
1.1.19.20 GFX_MAX_REQ_COUNT - MAX Requests Allowed - GAM
GFX_MAX_REQ_COUNT - MAX Requests Allowed - GAM
Register Space: MMIO: 0/2/0
Default Value: 0x43F20101
Address: 04074h-04077h
Programmable Request Count - GAM
DWord Bit Description
0 31:26 GAP Writes Max Request Limit Count
Default Value: 010000b
Access: R/W
This is the MAX number of Allowed Write Requests Count - These counters keep track of the
accepted write requests from all GAP clients (RCZ, HiZ,Stc, RCC, L3).
Minimum count value must be = 1
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 143
GFX_MAX_REQ_COUNT - MAX Requests Allowed - GAM
25:20 CVS Max Request Limit Count
Default Value: 111111b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each client. Requests are counted, regardless of kind of cycle (Miss/Hit/Present )
Minimum count value must be = 1
19 Reserved Bits
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
18:13 L3 Max Request Limit Count
Default Value: 010000b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each client. Requests are counted, regardless of kind of cycle (Miss/Hit/Present)
Minimum count value must be = 1
12 Reserved Bits
Default Value: 0b
Access: RO
Reserved Project: All Format: MBZ
11:6 Z Request Limit Count
Default Value: 000100b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each client. Requests are counted, regardless of kind of cycle (Miss/Hit/Present)
Minimum count value must be = 1
5:0 RCC Request Limit Count
Default Value: 000001b
Access: R/W
This is the MAX number of Allowed Requests Count - These counters keep track of the accepted
requests from each client. Requests are counted, regardless of kind of cycle (Miss/Hit/Present)
Minimum count value must be = 1
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 144
1.1.19.21 GAM_HWSP_REG - GAM Hardware Status Page Address Register
GAM_HWSP_REG - GAM Hardware Status Page Address Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04080h-04083h
This register is used to program the 4 KB-aligned System Memory address of the Hardware Status Page used to
report hardware status into (typically cacheable) System Memory. This address in this register is translated using
the Global GTT in memory. The mapping type of the GTT entry determines the snoop nature of the transaction to
memory.
DWord Bit Description
0 31:12 GAM HWSP Register
Default Value: 00000h
Access: R/W
11:0 Reserved Bits
Default Value: 000h
Access: RO
1.1.19.22 GFX_ENG_FR - Graphics Engine Fault Register
GFX_ENG_FR - Graphics Engine Fault Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04094h-04097h
Graphics Engine Fault Register
DWord Bit Description
0 31:12 Virtual Address of Fault
Default Value: 00000h
Access: R/W
This is the original Address of the Page that generated the First fault for this engine.
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 145
GFX_ENG_FR - Graphics Engine Fault Register
11 GTTSEL
Default Value: 0b
Access: R/W
This bit indicates if the valid bit happened while using PPGTT or GGTT: 0 - PPGTT, 1 - GGTT
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
10:3 SRCID of Fault
Default Value: 00h
Access: R/W
This is the Source ID of the unit that requested the cycle that generated the First Page fault for this
engine.
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
2:1 Fault Type
Default Value: 00b
Access: R/W
Type of Fault recorded:
00 - Page Fault.
01 - Invalid PD Fault
10 - Unloaded PD Fault
11 - Invalid and Unloaded PD fault
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
0 Valid Bit
Default Value: 0b
Access: R/W
This bit indicates that the first fault for this engine has been recorded. It can only be cleared by SW,
which will also clear the other fields.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 146
1.1.19.23 ERROR - Main Graphic Arbiter Error Report
ERROR - Main Graphic Arbiter Error Report
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 040A0h-040A3h
This register is used to report differet error conditions. Error bits are writable.
DWord Bit Description
0 31:16 Reserved Bits
Default Value: 0000h
Access: RO
Reserved Bits
15 Reserved Error Bits 15
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
14 Reserved Error Bits 14
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
13 Reserved Error Bits 13
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
12 Reserved Error Bits 12
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
11 Reserved Error Bits 11
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
10 Reserved Error Bits 10
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 147
ERROR - Main Graphic Arbiter Error Report
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
9 Reserved Error Bits 9
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
8 Unloaded PD Error
Default Value: 0b
Access: R/W
Unloaded PD error
The Cache Line containing a PD entry being accessed, was marked as invalid in the last PD load
cycle.
7 Reserved Error Bits 7
Default Value: 0b
Access: R/W
Reserved Error bits (Future expansion)
6 Page Directory Entry VTD Translation Error
Default Value: 0b
Access: R/W
Page Directory entry VTD translation error
PD entry’s VTD translation generated an error (HPA is not accessible for DMA read or write)
4 TLB Page VTD Translation Error
Default Value: 0b
Access: R/W
TLB Page VTD translation error
A TLB Page’s VTD translation generated an error (HPA is not accessible for DMA read or write)
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 148
ERROR - Main Graphic Arbiter Error Report
2 Invalid Page Directory Entry Error
Default Value: 0b
Access: R/W
Invalid Page Directory entry error
PD entry’s valid bit is 0
0 TLB Page Fault Error
Default Value: 0b
Access: R/W
TLB Page Fault error
A TLB Page’s GTT translation generated a page fault (GTT entry not valid)
1.1.19.24 DONE_REG - Gam Fub Done Lookup Register
DONE_REG - Gam Fub Done Lookup Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 040B0h-040B3h
Gam Fub Done Lookup Register
DWord Bit Description
0 31:0 Gam Fub Done Lookup Reg
Default Value: 00000000h
Access: RO
31 CVS Credit Fifo is Empty
30 CVS TLB Don’t have any Cycles
29 Z Credit fifo is empty
28 ZTLB Don’t have any cycles
27 RCC Credit Fifo is empty
26 RCC TLB Don’t have any cycles
25 L3 Credit fifo is empty
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 149
DONE_REG - Gam Fub Done Lookup Register
24 L3 TLB is don’t have any Cycles
23 VLF Credit fifo is empty
22 VLF TLB don’t have any cycles
21 CASC Credit fifo empty
20 CASC TLB don’t have any Cycles
19 Miss Fub Done
18 Read Stream Done
17 Read Steam Fifo is empty
16 Recycle Fifo in rstrm is empty
15 TLB Pend Done
14 TLB Pend PQ Array Is done
13 TLB pend PB Array is done
12 Read route fub is done
11 Gafm Data fifo is empty
10 GAP data fifo is empty
9 GAC data fifo is empty
8 Wrdp is done with all the cycles
7 Wrdp RID fifo is empty
6 No hold from midarb to RTSTRM
5 No hold from TLBPEND to MIDARB
4 VTD Mode
3 Tied to "1" - to be defined
2 Fence FSM are IDLE
1 Non PD Load Done
0 Tied to "1" - to be defined
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 150
1.1.19.25 GAC_HWSP_REG - GAC Hardware Status Page Address Register
GAC_HWSP_REG - GAC Hardware Status Page Address Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04180h-04183h
This register is used to program the 4 KB-aligned System Memory address of the Hardware Status Page used to
report hardware status into (typically cacheable) System Memory. The address in this register is translated using the
Global GTT in memory. The mapping type of the GTT entry determines the snoop nature of the transaction to
memory.
DWord Bit Description
0 31:12 GAC HWSP Register
Default Value: 00000h
Access: R/W
11:0 Reserved Bits
Default Value: 000h
Access: RO
1.1.19.26 MEDIA_ENG_FR - Media Engine Fault Register
MEDIA_ENG_FR - Media Engine Fault Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04194h-04197h
Media Engine Fault Register
DWord Bit Description
0 31:12 Virtual Address of Fault
Default Value: 00000h
Access: R/W
This is the original Address of the Page that generated the First fault for this engine.
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 151
MEDIA_ENG_FR - Media Engine Fault Register
11 GTTSEL
Default Value: 0b
Access: R/W
This bit indicates if the valid bit happened while using PPGTT or GGTT: 0 - PPGTT, 1 - GGTT
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
10:3 SRCID of Fault
Default Value: 00h
Access: R/W
This is the Source ID of the unit that requested the cycle that generated the First Page fault for this
engine.
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
2:1 Fault Type
Default Value: 00b
Access: R/W
Type of Fault recorded:
00 - Page Fault.
01 - Invalid PD Fault
10 - Unloaded PD Fault
11 - Invalid and Unloaded PD fault
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
0 Valid Bit
Default Value: 0b
Access: R/W
This bit indicates that the first fault for this engine has been recorded. It can only be cleared by SW,
which will also clear the other fields.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 152
1.1.19.27 GAB_HWSP_REG - GAB Hardware Status Page Address Register
GAB_HWSP_REG - GAB Hardware Status Page Address Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04280h-04283h
This register is used to program the 4 KB-aligned System Memory address of the Hardware Status Page used to
report hardware status into (typically cacheable) System Memory. The address in this register is translated using
the Global GTT in memory. The mapping type of the GTT entry determines the snoop nature of the transaction to
memory.
DWord Bit Description
0 31:12 GAB HWSP Register
Default Value: 00000h
Access: R/W
11:0 Reserved Bits
Default Value: 000h
Access: RO
1.1.19.28 BLT_ENG_FR - Blitter Engine Fault Register
BLT_ENG_FR - Blitter Engine Fault Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04294h-04297h
Blitter Engine Fault Register
DWord Bit Description
0 31:12 Virtual Address of Page Fault
Default Value: 00000h
Access: R/W
This is the original Address of the Page that generated the First fault for this engine.
