[mlir][tosa] Relax constraint on matmul verifier requiring equal operand types #155799
Conversation
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
…and types Removes the verifier constraint allowing support for matmul with different operand types such as fp8e5m2xfp8e4m3. Support for specific operand types strictly adhering to the TOSA specification will still be caught in the validation pass. Change-Id: I1453ded48326ea0460fa6caf52651c02b7d8c055
Member
|
@llvm/pr-subscribers-mlir-tosa @llvm/pr-subscribers-mlir Author: Luke Hutton (lhutton1) ChangesRemoves the verifier constraint allowing support for matmul with different operand types such as fp8e5m2xfp8e4m3. Support for specific operand types strictly adhering to the TOSA specification will still be caught in the validation pass. Full diff: https://github.com/llvm/llvm-project/pull/155799.diff 2 Files Affected:
diff --git a/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp b/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
index 3cafb199d2db3..d58acf6a53d5e 100644
--- a/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
+++ b/mlir/lib/Dialect/Tosa/IR/TosaOps.cpp
@@ -1605,12 +1605,6 @@ LogicalResult MatMulOp::verify() {
return emitOpError("expect quantized operands to have same widths, got ")
<< aQuantWidth << " and " << bQuantWidth;
}
- } else {
- // non-quantized element types
- if (aElementType != bElementType) {
- return emitOpError("expect same element type for inputs a and b, got ")
- << aElementType << " and " << bElementType;
- }
}
// check a_zp and b_zp
diff --git a/mlir/test/Dialect/Tosa/ops.mlir b/mlir/test/Dialect/Tosa/ops.mlir
index 30361a882afe5..6061c04662de8 100644
--- a/mlir/test/Dialect/Tosa/ops.mlir
+++ b/mlir/test/Dialect/Tosa/ops.mlir
@@ -934,6 +934,15 @@ func.func @test_matmul_f8E5M2(%arg0: tensor<1x14x19xf8E5M2>, %arg1: tensor<1x19x
return %0 : tensor<1x14x28xf16>
}
+// -----
+// CHECK-LABEL: test_matmul_f8E5M2_f8E4M3
+func.func @test_matmul_f8E5M2_f8E4M3(%arg0: tensor<1x14x19xf8E5M2>, %arg1: tensor<1x19x28xf8E4M3FN>) -> tensor<1x14x28xf16> {
+ %azp0 = "tosa.const"() <{values = dense<0.0> : tensor<1xf8E5M2>}> : () -> tensor<1xf8E5M2>
+ %bzp0 = "tosa.const"() <{values = dense<0.0> : tensor<1xf8E4M3FN>}> : () -> tensor<1xf8E4M3FN>
+ %0 = tosa.matmul %arg0, %arg1, %azp0, %bzp0 : (tensor<1x14x19xf8E5M2>, tensor<1x19x28xf8E4M3FN>, tensor<1xf8E5M2>, tensor<1xf8E4M3FN>) -> tensor<1x14x28xf16>
+ return %0 : tensor<1x14x28xf16>
+}
+
// -----
// CHECK-LABEL: max_pool2d_f8E5M2
func.func @test_max_pool2d_f8E5M2(%arg0: tensor<1x32x32x8xf8E5M2>) -> tensor<1x32x32x8xf8E5M2> {
|
Contributor
|
LGTM |
ckoparkar
added a commit
to ckoparkar/llvm-project
that referenced
this pull request
Sep 25, 2025
* main: (502 commits) GlobalISel: Adjust insert point when expanding G_[SU]DIVREM (llvm#160683) [LV] Add coverage for fixing-up scalar resume values (llvm#160492) AMDGPU: Convert wave_any test to use update_mc_test_checks [LV] Add partial reduction tests multiplying extend with constants. Revert "[MLIR] Implement remark emitting policies in MLIR" (llvm#160681) [NFC][InstSimplify] Refactor fminmax-folds.ll test (llvm#160504) [LoongArch] Pre-commit tests for [x]vldi instructions with special constant splats (llvm#159228) [BOLT] Fix dwarf5-dwoid-no-dwoname.s (llvm#160676) [lldb][test] Refactor and expand TestMemoryRegionDirtyPages.py (llvm#156035) [gn build] Port 833d5f0 AMDGPU: Ensure both wavesize features are not set (llvm#159234) [LoopInterchange] Bail out when finding a dependency with all `*` elements (llvm#149049) [libc++] Avoid constructing additional objects when using map::at (llvm#157866) [lldb][test] Make hex prefix optional in DWARF union types test [X86] Add missing prefixes to trunc-sat tests (llvm#160662) [AMDGPU] Fix vector legalization for bf16 valu ops (llvm#158439) [LoongArch][NFC] Pre-commit tests for `[x]vadda.{b/h/w/d}` [mlir][tosa] Relax constraint on matmul verifier requiring equal operand types (llvm#155799) [clang][Sema] Accept gnu format attributes (llvm#160255) [LoongArch][NFC] Add tests for element extraction from binary add operation (llvm#159725) ...
mahesh-attarde
pushed a commit
to mahesh-attarde/llvm-project
that referenced
this pull request
Oct 3, 2025
…and types (llvm#155799) Removes the verifier constraint allowing support for matmul with different operand types such as fp8e5m2xfp8e4m3. Support for specific operand types strictly adhering to the TOSA specification will still be caught in the validation pass.
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.
Removes the verifier constraint allowing support for matmul with different operand types such as fp8e5m2xfp8e4m3. Support for specific operand types strictly adhering to the TOSA specification will still be caught in the validation pass.