[mlir] Add optimization attrs for gpu-to-llvmspv function declarations and calls #99301
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
|
@llvm/pr-subscribers-mlir @llvm/pr-subscribers-mlir-gpu Author: Finlay (FMarno) ChangesAdds the attributes nounwind and willreturn to all function declarations. Adds Full diff: https://github.com/llvm/llvm-project/pull/99301.diff 2 Files Affected:
diff --git a/mlir/lib/Conversion/GPUToLLVMSPV/GPUToLLVMSPV.cpp b/mlir/lib/Conversion/GPUToLLVMSPV/GPUToLLVMSPV.cpp
index ebeb8f803d71d..81682a52c8c4b 100644
--- a/mlir/lib/Conversion/GPUToLLVMSPV/GPUToLLVMSPV.cpp
+++ b/mlir/lib/Conversion/GPUToLLVMSPV/GPUToLLVMSPV.cpp
@@ -40,11 +40,10 @@ namespace mlir {
// Helper Functions
//===----------------------------------------------------------------------===//
-static LLVM::LLVMFuncOp lookupOrCreateSPIRVFn(Operation *symbolTable,
- StringRef name,
- ArrayRef<Type> paramTypes,
- Type resultType,
- bool isConvergent = false) {
+static LLVM::LLVMFuncOp
+lookupOrCreateSPIRVFn(Operation *symbolTable, StringRef name,
+ ArrayRef<Type> paramTypes, Type resultType,
+ bool hasMemoryEffects = true, bool isConvergent = false) {
auto func = dyn_cast_or_null<LLVM::LLVMFuncOp>(
SymbolTable::lookupSymbolIn(symbolTable, name));
if (!func) {
@@ -53,6 +52,17 @@ static LLVM::LLVMFuncOp lookupOrCreateSPIRVFn(Operation *symbolTable,
symbolTable->getLoc(), name,
LLVM::LLVMFunctionType::get(resultType, paramTypes));
func.setCConv(LLVM::cconv::CConv::SPIR_FUNC);
+ func.setNoUnwind(true);
+ func.setWillReturn(true);
+ if (!hasMemoryEffects) {
+ // no externally observable effects
+ constexpr auto noModRef = mlir::LLVM::ModRefInfo::NoModRef;
+ auto memAttr = b.getAttr<LLVM::MemoryEffectsAttr>(
+ /*other*/ noModRef,
+ /*argMem*/ noModRef, /*inaccessibleMem*/ noModRef);
+ func.setMemoryAttr(memAttr);
+ }
+
func.setConvergent(isConvergent);
}
return func;
@@ -91,8 +101,9 @@ struct GPUBarrierConversion final : ConvertOpToLLVMPattern<gpu::BarrierOp> {
assert(moduleOp && "Expecting module");
Type flagTy = rewriter.getI32Type();
Type voidTy = rewriter.getType<LLVM::LLVMVoidType>();
- LLVM::LLVMFuncOp func = lookupOrCreateSPIRVFn(
- moduleOp, funcName, flagTy, voidTy, /*isConvergent=*/true);
+ LLVM::LLVMFuncOp func =
+ lookupOrCreateSPIRVFn(moduleOp, funcName, flagTy, voidTy,
+ /*hasMemoryEffects*/ true, /*isConvergent=*/true);
// Value used by SPIR-V backend to represent `CLK_LOCAL_MEM_FENCE`.
// See `llvm/lib/Target/SPIRV/SPIRVBuiltins.td`.