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 153
BLT_ENG_FR - Blitter Engine Fault Register
11 Blitter GTTSEL
Default Value: 0b
Access: R/W
This bit indicates if the valid bit happened while using PPGTT or GGTT: 0 - PPGTT, 1 - GGTT
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
10:3 SRCID of Fault
Default Value: 00h
Access: R/W
This is the Source ID of the unit that requested the cycle that generated the First Page fault for this
engine.
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
2:1 Fault Type
Default Value: 00b
Access: R/W
Type of Fault recorded:
00 - Page Fault.
01 - Invalid PD Fault
10 - Unloaded PD Fault
11 - Invalid and Unloaded PD fault
This value is locked and not updated on subsequent faults, until the valid bit of this register is cleared
by SW
0 Valid Bit
Default Value: 0b
Access: R/W
This bit indicates that the first fault for this engine has been recorded. It can only be cleared by SW,
which will also clear the other fields.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 154
1.1.19.29 TLB_RD_ADDR - TLB_RD_ADDRESS Register
TLB_RD_ADDR - TLB_RD_ADDRESS Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04700h-04703h
TLB Read Address
DWord Bit Description
0 31:10 Reserved Bits
Default Value: 0000000000000000000000b
Access: RO
Reserved Bits
9:0 TLB Read Address
Default Value: 0000000000b
Access: R/W
TLB Read Address
MSB<9:X> :
TLB Select <9:X> PAT MSB: Section of the PAT used.
PAT_MSB_VLFTLB 00000 32 entries - 32
PAT_MSB_CVSTLB 00001 32 entries - 32
PAT_MSB_RCCTLB 0001 64 entries - 64
PAT_MSB_ZTLB 001 128 entries - 128
PAT_MSB_L3TLB 01 160 entries - 256
PAT_MSB_CASCTLB 10 140 entries - 256
LSB <X:0> :
GEN RAM ADDRES in Selected TLB
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 155
1.1.19.30 TLB_RD_DATA - TLB_RD_DATA Register
TLB_RD_DATA - TLB_RD_DATA Register
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04704h-04707h
TLB_READ_DATA Register
DWord Bit Description
0 31:0 TLB_READ_DATA Register
Default Value: 00000000h
Access: RO
Return data
1.1.19.31 VLFTLB_VLD_0 - Valid Bit Vector 0 for VLF
VLFTLB_VLD_0 - Valid Bit Vector 0 for VLF
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04720h-04723h
This register contains the valid bits for entries 0-31 of VLFTLB
DWord Bit Description
0 31:0 Valid Bit Vector 0 for VLF
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.32 CVSTLB_VLD_0 - Valid Bit Vector 0 for CVS
CVSTLB_VLD_0 - Valid Bit Vector 0 for CVS
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 156
CVSTLB_VLD_0 - Valid Bit Vector 0 for CVS
Address: 04724h-04727h
This register contains the valid bits for entries 0-31 of CVSTLB
DWord Bit Description
0 31:0 Valid Bit Vector 0 for CVS
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.33 RCCTLB_VLD_0 - Valid Bit Vector 0 for RCC
RCCTLB_VLD_0 - Valid Bit Vector 0 for RCC
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04728h-0472Bh
This register contains the valid bits for entries 0-31 of RCCTLB
DWord Bit Description
0 31:0 Valid Bit Vector 0 for RCC
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.34 RCCTLB_VLD_1 - Valid Bit Vector 1 for RCC
RCCTLB_VLD_1 - Valid Bit Vector 1 for RCC
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 0472Ch-0472Fh
This register contains the valid bits for entries 0-31 of RCCTLB
DWord Bit Description
0 31:0 Valid Bit Vector 1 for RCC
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 157
1.1.19.35 ZTLB_VLD_0 - Valid Bit Vector 0 for Z
ZTLB_VLD_0 - Valid Bit Vector 0 for Z
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04730h-04733h
This register contains the valid bits for entries 0-31 of ZTLB
DWord Bit Description
0 31:0 Valid Bit Vector 0 for Z
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.36 ZTLB_VLD_1 - Valid Bit Vector 1 for Z
ZTLB_VLD_1 - Valid Bit Vector 1 for Z
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04734h-04737h
This register contains the valid bits for entries 0-31 of ZTLB
DWord Bit Description
0 31:0 Valid Bit Vector 1 for Z
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 158
1.1.19.37 ZTLB_VLD_2 - Valid Bit Vector 2 for Z
ZTLB_VLD_2 - Valid Bit Vector 2 for Z
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04738h-0473Bh
This register contains the valid bits for entries 0-31 of ZTLB
DWord Bit Description
0 31:0 Valid Bit Vector 2 for Z
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.38 ZTLB_VLD_3 - Valid Bit Vector 3 for Z
ZTLB_VLD_3 - Valid Bit Vector 3 for Z
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 0473Ch-0473Fh
This register contains the valid bits for entries 0-31 of ZTLB
DWord Bit Description
0 31:0 Valid Bit Vector 3 for Z
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 159
1.1.19.39 L3TLB_VLD_0 - Valid Bit Vector 0 for L3
L3TLB_VLD_0 - Valid Bit Vector 0 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04740h-04743h
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 0 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.40 L3TLB_VLD_1 - Valid Bit Vector 1 for L3
L3TLB_VLD_1 - Valid Bit Vector 1 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04744h-04747h
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 1 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 160
1.1.19.41 L3TLB_VLD_2 - Valid Bit Vector 2 for L3
L3TLB_VLD_2 - Valid Bit Vector 2 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04748h-0474Bh
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 2 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.42 L3TLB_VLD_3 - Valid Bit Vector 3 for L3
L3TLB_VLD_3 - Valid Bit Vector 3 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 0474Ch-0474Fh
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 3 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 161
1.1.19.43 L3TLB_VLD_4 - Valid Bit Vector 4 for L3
L3TLB_VLD_4 - Valid Bit Vector 4 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04750h-04753h
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 4 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.44 L3TLB_VLD_5 - Valid Bit Vector 5 for L3
L3TLB_VLD_5 - Valid Bit Vector 5 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04754h-04757h
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 5 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 162
1.1.19.45 L3TLB_VLD_6 - Valid Bit Vector 6 for L3
L3TLB_VLD_6 - Valid Bit Vector 6 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04758h-0475Bh
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 6 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.46 L3TLB_VLD_7 - Valid Bit Vector 7 for L3
L3TLB_VLD_7 - Valid Bit Vector 7 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 0475Ch-0475Fh
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 7 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 163
1.1.19.47 CASCTLB_VLD_0 - Valid Bit Vector 0 for CASC
CASCTLB_VLD_0 - Valid Bit Vector 0 for CASC
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04760h-04763h
This register contains the valid bits for entries 0-31 of CASCTLB
DWord Bit Description
0 31:0 Valid Bit Vector 0 for CASC
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.48 CASCTLB_VLD_1 - Valid Bit Vector 1 for CASC
CASCTLB_VLD_1 - Valid Bit Vector 1 for CASC
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04764h-04767h
This register contains the valid bits for entries 0-31 of CASCTLB
DWord Bit Description
0 31:0 Valid Bit Vector 1 for CASC
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 164
1.1.19.49 CASCTLB_VLD_2 - Valid Bit Vector 2 for CASC
CASCTLB_VLD_2 - Valid Bit Vector 2 for CASC
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04768h-0476Bh
This register contains the valid bits for entries 0-31 of CASCTLB
DWord Bit Description
0 31:0 Valid Bit Vector 2 for CASC
Default Value: 00000000h
Access: RO
Valid bits per entry
1.1.19.50 CASCTLB_VLD_3 - Valid Bit Vector 3 for CASC
L3TLB_VLD_3 - Valid Bit Vector 3 for L3
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 0474Ch-0474Fh
This register contains the valid bits for entries 0-31 of L3TLB
DWord Bit Description
0 31:0 Valid Bit Vector 3 for L3
Default Value: 00000000h
Access: RO
Valid bits per entry
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 165
1.1.19.51 CASCTLB_VLD_4 - Valid Bit Vector 4 for CASC
CASCTLB_VLD_4 - Valid Bit Vector 4 for CASC
Register Space: MMIO: 0/2/0
Default Value: 0x00000000
Address: 04770h-04773h
This register contains the valid bits for entries 0-31 of CASCTLB
DWord Bit Description
0 31:0 Valid Bit Vector 4 for CASC
Default Value: 00000000h
Access: RO
Valid bits per entry
1.2 Memory Interface Commands for Rendering Engine
1.2.1 Introduction
This chapter describes the formats of the “Memory Interface” commands, including brief descriptions of
their use. The functions performed by these commands are discussed fully in the Memory Interface
Functions Device Programming Environment chapter.
This chapter describes MI Commands for the original graphics processing engine. The term “for
Rendering Engine” in the title has been added to differentiate this chapter from a similar one describing
the MI commands for the Media Decode Engine.
The commands detailed in this chapter are used across products within the Ivy Bridge family. However,
slight changes may be present in some commands (i.e., for features added or removed), or some
commands may be removed entirely. Refer to the Preface chapter for product specific summary.
1.2.2 Software Synchronization Commands
To support mid-triangle interruption, certain commands need to be placed in a temporary location in
hardware until primitive commands are complete. This introduces out-of-order command execution.
Below show the commands that are affected. Note that the INSTPM register has a bit that is used to force
in-order execution. If set, however, mid-triangle modes like PSMI cannot be enabled.
