Skip to content

[mlir] Refactor ConvertVectorToLLVMPass options #128219

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 9 commits into from
Mar 10, 2025
Merged

[mlir] Refactor ConvertVectorToLLVMPass options #128219

Changes from 1 commit
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
6910060
Explicitly speciy all vector transform options on ConvertVectorToLLVM…
abulavin Feb 14, 2025
4b8967b
fixup: Reference vectorContractOptions instead of VectorTransformOptions
abulavin Feb 25, 2025
1e61aae
fixup: add check that non-default options are set correctly
abulavin Feb 26, 2025
4cc7f7d
fixup: Use the clearer NON-DEFAULT check-prefix
abulavin Feb 26, 2025
d2cca0c
fixup: correct vector.matmul typo to vector.matrix_multiply
abulavin Feb 27, 2025
0523d03
fixup: remove unused tablegen include
abulavin Feb 27, 2025
e3c44b4
fixup: remove redundnancy from lit test filename
abulavin Feb 27, 2025
ad26777
fixup: add documentation justifying the serialization test
abulavin Feb 27, 2025
c65d803
fixup: add back include for VectorTransformOptions.summary
abulavin Feb 28, 2025
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
13 changes: 9 additions & 4 deletions mlir/test/Conversion/VectorToLLVM/test-serialisable.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,9 +1,12 @@
// RUN: mlir-opt --convert-vector-to-llvm --dump-pass-pipeline %s 2>&1 | FileCheck %s

// Simple regression test that ensures ConvertVectorToLLVMPass options remain
// serialisable. We don't need to actually parse any IR to print the pass
// options. We just need to provide --dump-pass-pipeline

// RUN: mlir-opt --convert-vector-to-llvm --dump-pass-pipeline %s 2>&1 | FileCheck %s --check-prefix=DEFAULT

// RUN: mlir-opt --convert-vector-to-llvm='vector-contract-lowering=matmul vector-transpose-lowering=flat' \
// RUN: --dump-pass-pipeline 2>&1 | FileCheck %s --check-prefix=CHANGED

// CHECK: builtin.module(
// CHECK-SAME: convert-vector-to-llvm{
// CHECK-SAME: enable-amx={{[aA-zZ0-9]+}}
Expand All @@ -12,5 +15,7 @@
// CHECK-SAME: enable-x86vector={{[aA-zZ0-9]+}}
// CHECK-SAME: force-32bit-vector-indices={{[aA-zZ0-9]+}}
// CHECK-SAME: reassociate-fp-reductions={{[aA-zZ0-9]+}}
// CHECK-SAME: vector-contract-lowering={{[aA-zZ0-9]+}}
// CHECK-SAME: vector-transpose-lowering={{[aA-zZ0-9]+}}})
// DEFAULT: vector-contract-lowering=dot
// DEFAULT: vector-transpose-lowering=eltwise
// CHANGED: vector-contract-lowering=matmul
// CHANGED: vector-transpose-lowering=flat