@@ -134,8 +145,8 @@ struct LaunchConfigConversion : ConvertToLLVMPattern {
assert(moduleOp && "Expecting module");
Type dimTy = rewriter.getI32Type();
Type indexTy = getTypeConverter()->getIndexType();
- LLVM::LLVMFuncOp func =
- lookupOrCreateSPIRVFn(moduleOp, funcName, dimTy, indexTy);
+ LLVM::LLVMFuncOp func = lookupOrCreateSPIRVFn(
+ moduleOp, funcName, dimTy, indexTy, /*hasMemoryEffects*/ false);
Location loc = op->getLoc();
gpu::Dimension dim = getDimension(op);
@@ -268,9 +279,9 @@ struct GPUShuffleConversion final : ConvertOpToLLVMPattern<gpu::ShuffleOp> {
Type valueType = adaptor.getValue().getType();
Type offsetType = adaptor.getOffset().getType();
Type resultType = valueType;
- LLVM::LLVMFuncOp func =
- lookupOrCreateSPIRVFn(moduleOp, funcName, {valueType, offsetType},
- resultType, /*isConvergent=*/true);
+ LLVM::LLVMFuncOp func = lookupOrCreateSPIRVFn(
+ moduleOp, funcName, {valueType, offsetType}, resultType,
+ /*hasMemoryEffects*/ true, /*isConvergent=*/true);
Location loc = op->getLoc();
std::array<Value, 2> args{adaptor.getValue(), adaptor.getOffset()};
diff --git a/mlir/test/Conversion/GPUToLLVMSPV/gpu-to-llvm-spv.mlir b/mlir/test/Conversion/GPUToLLVMSPV/gpu-to-llvm-spv.mlir
index 1b0f89a9a573e..d094c786414fc 100644
--- a/mlir/test/Conversion/GPUToLLVMSPV/gpu-to-llvm-spv.mlir
+++ b/mlir/test/Conversion/GPUToLLVMSPV/gpu-to-llvm-spv.mlir
@@ -4,16 +4,46 @@
// RUN: | FileCheck --check-prefixes=CHECK-32,CHECK %s
gpu.module @builtins {
- // CHECK-64: llvm.func spir_funccc @_Z14get_num_groupsj(i32) -> i64
- // CHECK-64: llvm.func spir_funccc @_Z12get_local_idj(i32) -> i64
- // CHECK-64: llvm.func spir_funccc @_Z14get_local_sizej(i32) -> i64
- // CHECK-64: llvm.func spir_funccc @_Z13get_global_idj(i32) -> i64
- // CHECK-64: llvm.func spir_funccc @_Z12get_group_idj(i32) -> i64
- // CHECK-32: llvm.func spir_funccc @_Z14get_num_groupsj(i32) -> i32
- // CHECK-32: llvm.func spir_funccc @_Z12get_local_idj(i32) -> i32
- // CHECK-32: llvm.func spir_funccc @_Z14get_local_sizej(i32) -> i32
- // CHECK-32: llvm.func spir_funccc @_Z13get_global_idj(i32) -> i32
- // CHECK-32: llvm.func spir_funccc @_Z12get_group_idj(i32) -> i32
+ // CHECK-64: llvm.func spir_funccc @_Z14get_num_groupsj(i32) -> i64 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-64: llvm.func spir_funccc @_Z12get_local_idj(i32) -> i64 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-64: llvm.func spir_funccc @_Z14get_local_sizej(i32) -> i64 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-64: llvm.func spir_funccc @_Z13get_global_idj(i32) -> i64 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-64: llvm.func spir_funccc @_Z12get_group_idj(i32) -> i64 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-32: llvm.func spir_funccc @_Z14get_num_groupsj(i32) -> i32 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-32: llvm.func spir_funccc @_Z12get_local_idj(i32) -> i32 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-32: llvm.func spir_funccc @_Z14get_local_sizej(i32) -> i32 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-32: llvm.func spir_funccc @_Z13get_global_idj(i32) -> i32 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
+ // CHECK-32: llvm.func spir_funccc @_Z12get_group_idj(i32) -> i32 attributes {
+ // CHECK-SAME-DAG: memory = #llvm.memory_effects<other = none, argMem = none, inaccessibleMem = none>
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: will_return
// CHECK-LABEL: gpu_block_id
func.func @gpu_block_id() -> (index, index, index) {
@@ -104,7 +134,11 @@ gpu.module @builtins {
// -----
gpu.module @barriers {
- // CHECK: llvm.func spir_funccc @_Z7barrierj(i32) attributes {convergent}
+ // CHECK: llvm.func spir_funccc @_Z7barrierj(i32) attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
// CHECK-LABEL: gpu_barrier
func.func @gpu_barrier() {
@@ -120,10 +154,26 @@ gpu.module @barriers {
// Check `gpu.shuffle` conversion with default subgroup size.
