Skip to content

[SPIR-V][Doc] Add JointMatrixWorkItemLengthINTEL instruction to joint matrix extension #5781

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Apr 1, 2022
Merged

[SPIR-V][Doc] Add JointMatrixWorkItemLengthINTEL instruction to joint matrix extension #5781

Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
29dfc48
Add JointMatrixWorkItemLengthINTEL instruction to joint matrix extension
MrSidims Mar 10, 2022
9b1cdf0
Fix version
MrSidims Mar 10, 2022
bceac1a
Restrict result type
MrSidims Mar 10, 2022
73e2b6b
Update SPV_INTEL_joint_matrix.asciidoc
MrSidims Mar 16, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
58 changes: 45 additions & 13 deletions sycl/doc/design/spirv-extensions/SPV_INTEL_joint_matrix.asciidoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,7 @@
:OpJointMatrixSUMadINTEL_token: 6128
:OpJointMatrixUSMadINTEL_token: 6129
:OpJointMatrixUUMadINTEL_token: 6130
:OpJointMatrixWorkItemLengthINTEL_token: 6410

{extension_name}
================
Expand Down Expand Up @@ -52,8 +53,8 @@ please let us know!

[width="40%",cols="25,25"]
|========================================
| Last Modified Date | 2021-12-28
| Revision | 3
| Last Modified Date | 2022-03-10
| Revision | 4
|========================================

== Dependencies
Expand All @@ -66,7 +67,7 @@ This extension requires SPIR-V 1.0.
== Overview

This extension adds a type and instructions for joint matrices. Such matrices
are shared among a group of work items and is not private to each work item.
are shared among a group of work-items and is not private to each work-item.
The type introduced with this extension allows to specify memory scope and
layout of the matrix, including layouts optimized for particular hardware(AMX) .
New instructions also allow to specify synchronization scope.
Expand Down Expand Up @@ -104,6 +105,7 @@ OpJointMatrixMADINTEL
OpJointMatrixSUMADINTEL
OpJointMatrixUSMADINTEL
OpJointMatrixUUMADINTEL
OpJointMatrixWorkItemLengthINTEL

----

Expand All @@ -113,14 +115,15 @@ OpJointMatrixUUMADINTEL
[cols="70%,30%"]
[grid="rows"]
|====
|*{capability_name}* | {capability_token}
|*OpTypeJointMatrixINTEL* | {OpTypeJointMatrixINTEL_token}
|*OpJointMatrixLoadINTEL* | {OpJointMatrixLoadINTEL_token}
|*OpJointMatrixStoreINTEL* | {OpJointMatrixStoreINTEL_token}
|*OpJointMatrixMadINTEL* | {OpJointMatrixMadINTEL_token}
|*OpJointMatrixSUMadINTEL* | {OpJointMatrixSUMadINTEL_token}
|*OpJointMatrixUSMadINTEL* | {OpJointMatrixUSMadINTEL_token}
|*OpJointMatrixUUMadINTEL* | {OpJointMatrixUUMadINTEL_token}
|*{capability_name}* | {capability_token}
|*OpTypeJointMatrixINTEL* | {OpTypeJointMatrixINTEL_token}
|*OpJointMatrixLoadINTEL* | {OpJointMatrixLoadINTEL_token}
|*OpJointMatrixStoreINTEL* | {OpJointMatrixStoreINTEL_token}
|*OpJointMatrixMadINTEL* | {OpJointMatrixMadINTEL_token}
|*OpJointMatrixSUMadINTEL* | {OpJointMatrixSUMadINTEL_token}
|*OpJointMatrixUSMadINTEL* | {OpJointMatrixUSMadINTEL_token}
|*OpJointMatrixUUMadINTEL* | {OpJointMatrixUUMadINTEL_token}
|*OpJointMatrixWorkItemLengthINTEL* | {OpJointMatrixWorkItemLengthINTEL_token}
|====

== Modifications to the SPIR-V Specification, Version 1.5
Expand Down Expand Up @@ -207,7 +210,6 @@ result of a constant instruction with scalar 'integer type'. +
'Scope'
|=====


==== 3.37.8. Memory Instructions

[cols="1,1,7*3",width="100%"]
Expand Down Expand Up @@ -253,7 +255,6 @@ specifying the <<Memory_Operands,*memory operand*>> *None*. +
'Memory Access'
|=====


[cols="1,1,6*3",width="100%"]
|=====
7+|[[OpJointMatrixStoreINTEL]]*OpJointMatrixStoreINTEL* +
Expand Down Expand Up @@ -296,6 +297,36 @@ specifying the <<Memory_Operands,*memory operand*>> *None*. +
'Memory Access'
|=====

==== 3.37.12. Composite Instructions
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why this section is numbered 3.37.12 instead of 3.37.9?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

3.37.9 (or 3.42.9 in 1.6 SPIR-V spec) is for Function Instructions


Modify *OpVectorExtractDynamic* and *OpVectorInsertDynamic* to accept
<<OpTypeJointMatrixINTEL,*OpTypeJointMatrixINTEL*>> as the 'Vector' operand.
In this case the instructions operate on an implicit vector which repesents
part of the joint matrix and holds components owned by the current work-item.
If the 'index' operand of these instructions exceeds the value returned by
<<OpJointMatrixWorkItemLengthINTEL,*OpJointMatrixWorkItemLengthINTEL*>>,
behavior is undefined.

[cols="1,1,3*3",width="100%"]
|=====
4+|[[OpJointMatrixWorkItemLengthINTEL]]*OpJointMatrixWorkItemLengthINTEL* +
+
Return number of components owned by the current work-item in a joint matrix. +
+
'Result Type' must be an 32-bit unsigned integer type scalar. +
+
'Matrix' is the <<OpTypeJointMatrixINTEL,*OpTypeJointMatrixINTEL*>> to query the
number of the components. +

1+|Capability: +
*{capability_name}*
1+| 4 | {OpJointMatrixWorkItemLengthINTEL_token}
| '<id>' +
'Result Type'
| 'Result <id>'
| '<id>' +
'Matrix'
|=====

==== 3.37.13. Arithmetic Instructions

Expand Down Expand Up @@ -493,4 +524,5 @@ Revision History
|1|2021-02-16|Alexey Sotkin|Initial revision
|2|2021-09-06|Dmitry Sidorov|Split OpJointMatrixMadINTEL instruction into 4
|3|2021-12-28|Dmitry Sidorov|Add Joint Matrix to Composite definition
|4|2022-03-10|Dmitry Sidorov|Add OpJointMatrixWorkItemLengthINTEL instruction
|========================================