Command Qualifications
MI_NOOP When writing to the NOOPID register
MI_USER_INTERRUPT Always
MI_SEMAPHORE_MBOX Memory write
MI_STORE_DATA_IMM Always
MI_STORE_DATA_INDEX Always
MI_LOAD_REGISTER_IMM Always
MI_UPDATE_GTT Always
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 166
Command Qualifications
MI_STORE_REGISTER_MEM Register read is done in-order, register write done out-of-order
1.2.3 MI_ARB_CHECK
MI_ARB_CHECK
Source: RenderCS
Length Bias: 1
The MI_ARB_CHECK instruction is used to check the ring buffer double buffered head pointer (register UHPTR).
This instruction can be used to pre-empt the current execution of the ring buffer. Note that the valid bit in the updated
head pointer register needs to be set for the command streamer to be pre-empted.
Programming Notes
 The current head pointer is loaded with the updated head pointer register independent of the location of the
updated head
 If the current head pointer and the updated head pointer register are equal, hardware will automatically reset
the valid bit corresponding to the UHPTR
 This instruction can be in either a ring buffer or batch buffer.
 For pre-emption, the wrap count in the ring buffer head register is no longer maintained by hardware. The
hardware updates the wrap count to the value in the UHPTR register.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 05h MI_ARB_CHECK
Format: OpCode
22:0 Reserved
Format: MBZ
1.2.4 MI_ARB_ON_OFF
MI_ARB_ON_OFF
Source: RenderCS
Length Bias: 1
The MI_ARB_ON_OFF instruction is used to disable/enable context switching. Note that context switching will
remain disabled until re-enabled through use of this command.
This command will also prevent a switch in the case of waiting on events, running out of commands. These will
effectively hang the device if allowed to occur while arbitration is off (context switching is disabled.) This command
should always be used as an off-on pair with the sequence of instructions to be protected from context switch
between MI_ARB_OFF and MI_ARB_ON. Software must use this arbitration control with caution since it has the
potential to increase the response time of the Render Engine to pre-emption requests. This is a privileged command;
it will not be effective (will be converted to a no-op) if executed from within a non-privileged batch buffer.
DWord Bit Description
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 167
MI_ARB_ON_OFF
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 08h MI_ARB_ON_OFF
Format: OpCode
22:1 Reserved
Format: MBZ
0 Arbitration Enable
Format: Enable
This field enables or disables context switches due to pre-emption .
1.2.5 MI_BATCH_BUFFER_END
MI_BATCH_BUFFER_END
Source: RenderCS
Length Bias: 1
The MI_BATCH_BUFFER_END command is used to terminate the execution of commands stored in a batch buffer
initiated using a MI_BATCH_BUFFER_START command.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 0Ah MI_ BATCH_BUFFER_END
Format: OpCode
22:0 Reserved
Format: MBZ
1.2.6 MI_CONDITIONAL_BATCH_BUFFER_END
MI_CONDITIONAL_BATCH_BUFFER_END
Source: RenderCS
Length Bias: 2
The MI_BATCH_BUFFER_END command is used to conditionally terminate the execution of commands stored in a
batch buffer initiated using a MI_BATCH_BUFFER_START command.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 168
MI_CONDITIONAL_BATCH_BUFFER_END
Format: OpCode
28:23 MI Command Opcode
Default Value: 36h MI_CONDITIONAL_BATCH_BUFFER_END
Format: OpCode
22 Use Global GTT
Default Value: 0h
Format: U1
If set, this command will use the global GTT to translate the Compare Address and this command
must be executing from a privileged (secure) batch buffer. If clear, the PPGTT will be used to translate
the Compare Address.
21 Compare Semaphore
Default Value: 0h
Format: U1
If set, the value from the Compare Data Dword is compared to the value from the Compare Address in
memory. If the value at Compare Address is greater than the Compare Data Dword, execution of
current command buffer should continue. If clear, no comparison takes place.
19:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 0h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:0 Compare Data Dword
Data dword to compare memory. The Data dword is supplied by software to control execution of the
command buffer. If the compare is enabled and the data at Compare Address is greater than this
dword, the execution of the command buffer should continue.
2 31:3 Compare Address
Qword address to fetch Data Dword(DW0) from memory.
HW will compare the Data Dword(DW0) with Compare Data Dword
2:0 Reserved
Format: MBZ
1.2.7 MI_BATCH_BUFFER_START (Render)
MI_BATCH_BUFFER_START
Source: RenderCS
Length Bias: 2
The MI_BATCH_BUFFER_START command is used to initiate the execution of commands stored in a batch buffer.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 169
MI_BATCH_BUFFER_START
For restrictions on the location of batch buffers, see Batch Buffers in the Device Programming Interface chapter of MI
Functions.
Programming Notes
It is essential that the address location beyond the current page be populated inside the GTT. HW performs over-fetch
of the command addresses and any over-fetch requires a valid TLB entry. A single extra page beyond the batch buffer
is sufficient. Prior to sending batch buffer start command with clear command buffer enable set, software has to
ensure pipe is flushed explicitly by sending MI_FLUSH.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 31h MI_BATCH_BUFFER_START
Format: OpCode
22 Reserved
Format: MBZ
21:17 Reserved
Format: MBZ
16 Reserved
Format: MBZ
15 Reserved
Format: MBZ
14 Reserved
Format: MBZ
13 Reserved
Format: MBZ
11 Clear Command Buffer Enable
Format: Enable
The address of the batch buffer is an offset into the WOPCM area. This batch buffer needs to be
preceded by a MI_FLUSH command or PIPE_CONTROL with CS Stall set.
10 Reserved
Format: MBZ
8 Address Space Indicator
Description Project
SW must ensure the "Address Space Indicator" of the chained batch buffer to be same as the
initial batch buffer. Ex: If the MI_BATCH_BUFFER_START executed from Ring Buffer has
"Address Space Indicator" as "PPGTT" then all subsequent chained batch buffers (not second
level Batch Buffers) must be in "PPGTT".
Not complying to above programming will result in unknown behavior of HW.
Second level batch buffer can select its "Address space Indicator" independent of the parent
batch buffer.
This field must be '0' unless the Per-Process GTT Enable is '1'
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 170
MI_BATCH_BUFFER_START
Value Name Description
0h GGTT This batch buffer will be accessed via the GGTT.
1h PPGTT This batch buffer will be accessed via the PPGTT.
7:0 DWord Length
Default Value: 0h Excludes DWord (0,1)
Format: =n Total - Bias
1 31:2 Batch Buffer Start Address
Format: GraphicsAddress[31:2]BatchBuffer
This field specifies Bits 31:2 of the starting address of the batch buffer.
1:0 Reserved
Format: MBZ
1.2.7.1 Command Access of Privileged Memory
Memory space mapped through the global GTT is considered “privileged” memory. Commands that have
the capability of accessing both privileged and unprivileged (PPGTT space) memory will contain a bit that,
if set, will attempt a “privileged” access through the GGTT rather than an unprivileged access through the
context-local PPGTT.
“User mode” command buffers should not be able to access privileged memory under any circumstances.
These command buffers will be issued by the kernel mode driver with the batch buffer’s Buffer Security
Indicator set to “non-secure”. Commands in such a batch buffer are not allowed to access privileged
memory.
“Kernel mode” command buffers are allowed to access privileged memory. The batch buffers Buffer
Security indicator is set to “secure” in this case. In some of the commands that access memory in a
secure batch buffer, a bit is provided in the command to steer the access to Per process or Global virtual
space. Secure batch buffers are executed from the global GTT.
Commands in ring buffers and commands in batch buffers that are marked as secure (by the kernel mode
driver) are allowed to access both privileged and unprivileged memory and may choose on a command-
by-command basis.
GGTT and PPGTT Usage by Command
Command Address Allowed Access
MI_BATCH_BUFFER_START* Command Address Selectable
MI_DISPLAY_FLIP Display Buffer Base GGTT Only
MI_STORE_DATA_IMM* Storage Address Selectable
MI_STORE_DATA_INDEX** Storage Offset Selectable
MI_STORE_REGISTER_MEM* Storage Address Selectable
MI_SEMAPHORE_MBOX Semaphore Address Selectable
PIPE_CONTROL STDW Address Selectable
*Command has a GGTT/PPGTT selector added to it vs. previous products.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 171
**Added bit allows offset to apply to global HW Status Page or PP HW Status Page found in context
image.
2.1.7.2 Privileged Commands
A subset of the commands are privileged. These commands may be issued only from a secure batch
buffer or directly from a ring. If one of these commands is parsed in a non-secure batch buffer, an error is
flagged and the command is dropped. For commands that generates a write, the hardware will complete
the transaction but the byte enables are turned off. Batch buffers from the User mode driver are passed
directly to the kernel mode driver which does not validate them but issues them with the Security Indicator
set to ‘non-secure’ to protect the system from an attack using these privileged commands.
Privileged Commands
Function in non-privileged batch buffers
Privileged Command
MI_LOAD_REGISTER_IMM Byte enables are turned off
MI_UPDATE_GTT Byte enabled are turned off
MI_STORE_DATA_IMM
Command is translated using the Per-process GTT if Per-Process Virtual Address
Space is set
MI_STORE_DATA_INDEX
MI_STORE_REGISTER_MEM Command is translated and completed with byte enables turned off
MI_DISPLAY_FLIP Command is ignored by the hardware
Parsing one of the commands in the table above from a non-secure batch buffer will flag an error and
convert the command to a NOOP.