gpu.module @shuffles {
- // CHECK: llvm.func spir_funccc @_Z22sub_group_shuffle_downdj(f64, i32) -> f64 attributes {convergent}
- // CHECK: llvm.func spir_funccc @_Z20sub_group_shuffle_upfj(f32, i32) -> f32 attributes {convergent}
- // CHECK: llvm.func spir_funccc @_Z21sub_group_shuffle_xorlj(i64, i32) -> i64 attributes {convergent}
- // CHECK: llvm.func spir_funccc @_Z17sub_group_shuffleij(i32, i32) -> i32 attributes {convergent}
+ // CHECK: llvm.func spir_funccc @_Z22sub_group_shuffle_downdj(f64, i32) -> f64 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
+ // CHECK: llvm.func spir_funccc @_Z20sub_group_shuffle_upfj(f32, i32) -> f32 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
+ // CHECK: llvm.func spir_funccc @_Z21sub_group_shuffle_xorlj(i64, i32) -> i64 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
+ // CHECK: llvm.func spir_funccc @_Z17sub_group_shuffleij(i32, i32) -> i32 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
// CHECK-LABEL: gpu_shuffles
// CHECK-SAME: (%[[VAL_0:.*]]: i32, %[[VAL_1:.*]]: i32, %[[VAL_2:.*]]: i64, %[[VAL_3:.*]]: i32, %[[VAL_4:.*]]: f32, %[[VAL_5:.*]]: i32, %[[VAL_6:.*]]: f64, %[[VAL_7:.*]]: i32)
@@ -155,10 +205,26 @@ gpu.module @shuffles {
gpu.module @shuffles attributes {
spirv.target_env = #spirv.target_env<#spirv.vce<v1.4, [Kernel, Addresses, GroupNonUniformShuffle, Int64], []>, #spirv.resource_limits<subgroup_size = 16>>
} {
- // CHECK: llvm.func spir_funccc @_Z22sub_group_shuffle_downdj(f64, i32) -> f64 attributes {convergent}
- // CHECK: llvm.func spir_funccc @_Z20sub_group_shuffle_upfj(f32, i32) -> f32 attributes {convergent}
- // CHECK: llvm.func spir_funccc @_Z21sub_group_shuffle_xorlj(i64, i32) -> i64 attributes {convergent}
- // CHECK: llvm.func spir_funccc @_Z17sub_group_shuffleij(i32, i32) -> i32 attributes {convergent}
+ // CHECK: llvm.func spir_funccc @_Z22sub_group_shuffle_downdj(f64, i32) -> f64 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
+ // CHECK: llvm.func spir_funccc @_Z20sub_group_shuffle_upfj(f32, i32) -> f32 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
+ // CHECK: llvm.func spir_funccc @_Z21sub_group_shuffle_xorlj(i64, i32) -> i64 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
+ // CHECK: llvm.func spir_funccc @_Z17sub_group_shuffleij(i32, i32) -> i32 attributes {
+ // CHECK-SAME-DAG: no_unwind
+ // CHECK-SAME-DAG: convergent
+ // CHECK-NOT: memory = #llvm.memory_effects
+ // CHECK-SAME: }
// CHECK-LABEL: gpu_shuffles
// CHECK-SAME: (%[[VAL_0:.*]]: i32, %[[VAL_1:.*]]: i32, %[[VAL_2:.*]]: i64, %[[VAL_3:.*]]: i32, %[[VAL_4:.*]]: f32, %[[VAL_5:.*]]: i32, %[[VAL_6:.*]]: f64, %[[VAL_7:.*]]: i32)
|
FMarno
force-pushed
the
gpu_to_llvmspv_add_attrs
branch
from
151c7de to
4c5887e
Compare
victor-eds
reviewed
Jul 17, 2024
kuhar
changed the title
victor-eds
reviewed
Jul 18, 2024
victor-eds
approved these changes
Jul 18, 2024
FMarno
force-pushed
the
gpu_to_llvmspv_add_attrs
branch
from
c458336 to
54d16fe
Compare
Adds attributes including nounwind, willreturn and memory(none).
FMarno
force-pushed
the
gpu_to_llvmspv_add_attrs
branch
from
1c237ad to
a097821
Compare
FMarno
changed the title
victor-eds
approved these changes
Jul 23, 2024
Dinistro
approved these changes
Jul 24, 2024
yuxuanchen1997
pushed a commit
that referenced
this pull request
Jul 25, 2024
…s and calls (#99301) Summary: Adds the attributes nounwind and willreturn to all function declarations. Adds `memory(none)` equivalent to the id/dimension function declarations. The function declaration attributes are copied to the function calls. `nounwind` is legal because there are no exception in SPIR-V. I also do not see any reason why any of these functions would not return when used correctly. I'm confident that the get id/dim functions will have no externally observable memory effects, but think the convergent functions will have effects. Test Plan: Reviewers: Subscribers: Tasks: Tags: Differential Revision: https://phabricator.intern.facebook.com/D60250575
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Adds the attributes nounwind and willreturn to all function declarations. Adds
memory(none)equivalent to the id/dimension function declarations. The function declaration attributes are copied to the function calls.nounwindis legal because there are no exception in SPIR-V. I also do not see any reason why any of these functions would not return when used correctly.I'm confident that the get id/dim functions will have no externally observable memory effects, but think the convergent functions will have effects.