-
Notifications
You must be signed in to change notification settings - Fork 581
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.
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
[LLVMGPU] Pad to intrinsic shape in LLVMGPUPadAndVectorDistribute pipeline #18632
[LLVMGPU] Pad to intrinsic shape in LLVMGPUPadAndVectorDistribute pipeline #18632
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't understand the batch matmul test
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/test/ROCDL/pipeline_vector_distribute_gfx940.mlir
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/test/ROCDL/pipeline_vector_distribute_gfx940.mlir
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/test/ROCDL/pipeline_vector_distribute_gfx940.mlir
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/test/ROCDL/pipeline_vector_distribute_gfx940.mlir
Outdated
Show resolved
Hide resolved
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
0a2539e
to
abad5bc
Compare
I modified the first test to use lowering_config to decide tile sizes multiples. |
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/LLVMGPUPromoteMatmulToFitMMA.cpp
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/test/promote_matmul_to_fit_mma.mlir
Outdated
Show resolved
Hide resolved
compiler/src/iree/compiler/Codegen/LLVMGPU/test/ROCDL/pipeline_vector_distribute_gfx940.mlir
Outdated
Show resolved
Hide resolved
cef81b0
to
47934f0
Compare
This patch makes LLVMGPUPromoteToFitMMA pass pad to a multiple of intrinsic shape, instead of padding to 1.
Fixes #18602