Skip to content

[mlir] AMDGPUToROCDL: handle 1-element vectors #128266

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 1 commit into from
Feb 23, 2025
Merged

[mlir] AMDGPUToROCDL: handle 1-element vectors #128266

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
6 changes: 6 additions & 0 deletions mlir/lib/Conversion/AMDGPUToROCDL/AMDGPUToROCDL.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,6 +167,12 @@ struct RawBufferOpLowering : public ConvertOpToLLVMPattern<GpuOp> {
}
}
}
if (auto vecType = dyn_cast<VectorType>(llvmBufferValType)) {
// Buffer intrinsics doesn't support 1-element vectors, cast them to
// scalars.
if (vecType.getNumElements() == 1)
llvmBufferValType = vecType.getElementType();
}

SmallVector<Value, 6> args;
if (storeData) {
Expand Down
17 changes: 17 additions & 0 deletions mlir/test/Conversion/AMDGPUToROCDL/amdgpu-to-rocdl.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,6 +76,15 @@ func.func @gpu_gcn_raw_buffer_load_i32_oob_off(%buf: memref<64xi32>, %idx: i32)
func.return %0 : i32
}

// CHECK-LABEL: func @gpu_gcn_raw_buffer_load_1xi32
func.func @gpu_gcn_raw_buffer_load_1xi32(%buf: memref<64xi32>, %idx: i32) -> vector<1xi32> {
// CHECK: %[[ret:.*]] = rocdl.raw.ptr.buffer.load %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : i32
// CHECK: %[[cast:.*]] = llvm.bitcast %[[ret]] : i32 to vector<1xi32>
// CHECK: return %[[cast]]
%0 = amdgpu.raw_buffer_load {boundsCheck = true} %buf[%idx] : memref<64xi32>, i32 -> vector<1xi32>
func.return %0 : vector<1xi32>
}

// CHECK-LABEL: func @gpu_gcn_raw_buffer_load_2xi32
func.func @gpu_gcn_raw_buffer_load_2xi32(%buf: memref<64xi32>, %idx: i32) -> vector<2xi32> {
// CHECK: %[[ret:.*]] = rocdl.raw.ptr.buffer.load %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : vector<2xi32>
Expand Down Expand Up @@ -159,6 +168,14 @@ func.func @gpu_gcn_raw_buffer_store_i32(%value: i32, %buf: memref<64xi32>, %idx:
func.return
}

// CHECK-LABEL: func @gpu_gcn_raw_buffer_store_1xf32
func.func @gpu_gcn_raw_buffer_store_1xf32(%value: vector<1xf32>, %buf: memref<64xf32>, %idx: i32) {
// CHECK: %[[cast:.*]] = llvm.bitcast %{{.*}} : vector<1xf32> to f32
// CHECK: rocdl.raw.ptr.buffer.store %[[cast]], %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32
amdgpu.raw_buffer_store {boundsCheck = true} %value -> %buf[%idx] : vector<1xf32> -> memref<64xf32>, i32
func.return
}

// CHECK-LABEL: func @gpu_gcn_raw_buffer_store_2xi8
func.func @gpu_gcn_raw_buffer_store_2xi8(%value: vector<2xi8>, %buf: memref<64xi8>, %idx: i32) {
// CHECK: %[[cast:.*]] = llvm.bitcast %{{.*}} : vector<2xi8> to i16
Expand Down