1.2.7.2 User Mode Privileged Commands
A subset of the commands are privileged. These commands may be issued only from a secure batch
buffer or directly from a ring. If one of these commands is parsed in a non-secure batch buffer, an error is
flagged and the command is dropped. For commands that generates a write, the hardware will complete
the transaction but the byte enables are turned off. Batch buffers from the User mode driver are passed
directly to the kernel mode driver which does not validate them but issues them with the Security Indicator
set to ‘non-secure’ to protect the system from an attack using these privileged commands.
User Mode Privileged Commands
User Mode Privileged Command Function in non-privileged batch buffers
MI_LOAD_REGISTER_IMM Command is converted to NOOP
MI_UPDATE_GTT Command is converted to NOOP
MI_STORE_DATA_IMM
Command is converted to NOOP if Use Global GTT is enabled.
MI_STORE_DATA_INDEX
Command is converted to NOOP if Use Global GTT is enabled.
MI_STORE_REGISTER_MEM Command is converted to NOOP
MI_DISPLAY_FLIP Command is converted to NOOP
MI_ARB_ON_OFF Command is converted to NOOP
MI_ARB_CHECK Command is converted to NOOP
MI_WAIT_FOR_EVENT Command is converted to NOOP
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 172
1.2.8 MI_CLFLUSH
MI_CLFLUSH
Source: RenderCS
Length Bias: 2
Flushes out the page given in the command out to system memory. This command is specific to the render engine.
This command is not privileged.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 27h Store DW MI_CLFLUSH
Format: OpCode
22 Use Global GTT
This bit will be ignored and treated as if clear when executing from a non-privileged batch buffer. It is
allowed for this bit to be clear when executing this command from a privileged (secure) batch buffer.
This bit must be 1 if the Per Process GTT Enable bit is clear.
Value Name Description
0h Per Process
Graphics Address
1h Global Graphics This command will use the global GTT to translate the Address and this
Address command must be executing from a privileged (secure) batch buffer.
21:10 Reserved
Format: MBZ
9:0 DWord Length
Default Value: 0h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:12 Page Base Address
Format: GraphicsAddress[31:12]
4KB aligned Page Address which software requires hardware to flush to DRAM.
11:6 Starting Cacheline Offset
Format: U6 Zero based starting cacheline offset to the Page Base Address.
5:0 Reserved
Format: MBZ
2 31:16 Address
15:0 Page Base Address
Format: GraphicsAddress[47:32]
This field specifies the 4GB aligned base address of gfx 4GB virtual address space within the host’s
64-bit virtual address space.
3..n 31:0 DW Representing ½ Cache Line
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 173
MI_CLFLUSH
Format: MBZ
The information given to hardware is the DW itself, not the contents. Hardware uses the DW count of
the command to determine the offset from the base to flush out. The offset is ½ cache line (8 DW =
1HW) granular so for a full page, the command will need 4096 bytes / 4 bytes per DW / 8 DW per HW
= 128 DW.
Programming Notes
Always even number of "DW Representing 1/2 cacheline" terms must be programmed.
1.2.9 MI_DISPLAY_FLIP
MI_DISPLAY_FLIP
Source: RenderCS
Length Bias: 2
The MI_DISPLAY_FLIP command is used to request a specific display plane to switch (flip) to display a new buffer.
The buffer is specified with a starting address and pitch. The tiled attribute of the buffer start address is programmed
as part of the packet. The operation this command performs is also known as a “display flip request” operation – in
that the flip operation itself will occur at some point in the future. This command specifies when the flip operation is to
occur: either synchronously with vertical retrace to avoid tearing artifacts (possibly on a future frame), or
asynchronously (as soon as possible) to minimize rendering stalls at the cost of tearing artifacts.
Programming Notes
This command simply requests a display flip operation -- command execution then continues normally. There is no
guarantee that the flip (even if asynchronous) will occur prior to subsequent commands being executed. (Note that
completion of the PIPE_CONTROL command does not guarantee that outstanding flip operations have completed).
The MI_WAIT_FOR_EVENT command must be used to provide this synchronization to avoid back to back
MI_DISPLAY_FLIP commands to the same display plane – by pausing command execution until a pending flip has
actually completed. This synchronization can also be performed by use of the Display Flip Pending hardware
status.
After a display flip operation is requested, software is responsible for initiating any required synchronization with
subsequent buffer clear or rendering operations. For multi-buffering (e.g., double buffering) operations, this will
typically require updating SURFACE_STATE or the binding table to change the rendering (back) buffer. In addition,
prior to any subsequent clear or rendering operations, software must typically ensure that the new rendering buffer
is not actively being displayed. Again, the MI_WAIT_FOR_EVENT command or Display Flip Pending hardware
status can be used to provide this synchronization. See Display Flip Synchronization in the Device Programming
Interface chapter of MI Functions.
The display buffer command uses the X and Y offset for the tiled buffers from the Display Interface registers.
Software is allowed to change the offset via the MMIO interface irrespective of the flip commands enqueued in the
command stream. For tiled buffers, the display subsystem uses the X and Y offset in generation of the final request
to memory. The offset is always updated on the next vblank for both Synchronous and Asynch Flips. It is not
necessary to have a flip enqueued to update the X and Y offset The display buffer command uses the linear dword
offset for the linear buffers from the Display Interface registers. Software is allowed to change the offset via the
MMIO interface irrespective of the flip commands enqueued in the command stream. For linear buffers, the display
subsystem uses the dword offset in generation of the final request to memory. For synchronous flips the offset is
updated on the next vblank. It is not necessary to have a sync flip enqueued to update the dword offset. Linear
memory does not support asynchronous flips DWord 3 (Left Eye Display Buffer Base Address) must not be set with
synchronous flips or asynchronous flips.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 174
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 14h MI_DISPLAY_FLIP
Format: OpCode
22 Async Flip Indicator
Format: Enable
This bit should always be set if DW2 [1:0] == ‘01’ (async flip). This field is required due to HW
limitations. This bit is used by the render pipe while DW2 is used by the display hardware.
21:19 Display (Plane) Select
Format: U3
This field selects which display plane is to perform the flip operation.
Value Name Project
0h Display Plane A
1h Display Plane B
2h Display Sprite A
3h Display Sprite B
4h Display Plane C
5h Display Sprite C
18:8 Reserved
Format: MBZ
7:0 DWord Length
Format: =n
Total Length - 2. Excludes DWord (0,1).
For Synchronous Flips and Asynchronous Flips, this field must be programmed to 1h for a total length
of 3.
Value Name
0h [Default]
1h For Synchronous Flips and Asynchronous Flips
1 31 Reserved
Format: MBZ
30:16 Reserved
Format: MBZ
15:6 Display Buffer Pitch
Default Value: 0h
Format: U10
For Synchronous Flips, this field specifies the 64-byte aligned pitch of the new display buffer. For
Asynchronous Flips, this parameter is programmed so that all the flips in a flip chain should maintain
the same pitch as programmed with the last synchronous flip or direct through MMIO.
5:1 Reserved
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 175
MI_DISPLAY_FLIP
Format: MBZ
0 Tile Parameter
Format: Enable
For Asynchronous Flips, this parameter cannot be changed. All the flips in a flip chain should maintain
the same tile parameter as programmed with the last synchronous flip or direct thru mmio.
Value Name Description
0h Linear [Default] For Syncronous Flips Only
1h Tiled X
2 31:12 Display Buffer Base Address
Format: GraphicsAddress[31:12]
This field specifies Bits 31:12 of the Graphics Address of the new display buffer.
Programming Notes
The Display buffer must reside completely in Main Memory
This address is always translated via the global (rather than per-process) GTT
11:3 Reserved
Format: MBZ
1:0 Flip Type
This field specifies whether the flip operation should be performed asynchronously to vertical retrace.
Value Name Description
00b Sync Flip The flip will occur during the vertical blanking interval – thus avoiding any
[Default] tearing artifacts.
01b Async Flip The flip will occur "as soon as possible" – and may exhibit tearing artifacts
11b Reserved
Programming Notes
Asynch flips are Supported on X-Tiled Frame buffers only.
For Asynch Flips the Buffers used must be 32KB aligned.
1.2.10 MI_FLUSH
MI_FLUSH
Source: RenderCS
Length Bias: 1
Description Project
The MI_FLUSH command is used to perform an internal flush operation. The parser pauses on an internal
flush until all drawing engines have completed any pending operations and the read caches are invalidated
including the texture cache accessed via the Sampler or the data port. In addition, this command can also be
used to:
2.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 176
MI_FLUSH
Flush any dirty data in the Render Cache to memory. This is done by default, however this can be
inhibited.
3.
Invalidate the state and command cache.
Usage note: After this command is completed and followed by a Store DWord-type command, CPU access
to graphics memory will be coherent (assuming the Render Cache flush is not inhibited). This command is
specific to the render engine. Other engines use MI_FLUSH_DW.
In order to use this command, bit 12 in the MI_MODE(0x209c) must be enabled.
If GFX_MODE(0x229C) bit 13, this command will cause a config write to MMIO register space with the
address 0x4f100.
MI_FLUSH command is no longer validated or supported. Use at your own risk.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 04h MI_FLUSH
Format: OpCode
22:7 Reserved
Format: MBZ
5 Indirect State Pointers Disable
Format: Disable
At the completion of the flush, the indirect state pointers in the hardware will be considered as invalid ie
the indirect pointers will not be restored for the context.
