[0035] Address a few linalg typos and clarifications#800
Open
pow2clk wants to merge 2 commits intomicrosoft:mainfrom
Open
[0035] Address a few linalg typos and clarifications#800pow2clk wants to merge 2 commits intomicrosoft:mainfrom
pow2clk wants to merge 2 commits intomicrosoft:mainfrom
Conversation
Mostly changes the inlined header to be consistent with itself, other parts of the spec and other corrections for compilability Renames new vector types to not clash with previous spec in common header fixes microsoft#796 Finish update to SIZE_TYPE in header. Remove unused IsNativeScalar boolean. These finish the cleanups that microsoft#712 started to fix microsoft#570 Finish update to align mat/vec multiplies with coopvec spec. Finishes microsoft#741 Update mat/vec multiply operations to not take a scope parameter, limiting them to thread scope Fixes microsoft#797 Remove unused template type for accumulate() methods fixes microsoft#795 Renumbered componenttype enum to be consistent with internal numbers fixes microsoft#798 Added `p` prefix to overload names of groupshared dxil ops for consistency. Corrects typo in microsoft#769 fixing microsoft#756 Consolidated dxil op overloads that must be the same for correctness. This required reordering parameters in some cases to preserve in-order usage fixes microsoft#799 Corrected latent mention of linAlgMatrixMulOp
V-FEXrt
reviewed
Feb 18, 2026
Revert changes to typenames in linalg.h meant to prevent clashing with coopvec as those types have been moved to another header. Correct list of supported type enums and numbers. Correct commas in dxil op parameter descriptions Revert dxil intrinsics with always identical extended overloads to have an overload for each overloaded parameter. Rmove .p overload types. Brings them into consitency by eliminating them all as agreed
llvm-beanz
reviewed
Mar 4, 2026
|
|
||
| template <ComponentEnum LHSTy, ComponentEnum RHSTy, | ||
| MatrixUseEnum UseLocal = Use> | ||
| template <ComponentEnum T, MatrixUseEnum UseLocal = Use> |
Collaborator
There was a problem hiding this comment.
Suggested change
| template <ComponentEnum T, MatrixUseEnum UseLocal = Use> | |
| template <ComponentEnum ElTy, MatrixUseEnum UseLocal = Use> |
I prefer to not use single-letter template argument names except in cases like the M, N, and K dimensions where the names directly match the documentation.
| #define __MATRIX_SCALAR_COMPONENT_MAPPING(enum_val, type) \ | ||
| template <> struct ComponentTypeTraits<enum_val> { \ | ||
| using Type = type; \ | ||
| static const bool IsNativeScalar = true; \ |
Collaborator
There was a problem hiding this comment.
I suspect we may need these since they may need these, so I'd rather not remove them at least yet.
|
|
||
| ## Appendix 1: HLSL Header | ||
|
|
||
| [Compiler Explorer](https://godbolt.org/z/habj4EnaW) |
Collaborator
There was a problem hiding this comment.
Can you update this with the updated code?
llvm-beanz
reviewed
Mar 6, 2026
| F32 = 9, | ||
| F64 = 10, | ||
| PackedS8x32 = 17, | ||
| PackedU8x32 = 18, |
Collaborator
There was a problem hiding this comment.
On reflection, while I do think we should have the value of these enumerations match the DXIL constants, we should probably not include the Packed types here since they're not valid element types in the linalg templates.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Mostly changes the inlined header to be consistent with itself, other parts of the spec and other corrections for compilability
Finish update to SIZE_TYPE in header.
Remove unused IsNativeScalar boolean.
These finish the cleanups that #712 started to fix #570
Finish update to align mat/vec multiplies with coopvec spec. Finishes #741
Update mat/vec multiply operations to not take a scope parameter, limiting them to thread scope Fixes #797
Remove unused template type for accumulate() methods fixes #795
Renumbered componenttype enum to be consistent with internal numbers fixes #798
Removed latent
pprefix to overload names of groupshared dxil ops for consistency. Corrects omission in #769 fixing #756Corrected latent mention of linAlgMatrixMulOp