Skip to content

[AMD] DCE/canonicalize true epilogue conditionals #6314

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 7 commits into from
Apr 1, 2025
Merged

[AMD] DCE/canonicalize true epilogue conditionals #6314

Show file tree
Hide file tree
Changes from 5 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
64eb6d9
[AMD] DCE/canonicalize true epilogue conditionals
makslevental Mar 26, 2025
95d7dc0
add fold-true-cmpi pattern/test pass
makslevental Mar 28, 2025
efefa95
add fold-true-cmpi pattern to StreamPipeline.cpp
makslevental Mar 28, 2025
0e55f55
add tests
makslevental Mar 28, 2025
9bc5cec
special case in test_assume
makslevental Apr 1, 2025
e6703de
address comments
makslevental Apr 1, 2025
74a5f63
fix bitPosition error
makslevental Apr 1, 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
2 changes: 2 additions & 0 deletions bin/RegisterTritonDialects.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,6 +34,7 @@ void registerTestAlignmentPass();
void registerTestAllocationPass();
void registerTestMembarPass();
void registerTestTritonAMDGPURangeAnalysis();
void registerTestTritonAMDGPUFoldTrueCmpIOp();
} // namespace test
} // namespace mlir

Expand All @@ -47,6 +48,7 @@ inline void registerTritonDialects(mlir::DialectRegistry &registry) {
mlir::test::registerTestAllocationPass();
mlir::test::registerTestMembarPass();
mlir::test::registerTestTritonAMDGPURangeAnalysis();
mlir::test::registerTestTritonAMDGPUFoldTrueCmpIOp();
mlir::triton::registerConvertTritonToTritonGPUPass();
mlir::triton::gpu::registerAllocateSharedMemoryPass();
mlir::triton::gpu::registerTritonGPUAllocateWarpGroups();
Expand Down
5 changes: 4 additions & 1 deletion python/test/unit/language/test_core.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4666,7 +4666,10 @@ def _kernel(out_ptr, N: tl.constexpr, BLOCK_N: tl.constexpr):
if is_interpreter():
return

assert 'llvm.assume' in pgm.asm['llir']
assert 'llvm.intr.assume' in pgm.asm['ttgir']
# stream pipeliner on AMD folds true cmpi ops to %true (Which llvm itself then dces)
if not is_hip():
assert 'llvm.assume' in pgm.asm['llir']


# ---------------
Expand Down
158 changes: 158 additions & 0 deletions test/TritonGPU/amd/amd-fold-true-cmpi.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,158 @@
// RUN: triton-opt %s -split-input-file -allow-unregistered-dialect -test-tritonamdgpu-fold-true-cmpi -canonicalize | FileCheck %s

module attributes {"ttg.num-warps" = 4 : i32} {
tt.func @cmpsle(%arg0: !tt.ptr<f32>) -> i1 {
%c0 = arith.constant 0 : i32
%c1024_i32 = arith.constant 1024 : i32
%cmpsle = arith.cmpi sle, %c0, %c1024_i32 : i32
tt.return %cmpsle: i1
}
}

// CHECK-LABEL: tt.func @cmpsle(
// CHECK-SAME: %[[VAL_0:.*]]: !tt.ptr<f32>) -> i1 {
// CHECK: %[[VAL_1:.*]] = arith.constant true
// CHECK: tt.return %[[VAL_1]] : i1
// CHECK: }

// -----

module attributes {"ttg.num-warps" = 4 : i32} {
tt.func @assumepid(%arg0: !tt.ptr<f32>) -> tensor<1024xf32> {
%c0 = arith.constant 0 : i32
%c1024_i32 = arith.constant 1024 : i32
%pid = tt.get_program_id x : i32
%cmpsle = arith.cmpi sle, %pid, %c1024_i32 : i32
llvm.intr.assume %cmpsle : i1
%cmpsge = arith.cmpi sge, %pid, %c0 : i32
llvm.intr.assume %cmpsge : i1
%1 = arith.muli %pid, %c1024_i32 : i32
%2 = tt.addptr %arg0, %1 : !tt.ptr<f32>, i32
%3 = tt.splat %2 : !tt.ptr<f32> -> tensor<1024x!tt.ptr<f32>>
%4 = tt.load %3 : tensor<1024x!tt.ptr<f32>>
tt.return %4 : tensor<1024xf32>
}
}