4 Generic Media State Clear
Format: Disable
If set, all generic media state context information will not be included with the next context save,
assuming no new state is initiated after the flush. If clear, the generic media state context save state
will not be affected. An MI_FLUSH with this bit set should be issued once all the Media Objects that will
be processed by a given persistent root thread have been issued or when an MI_SET_CONTEXT
switching from a generic media context to a 3D context completes. When using MI_SET_CONTEXT,
once state is programmed, it will be saved and restarted as part of any context each time that context is
saved/restored until an MI_FLUSH with this bit set is issued in that context.
3 Global Snapshot Count Reset
Format: Boolean
The Statistics Counters are also reset; SW should never set this bit during normal operation since the
Statistics Counters are intended to be free running.
Value Name Description
0h Don’t Do not reset the snapshot counts or Statistics Counters.
Reset
1h Reset Reset the snapshot count for all the units and reset the Statistics Counters except as
noted above.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 177
MI_FLUSH
Programming Notes
TIMESTAMP are not reset by MI_FLUSH with this bit set. TIMESTAMP and PS_DEPTH_COUNT can
be reset by writing 0 to them
2 Render Cache Flush Inhibit
Format: Boolean
If set, the Render Cache is not flushed as part of the processing of this command.
Value Name Description
0h Flush Flush the Render Cache
1h Don’t Flush Do not flush the Render Cache
1 State/Instruction Cache Invalidate
Format: Boolean
If set, Invalidates the State and Instruction Cache
Value Name Description
0h Don’t Invalidate Leave State/Instruction Cache unaffected
1h Invalidate Invalidate State/Instruction Cache
0 Reserved
Format: MBZ
1.2.11 MI_LOAD_REGISTER_IMM
MI_LOAD_REGISTER_IMM
Source: RenderCS
Length Bias: 2
The MI_LOAD_REGISTER_IMM command requests a write of up to a DWord constant supplied in the command to
the specified Register Offset (i.e., offset into Memory-Mapped Register Range).
Programming Notes Project
A stalling flush must be sent down pipeline before issuing this command. The behavior of this command is
controlled by Dword 3, Bit 8 (Disable Register Access) of the RINGBUF register. If this command is
disallowed then the command stream converts it to a NOOP.
If this command is executed from a BB then the behavior of this command is controlled by Dword 0, Bit 8
(Security Indicator) of the BATCH_BUFFER_START Command. If the batch buffer is insecure then the
command stream converts this command to a NOOP. Note that the corresponding ring buffer must allow a
register update for this command to execute.
To ensure this command gets executed before upcoming commands in the ring, either a stalling
pipeControl should be sent after this command, or MMIO 0x20C0 bit 7 should be set to 1.
When base address of 0x180000 is added to the Register Offset, when executed will result in updating of
the register in the other GT in GTB mode of operation then the GT from which this instruction is executed.
When this instruction is executed by Command Streamer with COREID-0 will result in updating the register
in GT with COREID-1 and vice versa, when base address of 0x180000 is added to the register offset.
The following addresses should NOT be used for LRIs:
1. 0x8800 - 0x88FF
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 178
MI_LOAD_REGISTER_IMM
2. >= 0xC0000
Limited LRI cycles to the Display Engine 0x40000-0xBFFFF) are allowed, but must be spaced to allow only
one pending at a time. This can be done by issuing an SRM to the same address immediately after each
LRI.
MI_LOAD_REGISTER_IMM command to program Scanline Register followed by Wait For Event command
with Scanline Wait, should always be programmed in the same cacheline together without any commands
(including pipe control) in between and also should be submitted in the same ring dispatch.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 22h MI_LOAD_REGISTER_IMM
Format: OpCode
22:12 Reserved
Format: MBZ
11:8 Byte Write Disables
Format: Enable[4] Bit 8 corresponds to Data DWord [7:0]
Range: Must specify a valid register write operation
If [11:8] is '1111b', then this command will behave as a NOOP.
Otherwise, the value is forwarded to the destination register.
7:0 DWord Length
Default Value: 1h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:2 Register Offset
Format: MmioAddress[31:2]
This field specifies bits [31:2] of the offset into the Memory Mapped Register Range (i.e., this field
specifies a DWord offset). When the base address of 0x180000 is added to the Register Offset, when
executed will result in updating of the register in the other GT in GTB mode of operation then the GT
from which this instruction is executed. When this instruction is executed by Command Streamer with
COREID-0 will result in updating the register in GT with COREID-1 and vice versa, when base address
of 0x180000 is added to the register offset.
1:0 Reserved
Format: MBZ
2 31:0 Data DWord
Mask: Bytes Write Disables
Format: U32
This field specifies the DWord value to be written to the targeted location.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 179
1.2.12 MI_NOOP
MI_NOOP
Source: RenderCS
Length Bias: 1
The MI_NOOP command basically performs a “no operation” in the command stream and is typically used to pad the
command stream (e.g., in order to pad out a batch buffer to a QWord boundary). However, there is one minor
(optional) function this command can perform – a 22-bit value can be loaded into the MI NOPID register. This
provides a general-purpose command stream tagging (“breadcrumb”) mechanism (e.g., to provide sequencing
information for a subsequent breakpoint interrupt).
Programming Notes Project
Performance : The MI_NOOP process time is reduced to 1 clock. An example use of the improved NOOP
throughput is for some multi-pass media applications where some unwanted media object commands are
replaced by MI_NOOP commands without repacking the commands in a batch buffer.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
28:23 MI Command Opcode
Default Value: 0h MI_NOOP
22 Identification Number Register Write Enable
Format: Enable
This field enables the value in the Identification Number field to be written into the MI NOPID
register. If disabled, that register is unmodified, making this command an effective "no operation"
function.
Value Name Description
0h Disable Do not write the NOP_ID register.
1h Enable Write the NOP_ID register.
21:0 Identification Number
Format: U22
This field contains a 22-bit number which can be written to the MI NOPID register.
1.2.13 Surface Probing
These commands are only valid when the “Surface Fault Enable” bit is set in the GFX_MODE register.
1.2.14 MI_REPORT_HEAD
MI_REPORT_HEAD
Source: RenderCS
Length Bias: 1
The MI_REPORT_HEAD command causes the Head Pointer value of the active ring buffer to be written to a
cacheable (snooped) system memory location. The location written is relative to the address programmed in the
Hardware Status Page Address Register.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 180
MI_REPORT_HEAD
Programming Notes
This command must not be executed from a Batch Buffer. (Refer to the description of the HWS_PGA register.)
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 07h MI_REPORT_HEAD
Format: OpCode
22:0 Reserved
Format: MBZ
1.2.15 MI_SEMAPHORE_MBOX
MI_SEMAPHORE_MBOX
Source: RenderCS
Length Bias: 2
This command is provided as alternative to MI_SEMAPHORE to provide mailbox-type semaphores where there is
no update of the semaphore by the checking process (the consumer). Single-bit compare-and-update semantics are
also provided. In either case, atomic access of semaphores need not be guaranteed by hardware as with the
previous command. This command should eventually supersede the previous command.
Synchronization between contexts (especially between contexts running on two different engines) is provided by the
MI_SEMAPHORE_MBOX command. Note that contexts attempting to synchronize in this fashion must be able to
access a common_sli memory location. This means the contexts must share the same virtual address space (have
the same page directory), must have a common physical page mapped into both of their respective address spaces,
or the semaphore commands must be executing from a secure batch buffer or directly from a ring with the Use
Global GTT bit set such that they are privileged and will use the (always shared) global GTT.
MI_SEMAPHORE with the Update Semaphore bit set (and the Compare Semaphore bit clear) implements the
Signal command, while the Wait command is indicated by Compare Semaphore being set. Note that Wait can
cause a context switch. Signal increments unconditionally.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 16h MI_SEMAPHORE_MBOX
Format: OpCode
22 Use Global GTT
If set, this command will use the global GTT to translate the Semaphore Address and this command
must be executing from a privileged (secure) batch buffer. If clear, the PPGTT will be used to translate
the Semaphore Address.
This bit will be ignored (and treated as if clear) if this command is executed from a non-privileged
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 181
MI_SEMAPHORE_MBOX
batch buffer. It is allowed for this bit to be clear when executing this command from a privileged
(secure) batch buffer or directly from a ring buffer.
Programming Notes
This field is only valid when Compare Register Field is reset.
21 Update Semaphore
If set, the value from the Semaphore Data Dword is written to memory. If Compare Semaphore is
also set, the semaphore is not updated if the semaphore comparison fails. If clear, the data at
Semaphore Address is not changed.
Programming Notes
This field should be always clear when Compare Register Field is set.
20 Compare Semaphore
If set, the value from the Semaphore Data Dword is compared to the value from the Semaphore
Address in memory when Compare Register is clear. If set, the value from the Semaphore Data
Dword is compared to the value from MMIO Register selected by Register Select field when
Compare Register is set. If the value at Semaphore Address/MMIO Register is greater than the
Semaphore Data Dword, execution is continued from the current command buffer. If clear, no
comparison takes place. Update Semaphoremust be set in this case.
19 Reserved
Format: MBZ
18 Compare Register
If set, data in MMIO register will be used for compare. If clear, data in memory will be used for
compare.
Programming Notes
Compare Register field should be always set.
17:16 Register Select
Format: Register Select
If Compare Register is set in bit[18], this field indicates which register will be used.
Value Name Description
0h RVSYNC VCS Register
2h RBSYNC BCS Register
3h Use General Register Select
15:14 Reserved
Format: MBZ
13:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 0h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:0 Semaphore Data Dword
Format: U32
Data dword to compare/update memory. The Data dword is supplied by software to control execution
of the command buffer. If the compare is enabled and the data at Semaphore Address is greater than
this dword, the execution of the command buffer continues.
2 31:2 PointerBitFieldName/MMIO Register Address
Format: GraphicsVirtualAddress[31:2]Semaphore
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 182
MI_SEMAPHORE_MBOX
If Compare Register bit[18] is cleared, this field is the Graphics Memory Address of the 32-bit value for
the semaphore. If Compare Register bit[18] is set, this field is the MMIO address of the register for the
semaphore.
1:0 Reserved
Format: MBZ
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 183
1.2.16 MI_SET_CONTEXT
MI_SET_CONTEXT
Source: RenderCS
Length Bias: 2
The MI_SET_CONTEXT command is used to specify the logical context associated with the hardware context. A
logical context is an area in memory used to store hardware context information, and the context is referenced via a
2KB-aligned pointer. If the (new) logical context is different (i.e., at a different memory address), the device saves
the current HW context values to the current logical context address, and then restores (loads) the new logical
context by reading the context from the new address and loading it into the hardware context state. If the logical
context address specified in this command matches the current logical context address, this command is effectively
treated as a NOOP. Specific to the Render command stream only.
This command also includes some controls over the context save/restore process.
 The Force Restore bit can be used to refresh the on-chip device state from the same memory address if the
indirect state buffers have been modified.
 The Restore Inhibit bit can be used to prevent the new context from being loaded at all. This must be used to
prevent an uninitialized context from being loaded. Once software has initialized a context (by setting all state
variables to initial values via commands), the context can then be stored and restored normally.
 This command needs to be always followed by a single MI_NOOP instruction to workaround a silicon issue.
 When switching from a generic media context to a 3D context, the generic media state must be cleared via the
Generic Media State Clear bit 16 in PIPE_CONTROL (or bit 4 in MI_FLUSH) before saving 3D context.
 MI_SET_CONTEXT commands are permitted only within a ring buffer (not within a batch buffer).
Programming Notes Project
Workaround : If Flush TLB Invalidation Mode is enabled it is the driver’s responsibility to invalidate the
TLBs at least once after the previous context switch after any GTT mappings changed (including new GTT
entries). This can be done by a pipelined PIPE_CONTROL with TLB inv bit set immediately before
MI_SET_CONTEXT.
MI_ARB_ON_OFF with 'Arbitration Enable Reset' set should be programmed before an MI_SET_CONTEXT
command. MI_ARB_ON_OFF with 'Arbitration Enable' set should be programmed after an
MI_SET_CONTEXT command. This programming ensures that PSMI context switch flows do not conflict
with MI_SET_CONTEXT flows.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 18h MI_SET_CONTEXT
Format: OpCode
22:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 0h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:12 Logical Context Address
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 184
MI_SET_CONTEXT
Format: GraphicsAddress[31:12]LogicalContext
Description Project
This field contains the 4KB-aligned graphics memory address of the Logical Context that is to
be loaded into the hardware context. If this address is equal to the CCID register associated
with the current ring, no load will occur. Prior to loading this new context, the device will save
the existing context as required. After the context switch operation completes, this address will
be loaded into the associated CCID register.
This field needs to be 4KB aligned virtual address.
11:10 Reserved
Format: MBZ
9 Reserved
Format: MBZ
8 Reserved, Must be 1
Format: Must Be One
7:5 Reserved
Format: MBZ
4 Reserved
Format: MBZ
3 Extended State Save Enable
Format: Enable
If set, the extended state identified in the Logical Context Data section of the Memory Data Formats
chapter is saved as part of switching away from this logical context. This bit will be stored in the
associated CCID register to control the context save operation when switching away from this context
(as part of a subsequent MI_SET_CONTEXT command). This bit must be 1 when RS2 power state is
enabled (via MCHBAR, offset 0x11B8)
2 Extended State Restore Enable
Format: Enable
If set, the extended state identified in the Logical Context Data section of the Memory Data Formats
chapter is loaded (or restored) as part of switching to this logical context. This method can be used to
restore things such as filter coefficients using the indirect state restore followed by a restore of the
extended logical context data. This bit affects the switch (if required) to the context specified in Logical
Context Address. This bit will also be stored in the associated CCID register to control a subsequent
context save operation when switching to this context (as part of a subsequent ring buffer switch). This
bit must be 1 when RS2 power state is enabled (via MCHBAR, offset 0x11B8)
1 Force Restore
When switching to this logical context a comparison between Logical Context Address and the contests
of the CCID register is performed. Normally, matching addresses prevent a context restore from
occurring; however, when this bit is set a context restore is forced to occur. This bit cannot be set with
Restore Inhibit. Note: This bit is not saved in the associated CCID register. It only affects the
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 185
MI_SET_CONTEXT
processing of this command.
0 Restore Inhibit
If set, the restore of the HW context from the logical context specified by Logical Context Address is
inhibited (i.e., the existing HW context values are maintained). This bit must be used to prevent the
loading of an uninitialized logical context. If clear, the context switch proceeds normally. This bit cannot
be set with Force Restore. Note: This bit is not saved in the associated CCID register. It only affects the
processing of this command.
1.2.17 MI_STORE_DATA_IMM
MI_STORE_DATA_IMM
Project: All
Source: RenderCS
Length Bias: 2
The MI_STORE_DATA_IMM command requests a write of the QWord constant supplied in the packet to the
specified Memory Address. As the write targets a System Memory Address, the write operation is coherent with the
CPU cache (i.e., the processor cache is snooped).
Programming Notes
This command should not be used within a "non-privilege" batch buffer to access global virtual space, doing so will
be treated as privilege access violation. Refer “User Mode Privilege Command” in MI_BATCH_BUFFER_START
command section to know HW behavior on encountering privilege access violation. This command can be used
within ring buffers and/or privilege batch buffers to access global virtual space.
This command can be used for general software synchronization through variables in cacheable memory (i.e.,
where software does not need to poll un-cached memory or device registers).
This command simply initiates the write operation with command execution proceeding normally. Although the
write operation is guaranteed to complete eventually, there is no mechanism to synchronize command execution
with the completion (or even initiation) of these operations.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 20h MI_STORE_DATA_IMM
Format: OpCode
22 Use Global GTT
Project: All
Format: Boolean
If set, this command will use the global GTT to translate the Address and this command must be
executing from a privileged (secure) batch buffer. If clear, the PPGTT will be used. It is allowed for this
bit to be clear when executing this command from a privileged (secure) batch buffer. This bit must be
‘1’ if the Per Process GTT Enable bit is clear.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 186
MI_STORE_DATA_IMM
21 Reserved
Format: MBZ
20:10 Reserved
Format: MBZ
9:0 DWord Length
Default Value: 2h
Format: =n Total Length - 2. Excludes DWord (0,1)
Programming Notes
DWord Length programmed must not exceed 0x3FE
1 31:0 Reserved
Format: MBZ
2 31:2 Address
Format: GraphicsAddress[31:2]U32(2)
This field specifies Bits 31:2 of the Address where the DWord will be stored. As the store address must
be DWord-aligned, Bits 1:0 of that address MBZ. This address must be 8B aligned for a store "QW"
command.
1:0 Reserved
Format: MBZ
3 31:0 Data DWord 0
Format: U32
This field specifies the DWord value to be written to the targeted location.For a QWord write this
DWord is the lower DWord of the QWord to be reported (DW 0).
4 31:0 Data DWord 1
Format: U32
This field specifies the upper DWord value to be written to the targeted QWord location (DW 1).
1.2.18 MI_STORE_DATA_INDEX
MI_STORE_DATA_INDEX
Source: RenderCS
Length Bias: 2
The MI_STORE_DATA_INDEX command requests a write of the data constant supplied in the packet to the
specified offset from the System Address defined by the Hardware Status Page Address Register. As the write
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 187
MI_STORE_DATA_INDEX
targets a System Address, the write operation is coherent with the CPU cache (i.e., the processor cache is snooped).
Programming Notes
-
 Use of this command with an invalid or uninitialized value in the Hardware Status Page Address Register is
UNDEFINED.
 This command can be used for general software synchronization through variables in cacheable memory (i.e.,
where software does not need to poll uncached memory or device registers).
 This command simply initiates the write operation with command execution proceeding normally. Although the
write operation is guaranteed to complete eventually, there is no mechanism to synchronize command
execution with the completion (or even initiation) of these operations.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 21h MI_STORE_DATA_INDEX
Format: OpCode
22 Reserved
21 Reserved
Format: MBZ
20:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 1h
Format: =n Total Length - 2. Excludes DWord (0,1 ) = 1 for DWord, 2 for QWord.
1 31:12 Reserved
Format: MBZ
11:2 Offset
Format: U10 zero-based DWord offset into the HW status page.
Format: HardwareStatusPageOffset[11:2]U32
This field specifies the offset (into the hardware status page) to which the data will be written. Note that
the first few DWords of this status page are reserved for special-purpose data storage – targeting these
reserved locations via this command is UNDEFINED. This address must be 8B aligned for a store QW
command.