// CHECK-LABEL: tt.func @assumepid(
// CHECK-SAME: %[[VAL_0:.*]]: !tt.ptr<f32>) -> tensor<1024xf32> {
// CHECK: %[[VAL_1:.*]] = arith.constant true
// CHECK: %[[VAL_2:.*]] = arith.constant 1024 : i32
// CHECK: %[[VAL_3:.*]] = tt.get_program_id x : i32
// CHECK: llvm.intr.assume %[[VAL_1]] : i1
// CHECK: llvm.intr.assume %[[VAL_1]] : i1
// CHECK: %[[VAL_4:.*]] = arith.muli %[[VAL_3]], %[[VAL_2]] : i32
// CHECK: %[[VAL_5:.*]] = tt.addptr %[[VAL_0]], %[[VAL_4]] : !tt.ptr<f32>, i32
// CHECK: %[[VAL_6:.*]] = tt.splat %[[VAL_5]] : !tt.ptr<f32> -> tensor<1024x!tt.ptr<f32>>
// CHECK: %[[VAL_7:.*]] = tt.load %[[VAL_6]] : tensor<1024x!tt.ptr<f32>>
// CHECK: tt.return %[[VAL_7]] : tensor<1024xf32>
// CHECK: }

// -----

#blocked = #ttg.blocked<{sizePerThread = [1, 4], threadsPerWarp = [1, 32], warpsPerCTA = [4, 1], order = [1, 0]}>
#blocked1 = #ttg.blocked<{sizePerThread = [1, 4], threadsPerWarp = [4, 8], warpsPerCTA = [4, 1], order = [1, 0]}>
#mma = #ttg.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = []}>
#shared = #ttg.swizzled_shared<{vec = 8, perPhase = 2, maxPhase = 4, order = [1, 0]}>
#shared1 = #ttg.swizzled_shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [1, 0]}>
#smem = #ttg.shared_memory
module attributes {"ttg.num-ctas" = 1 : i32, "ttg.num-warps" = 4 : i32} {
tt.func @assume_matmul(%arg0: index, %arg1: index, %arg2: index, %arg3: !tt.ptr<f16>, %arg4: !tt.ptr<f16>) -> tensor<128x128xf32, #mma> {
%c-1 = arith.constant -1 : index
%c1 = arith.constant 1 : index
%c0 = arith.constant 0 : index
%c1_i32 = arith.constant 1 : i32
%c0_i32 = arith.constant 0 : i32
%true = arith.constant true
%cst = arith.constant dense<4.000000e+00> : tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #mma, kWidth = 2}>>
%cst_0 = arith.constant dense<4> : tensor<32x128xi32, #blocked>
%cst_1 = arith.constant dense<4> : tensor<128x32xi32, #blocked1>
%cst_2 = arith.constant dense<0.000000e+00> : tensor<128x128xf32, #mma>
%cst_3 = arith.constant dense<0.000000e+00> : tensor<32x128xf16, #blocked>
%0 = tt.splat %arg3 : !tt.ptr<f16> -> tensor<128x32x!tt.ptr<f16>, #blocked1>
%1 = tt.make_range {end = 32 : i32, start = 0 : i32} : tensor<32xi32, #ttg.slice<{dim = 0, parent = #blocked1}>>
%2 = tt.expand_dims %1 {axis = 0 : i32} : tensor<32xi32, #ttg.slice<{dim = 0, parent = #blocked1}>> -> tensor<1x32xi32, #blocked1>
%3 = tt.broadcast %2 : tensor<1x32xi32, #blocked1> -> tensor<128x32xi32, #blocked1>
%4 = tt.addptr %0, %3 : tensor<128x32x!tt.ptr<f16>, #blocked1>, tensor<128x32xi32, #blocked1>
%5 = tt.splat %arg4 : !tt.ptr<f16> -> tensor<32x128x!tt.ptr<f16>, #blocked>
%6 = tt.make_range {end = 128 : i32, start = 0 : i32} : tensor<128xi32, #ttg.slice<{dim = 0, parent = #blocked}>>
%7 = tt.expand_dims %6 {axis = 0 : i32} : tensor<128xi32, #ttg.slice<{dim = 0, parent = #blocked}>> -> tensor<1x128xi32, #blocked>
%8 = tt.broadcast %7 : tensor<1x128xi32, #blocked> -> tensor<32x128xi32, #blocked>
%9 = tt.addptr %5, %8 : tensor<32x128x!tt.ptr<f16>, #blocked>, tensor<32x128xi32, #blocked>
%10 = ttg.local_alloc : () -> !ttg.memdesc<1x128x32xf16, #shared, #smem, mutable>
%11 = ttg.local_alloc : () -> !ttg.memdesc<1x32x128xf16, #shared1, #smem, mutable>
%12 = arith.cmpi slt, %arg0, %arg1 : index
%13 = tt.splat %12 : i1 -> tensor<128x32xi1, #blocked1>
%14 = tt.load %4, %13 {OpIdx = #amdgpu.OpIdx<0>} : tensor<128x32x!tt.ptr<f16>, #blocked1>
%15 = tt.splat %12 : i1 -> tensor<32x128xi1, #blocked>
%16 = tt.load %9, %15, %cst_3 : tensor<32x128x!tt.ptr<f16>, #blocked>
%17 = ttg.memdesc_subview %10[%c0_i32, %c0_i32, %c0_i32] : !ttg.memdesc<1x128x32xf16, #shared, #smem, mutable> -> !ttg.memdesc<128x32xf16, #shared, #smem, mutable>
ttg.local_store %14, %17 {OpIdx = #amdgpu.OpIdx<0>} : tensor<128x32xf16, #blocked1> -> !ttg.memdesc<128x32xf16, #shared, #smem, mutable>