Value Name
[16, 1023]
1:0 Reserved
Format: MBZ
2 31:0 Data DWord 0
Format: U32
This field specifies the DWord value to be written to the targeted location.For a QWord write this
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 188
MI_STORE_DATA_INDEX
DWord is the lower DWord of the QWord to be reported (DW 0).
3 31:0 Data DWord 1
Format: U32
This field specifies the upper DWord value to be written to the targeted QWord location (DW 1).
1.2.19 MI_STORE_REGISTER_MEM
MI_STORE_REGISTER_MEM
Project: All
Source: RenderCS
Length Bias: 2
The MI_STORE_REGISTER_MEM command requests a register read from a specified memory mapped register
location in the device and store of that DWord to memory. The register address is specified along with the command
to perform the read.
Programming Notes
The command temporarily halts command execution.
The memory address for the write is snooped on the host bus.
This command should not be used from within a "non-privilege" batch buffer to access global virtual space. doing
so will be treated as privilege access violation. Refer "User Mode Privilege Command" in
MI_BATCH_BUFFER_START command section to know HW behavior on encountering privilege access violation.
This command can be used within ring buffers and/or "privilege" batch buffers to access global virtual space.
This command will cause undefined data to be written to memory if given register addresses for the
PGTBL_CTL_0 or FENCE registers.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 24h MI_STORE_REGISTER_MEM
Format: OpCode
22 Use Global GTT
It is allowed for this bit to be set when executing this command from a privileged (secure) batch or ring
buffer. This bit must be clear when programmed from within a non-privileged batch buffer. This bit
must be 1 if the Per Process GTT Enable bit is clear.
Value Name Description Project
0h Per Process
Graphics Address
1h Global Graphics This command will use the global GTT to translate the Address and
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 189
MI_STORE_REGISTER_MEM
Address this command must be executing from a privileged (secure) batch
buffer.
21 Reserved
Format: MBZ
20:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 1h Excludes DWord (0,1)
Format: =n Total Length - 2
1 31:26 Reserved
Format: MBZ
25:2 Register Address
Format: MMIOAddress[25:2]MMIO_Register
This field specifies Bits 25:2 of the Register offset the DWord will be read from. As the register
address must be DWord-aligned, Bits 1:0 of that address MBZ.
Programming Notes
 Storing a VGA register is not permitted and will store an UNDEFINED value.
 The values of PGTBL_CTL0 or any of the FENCE registers cannot be stored to memory;
UNDEFINED values will be written to memory if the addresses of these registers are specified.
1:0 Reserved
Format: MBZ
2 31:2 Memory Address
Format: GraphicsAddress[31:2]MMIO_Register
This field specifies the address of the memory location where the register value specified in the
DWord above will be written. The address specifies the DWord location of the data.Range =
GraphicsVirtualAddress[31:2] for a DWord register
1:0 Reserved
Format: MBZ
1.2.20 MI_SUSPEND_FLUSH
MI_SUSPEND_FLUSH
Source: RenderCS
Length Bias: 1
Description Project
Blocks MMIO sync flush or any flushes related to VT-d while enabled.
Programming Notes Project
SW must ensure MI_SUSPEND_FLUSH with "Suspend Flush" enabled have a corresponding
MI_SUSPEND_FLUSH with "Suspend Flush" disabled in the same ring dispatch. SW must also ensure not
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 190
MI_SUSPEND_FLUSH
to program MI_WAIT_FOR_EVENT command when "Suspend Flush" is enabled.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 0Bh MI_SUSPEND_FLUSH
Format: OpCode
22:1 Reserved
Format: MBZ
0 Suspend Flush
Format: Enable
Description Project
This field suspends flush due and IOTLB invalidation.
Programming Notes Project
Workaround: Make sure that Suspend Flush Enabled and Suspend Flush Disabled are in
the same tail update.
1.2.21 MI_UPDATE_GTT
MI_UPDATE_GTT
Source: RenderCS
Length Bias: 2
The MI_UPDATE_GTT command is used to update GTT page table entries in a coherent manner and at a
predictable place in the command flow.
An MI_FLUSH should be placed before this command, since work associated with preceding commands that are
still in the pipeline may be referencing GTT entries that will be changed by its execution. The flush also invalidates
TLBs and read caches that may become invalid as a result of the changed GTT entries. MI_FLUSH is not required if
it can be guaranteed that the pipeline is free of any work that relies on changing GTT entries (such as
MI_UPDATE_GTT contained in a paging DMA buffer that is doing only update/mapping activities and no rendering).
This is a privileged command.
Note that MI_UPDATE_GTT is mainly for the pages that are strictly used by GT. If driver chooses to update the
CPU used pages thru MI_UPDATE_GTT, it needs to write any value to MMIO address 0x101008 to ensure system
agent TLBs are invalidated before the new pages can be used.
PPGTT updates cannot be done via MI_UPDATE_GTT; gfx driver will have to use MI_STORE_DATA_IMM for
PPGTT inline updates.
DWord Bit Description
0 31:29 Command Type
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 191
MI_UPDATE_GTT
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 23h MI_UPDATE_GTT
Format: OpCode
22 Use Global GTT
Reserved: Must be 1h. Updating Per Process Graphics Address is not supported.
Value Name Description
0h Per Process Graphics Address
1h Global Graphics Address
21:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 0h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:12 Entry Address
Format: GraphicsAddress[31:12]
This field simply holds the DW offset of the first table entry to be modified. Note that one or more of
the upper bits may need to be 0, i.e., for a 2G aperture, bit 31 MBZ.
11:0 Reserved
Format: MBZ
2..n 31:0 Entry Data
Format: Table Entry
This Dword becomes the new page table entry. See PPGTT/Global GTT Table Entries (PTEs) in
Memory Interface Registers.
1.2.22 MI_USER_INTERRUPT
MI_USER_INTERRUPT
Source: RenderCS
Length Bias: 1
The MI_USER_INTERRUPT command is used to generate a User Interrupt condition. The parser will continue
parsing after processing this command. See User Interrupt.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 02h MI_USER_INTERRUPT
Format: OpCode
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 192
MI_USER_INTERRUPT
22:0 Reserved
Format: MBZ
1.2.23 MI_WAIT_FOR_EVENT
MI_WAIT_FOR_EVENT
Source: RenderCS
Length Bias: 1
Description Project
The MI_WAIT_FOR_EVENT command is used to pause command stream processing of this pipe only
until a specific event occurs or while a specific condition exists. See Wait Events/Conditions, Device
Programming Interface in MI Functions. Only one event/condition can be specified. Specifying multiple
events is UNDEFINED.
Once parsed, the parser will halt (and suspend command arbitration) until the event/condition occurs. Note
that if a specified condition does not exist (the condition code is inactive) at the time the parser executes
this command, the parser proceeds, treating this command as a no-operation.
If CSunit is waiting for V-blank or flip done, HW can go into RC1/RC6 state.
MI_NOOP setting NOP register (or any other benign command) must be set after MI_WAIT_FOR_EVENT
under the following conditions:
 Back-to-back MI_WAIT_FOR_EVENT commands
 MI_WAIT_FOR_EVENT is the last command before head = tail
Events must be unmasked in the Display Engine Render Response Mask Register
(DE RRMR 0x44050) prior to waiting for them with a MI_WAIT_FOR_EVENT command, or in the case of
flips
or scanlines, prior to starting the flip or loading the scanline. Unmasked events will wake command
streamer as they occur, so for improved power savings it is recommended to only unmask events that are
required. Programming the DE RRMR register can be done through MMIO or a
LOAD_REGISTER_IMMEDIATE
command.
Programming Notes Project
Software must always program MI_NOOP command with “Identification Number Register Write Enable” set
following MI_WAIT_FOR_EVENT command to avoid know HW issue.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 03h MI_WAIT_FOR_EVENT
Format: OpCode
22 Display Pipe C Horizontal Blank Wait Enable
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 193
MI_WAIT_FOR_EVENT
Format: Enable
This field enables a wait until the start of next Display Pipe C Horizontal Blank event occurs. This
event is described as the start of the next Display C Horizontal blank period. Note that this can cause
a wait for up to a line.
21 Display Pipe C Vertical Blank Wait Enable
Format: Enable
This field enables a wait until the next Display Pipe C Vertical Blank event occurs. This event is
described as the start of the next Display C vertical blank period. Note that this can cause a wait for up
to an entire refresh period.
20 Display Sprite C Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Sprite C Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers).
19:16 Condition Code Wait Select
This field enables a wait for the duration that the corresponding condition code is active. These enable
select one of 15 condition codes in the EXCC register, that cause the parser to wait until that
condition-code in the EXCC is cleared.
Value Name Description
0h Not enabled Condition Code Wait Not Enabled
1h-5h Enable Condition Code Select Enabled; selects one of 5 codes, 0 – 4
6h-15h Reserved
Programming Notes
Note that not all condition codes are implemented. The parser operation is UNDEFINED if an
unimplemented condition code is selected by this field. The description of the EXCC register (Memory
Interface Registers) lists the codes that are implemented.
15 Display Plane C Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane C “Flip Pending” condition. If a flip request
is pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers).
14 Display Pipe C Scan Line Wait Enable
Format: Enable
This field enables a wait while a Display Pipe C Scan Line condition exists. This condition is defined as
the start of the scan line specified in the Pipe C Display Scan Line Count Range Compare Register.
13 Display Pipe B Horizontal Blank Wait Enable
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 194
MI_WAIT_FOR_EVENT
Format: Enable
This field enables a wait until the start of next Display Pipe B “Horizontal Blank” event occurs. This
event is described as the start of the next Display B Horizontal blank period. Note that this can cause a
wait for up to a line.
12 Reserved
Format: MBZ
11 Display Pipe B Vertical Blank Wait Enable
Format: U32
This field enables a wait until the next Display Pipe B “Vertical Blank” event occurs. This event is
described as the start of the next Display Pipe B vertical blank period. Note that this can cause a wait
for up to an entire refresh period.
10 Display Sprite B Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Sprite B “Flip Pending” condition. If a flip request
is pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers).
9 Display Plane B Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane B Flip Pending condition. If a flip request is
pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers.
8 Display Pipe B Scan Line Wait Enable
Format: Enable
This field enables a wait while a Display Pipe B Scan Line condition exists. This condition is defined as
the start of the scan line specified in the Pipe B Display Scan Line Count Range Compare Register.
7:6 Reserved
Format: MBZ
5 Display Pipe A Horizontal Blank Wait Enable
Format: U32
This field enables a wait until the start of next Display Pipe A Horizontal Blank event occurs. This
event is described as the start of the next Display A Horizontal blank period. Note that this can cause a
wait for up to a line.
4 Reserved
Format: MBZ
3 Display Pipe A Vertical Blank Wait Enable
Format: U32
This field enables a wait until the next Display Pipe A “Vertical Blank” event occurs. This event is
described as the start of the next Display Pipe A vertical blank period. Note that this can cause a wait
for up to an entire refresh period.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 195
MI_WAIT_FOR_EVENT
2 Display Sprite A Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Sprite A “Flip Pending” condition. If a flip request
is pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers).
1 Display Plane A Flip Pending Wait Enable
Format: Enable
This field enables a wait for the duration of a Display Plane A “Flip Pending” condition. If a flip request
is pending, the parser will wait until the flip operation has completed (i.e., the new front buffer address
has now been loaded into the active front buffer registers).
0 Display Pipe A Scan Line Wait Enable
Format: Enable
This field enables a wait while a Display Pipe A “Scan Line” condition exists. This condition is defined
as the start of the scan line specified in the Pipe A Display Scan Line Count Range Compare Register.
1.2.24 MI_LOAD_REGISTER_MEM
MI_LOAD_REGISTER_MEM
Source: RenderCS
Length Bias: 2
The MI_LOAD_REGISTER_MEM command requests from a memory location and stores that DWord to a register.
Programming Notes
The command temporarily halts commands that will cause cycles down the 3D pipeline.
This command should not be used within a non-privilege batch buffer to access global virtual space, doing so will
be treated as privilege access violation. Refer "User Mode Privilege Command" in MI_BATCH_BUFFER_START
command section to know HW behavior on encountering privilege access violation.
This command is not allowed to update the privilege register range when executed from a non-privilege batch
buffer.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 29h MI_LOAD_REGISTER_MEM
Format: OpCode
22 Use Global GTT
This bit if set when executing from a non-privileged batch buffer will be treated as privilege access
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 196
MI_LOAD_REGISTER_MEM
violation. It is allowed for this bit to be clear when executing this command from a privileged (secure)
batch buffer or ring buffer.
Value Name Description
0h Per Process
Graphics Address
1h Global Graphics This command will use the global GTT to translate the Address and this
Address command must be executing from a privileged (secure) batch buffer.
21 Async Mode Enable
Description Project
If this bit is set then the command stream will not wait for completion of this command before
executing the next command. Please refer to the LOAD_INDIRECT and Predicate registers
for usage of this bit.
20:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 1h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:26 Reserved
Format: MBZ
25:2 Register Address
Format: MMIOAddress[22:2]MMIO_Register
This field specifies Bits 25:2 of the Register offset the DWord will be written to. As the register address
must be DWord-aligned, Bits 1:0 of that address MBZ.
1:0 Reserved
Format: MBZ
2 31:2 Memory Address
Format: GraphicsAddress[31:2]MMIO_Register
This field specifies the address of the memory location where the register value specified in the
DWord above will read from. The address specifies the DWord location of the data.
Range = GraphicsVirtualAddress[31:2] for a DWord register
1:0 Reserved
Format: MBZ
1.2.25 MI_URB_CLEAR
MI_URB_CLEAR
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 197
MI_URB_CLEAR
Source: RenderCS
Length Bias: 2
The MI_URB_CLEAR command allows SW to clear (write zero) to a section in the URB.
Programming Notes
 The command temporarily halts command execution.
 This command is part of context save/restore. Only the last instance will be part of context.
 This command requires the 3D pipeline to be flushed before execution.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 19h MI_URB_CLEAR
Format: OpCode
22:8 Reserved
Format: MBZ
7:0 DWord Length
Default Value: 0h
Format: =n Total Length - 2. Excludes DWord (0,1).
1 31:30 Reserved
Format: MBZ
29 Reserved
Format: MBZ
28:16 URB Clear Length
This field specifies the number of 256b entries in the URB to be cleared to zero.
Value Name
[0,8191]
15 Reserved
Format: MBZ
14 Reserved
Format: MBZ
13:0 URB Address
Format: URBAddress[18:5] 256b aligned
This field specifies Bits 18:5 of the URB Address
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 198
1.2.26 MI_PREDICATE
The MI_PREDICATE command is used to control the Predicate state bit, which in turn can be used to
enable/disable the processing of 3DPRIMITIVE commands.
MI_PREDICATE
Source: RenderCS
Length Bias: 1
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Format: OpCode
28:23 MI Command Opcode
Default Value: 0Ch MI_PREDICATE
Format: OpCode
22:8 Reserved
Format: MBZ
7:6 Load Operation
This field controls if/how the Predicate state bit is modified.
Value Name Description
0h LOADOP_KEEP The Predicate state bit is unmodified.
1h Reserved
2h LOADOP_LOAD The Predicate state bit is loaded with the combine operation result.
3h LOADOP_LOADINV The Predicate state bit is loaded with the inverted combine operation result.
5 Reserved
Format: MBZ
4:3 Combine Operation
This field controls if/how the result of the compare operation is combined with the current Predicate
state bit.
Value Name Description
0h COMBINEOP_SET The combine operation output the compare result unmodified.
1h COMBINEOP_AND The combine operation outputs the AND of the compare result and the
current Predicate state bit.
2h COMBINEOP_OR The combine operation outputs the OR of the compare result and the
current Predicate state bit.
3h COMBINEOP_XOR The combine operation outputs the XOR of the compare result and the
current Predicate state bit.
2 Reserved
Format: MBZ
1:0 Compare Operation
This field controls how Data DWord 0 and Data DWord 1 fields are used to generate a compare
operation result and possibly modify the PredicateData register.
Value Name Description
0h COMPAREOP_TRUE The compare operation outputs TRUE. The PredicateData
register is unmodified.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 199
MI_PREDICATE
1h COMPAREOP_FALSE The compare operation outputs FALSE. The PredicateData
register is unmodified.
2h COMPAREOP_SRCS_EQUAL (MItemp0 – MItemp1) is computed and loaded into the
PredicateData register. The compare operation outputs
(MItemp0 == MItemp1).
3h COMPAREOP_DELTAS_EQUAL (MItemp0 – MItemp1) is computed and compared to the
PredicateData register. If the values are equal, the compare
operation outputs TRUE, otherwise it outputs FALSE. The
PredicateData register is unmodified.
1.2.26.1 Predicated Rendering Support in HW
DX10 defines predicated rendering, where sequences of rendering commands can be discarded based
on the result of a previous predicate test. A new state bit, Predicate, has been added to the command
stream. In addition, a PredicateEnable bit is added to 3DPRIMITIVE. When the PredicateEnable bit is set,
the command is ignored if the Predicate state bit is set.
A new command, MI_PREDICATE, is added. It contains several control fields which specify how the
Predicate bit is generated.
Refer to the diagram below and the command description for details.
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 200
MI_PREDICATE Function
MI_LOAD_REGISTER_MEM commands can be used to load the MItemp0, MItemp1 and PredicateData
registers prior to MI_PREDIATE. In order to ensure the memory sources of the
MI_LOAD_REGISTER_MEM commands are coherent with previous 3D_PIPECONTROL store-dword
operations, software can use the new Pipe Control Flush Enable bit in the PIPE_CONTROL command.
1.2.27 MI_TOPOLOGY_FILTER
MI_TOPOLOGY_FILTER
Source: RenderCS
Length Bias: 1
This command is used to specify a specific 3DPrimType value, where the CS will ignore all 3DPRIMITIVE commands
that do no have a matching 3DPrimType. This primitive culling is optional (turned off by using this command with a
Topology Filter Value of 0). This command is specific to the Render command stream only.
DWord Bit Description
0 31:29 Command Type
Default Value: 0h MI_COMMAND
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 201
MI_TOPOLOGY_FILTER
Format: OpCode
28:23 MI Command Opcode
Default Value: 0Dh MI_TOPOLOGY_FILTER
Format: OpCode
22:6 Reserved
Format: MBZ
5:0 Topology Filter Value
Format: 3D_PrimTopoType
When non-zero, the CS will discard all 3DPRIMITIVE commands which do not match the specified
3DPrimTopologyType. When zero, no filtering is performed (normal operation).
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 202
Revision History
Revision Number Description Revision Date
1.0 First 2012 OpenSource edition May 2012
§§
Doc Ref #: IHD-OS-V1 Pt 3 – 05 12 5/31/2012 203
Reference in new issue View Git Blame Copy Permalink