%18 = ttg.memdesc_subview %11[%c0_i32, %c0_i32, %c0_i32] : !ttg.memdesc<1x32x128xf16, #shared1, #smem, mutable> -> !ttg.memdesc<32x128xf16, #shared1, #smem, mutable>
ttg.local_store %16, %18 : tensor<32x128xf16, #blocked> -> !ttg.memdesc<32x128xf16, #shared1, #smem, mutable>
%19 = arith.subi %arg1, %arg2 : index
%20:6 = scf.for %arg5 = %arg0 to %19 step %arg2 iter_args(%arg6 = %4, %arg7 = %9, %arg8 = %cst_2, %arg9 = %c0_i32, %arg10 = %17, %arg11 = %18) -> (tensor<128x32x!tt.ptr<f16>, #blocked1>, tensor<32x128x!tt.ptr<f16>, #blocked>, tensor<128x128xf32, #mma>, i32, !ttg.memdesc<128x32xf16, #shared, #smem, mutable>, !ttg.memdesc<32x128xf16, #shared1, #smem, mutable>) {
%33 = tt.addptr %arg6, %cst_1 : tensor<128x32x!tt.ptr<f16>, #blocked1>, tensor<128x32xi32, #blocked1>
%34 = tt.addptr %arg7, %cst_0 : tensor<32x128x!tt.ptr<f16>, #blocked>, tensor<32x128xi32, #blocked>
llvm.intr.assume %true : i1
%35 = tt.load %33 {OpIdx = #amdgpu.OpIdx<0>} : tensor<128x32x!tt.ptr<f16>, #blocked1>
%36 = ttg.local_load %arg10 : !ttg.memdesc<128x32xf16, #shared, #smem, mutable> -> tensor<128x32xf16, #ttg.dot_op<{opIdx = 0, parent = #mma, kWidth = 2}>>
%37 = tt.load %34 : tensor<32x128x!tt.ptr<f16>, #blocked>
%38 = ttg.local_load %arg11 : !ttg.memdesc<32x128xf16, #shared1, #smem, mutable> -> tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #mma, kWidth = 2}>>
%39 = arith.mulf %38, %cst : tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #mma, kWidth = 2}>>
%40 = tt.dot %36, %39, %arg8 : tensor<128x32xf16, #ttg.dot_op<{opIdx = 0, parent = #mma, kWidth = 2}>> * tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #mma, kWidth = 2}>> -> tensor<128x128xf32, #mma>
%41 = arith.addi %arg9, %c1_i32 : i32
%42 = arith.cmpi slt, %41, %c1_i32 : i32
%43 = arith.select %42, %41, %c0_i32 : i32
%44 = ttg.memdesc_subview %10[%43, %c0_i32, %c0_i32] : !ttg.memdesc<1x128x32xf16, #shared, #smem, mutable> -> !ttg.memdesc<128x32xf16, #shared, #smem, mutable>
ttg.local_store %35, %44 {OpIdx = #amdgpu.OpIdx<0>} : tensor<128x32xf16, #blocked1> -> !ttg.memdesc<128x32xf16, #shared, #smem, mutable>
%45 = ttg.memdesc_subview %11[%43, %c0_i32, %c0_i32] : !ttg.memdesc<1x32x128xf16, #shared1, #smem, mutable> -> !ttg.memdesc<32x128xf16, #shared1, #smem, mutable>
ttg.local_store %37, %45 : tensor<32x128xf16, #blocked> -> !ttg.memdesc<32x128xf16, #shared1, #smem, mutable>
scf.yield %33, %34, %40, %43, %44, %45 : tensor<128x32x!tt.ptr<f16>, #blocked1>, tensor<32x128x!tt.ptr<f16>, #blocked>, tensor<128x128xf32, #mma>, i32, !ttg.memdesc<128x32xf16, #shared, #smem, mutable>, !ttg.memdesc<32x128xf16, #shared1, #smem, mutable>
}
%21 = arith.cmpi slt, %arg2, %c0 : index
%22 = arith.select %21, %c1, %c-1 : index
%23 = arith.subi %arg1, %arg0 : index
%24 = arith.addi %23, %arg2 : index
%25 = arith.addi %24, %22 : index
%26 = arith.divsi %25, %arg2 : index
%28 = ttg.local_load %20#4 : !ttg.memdesc<128x32xf16, #shared, #smem, mutable> -> tensor<128x32xf16, #ttg.dot_op<{opIdx = 0, parent = #mma, kWidth = 2}>>
%29 = ttg.local_load %20#5 : !ttg.memdesc<32x128xf16, #shared1, #smem, mutable> -> tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #mma, kWidth = 2}>>
%30 = arith.mulf %29, %cst : tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #mma, kWidth = 2}>>
%27 = arith.cmpi sge, %26, %c1 : index
llvm.intr.assume %27 : i1
%31 = scf.if %27 -> (tensor<128x128xf32, #mma>) {
%33 = tt.dot %28, %30, %20#2 : tensor<128x32xf16, #ttg.dot_op<{opIdx = 0, parent = #mma, kWidth = 2}>> * tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #mma, kWidth = 2}>> -> tensor<128x128xf32, #mma>
scf.yield %33 : tensor<128x128xf32, #mma>
} else {
scf.yield %20#2 : tensor<128x128xf32, #mma>
}
%32 = arith.select %27, %31, %20#2 : tensor<128x128xf32, #mma>
ttg.local_dealloc %10 : !ttg.memdesc<1x128x32xf16, #shared, #smem, mutable>
ttg.local_dealloc %11 : !ttg.memdesc<1x32x128xf16, #shared1, #smem, mutable>
tt.return %32 : tensor<128x128xf32, #mma>
}
}

// CHECK: #[[$ATTR_2:.+]] = #ttg.nvidia_mma<{versionMajor = 2, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = []}>
// CHECK: #[[$ATTR_3:.+]] = #ttg.swizzled_shared<{vec = 8, perPhase = 2, maxPhase = 4, order = [1, 0]}>
// CHECK: #[[$ATTR_4:.+]] = #ttg.swizzled_shared<{vec = 8, perPhase = 1, maxPhase = 8, order = [1, 0]}>
// CHECK: #[[$ATTR_5:.+]] = #ttg.shared_memory

// CHECK-LABEL: tt.func @assume_matmul(
// CHECK: %[[VAL_7:.*]] = arith.constant true
// CHECK: %[[VAL_8:.*]] = arith.constant dense<4.000000e+00> : tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #[[$ATTR_2]], kWidth = 2}>>
// CHECK: %[[VAL_23:.*]] = ttg.local_alloc : () -> !ttg.memdesc<1x128x32xf16, #[[$ATTR_3]], #[[$ATTR_5]], mutable>
// CHECK: %[[VAL_24:.*]] = ttg.local_alloc : () -> !ttg.memdesc<1x32x128xf16, #[[$ATTR_4]], #[[$ATTR_5]], mutable>
// CHECK: %[[VAL_33:.*]]:6 = scf.for
// CHECK: scf.yield
// CHECK: }
// CHECK-NEXT: %[[VAL_54:.*]] = ttg.local_load %[[VAL_55:.*]]#4 : !ttg.memdesc<128x32xf16, #[[$ATTR_3]], #[[$ATTR_5]], mutable> -> tensor<128x32xf16, #ttg.dot_op<{opIdx = 0, parent = #[[$ATTR_2]], kWidth = 2}>>
// CHECK-NEXT: %[[VAL_56:.*]] = ttg.local_load %[[VAL_55]]#5 : !ttg.memdesc<32x128xf16, #[[$ATTR_4]], #[[$ATTR_5]], mutable> -> tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #[[$ATTR_2]], kWidth = 2}>>
// CHECK-NEXT: %[[VAL_57:.*]] = arith.mulf %[[VAL_56]], %[[VAL_8]] : tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #[[$ATTR_2]], kWidth = 2}>>
// CHECK-NEXT: llvm.intr.assume %[[VAL_7]] : i1
// CHECK-NEXT: %[[VAL_58:.*]] = tt.dot %[[VAL_54]], %[[VAL_57]], %[[VAL_55]]#2 : tensor<128x32xf16, #ttg.dot_op<{opIdx = 0, parent = #[[$ATTR_2]], kWidth = 2}>> * tensor<32x128xf16, #ttg.dot_op<{opIdx = 1, parent = #[[$ATTR_2]], kWidth = 2}>> -> tensor<128x128xf32, #[[$ATTR_2]]>
// CHECK-NEXT: ttg.local_dealloc %[[VAL_23]] : !ttg.memdesc<1x128x32xf16, #[[$ATTR_3]], #[[$ATTR_5]], mutable>
// CHECK-NEXT: ttg.local_dealloc %[[VAL_24]] : !ttg.memdesc<1x32x128xf16, #[[$ATTR_4]], #[[$ATTR_5]], mutable>
// CHECK-NEXT: tt.return %[[VAL_58]] : tensor<128x128xf32, #[[$ATTR_2]]>
// CHECK-NEXT: }
Loading
Loading