Skip to content

[mlir][gpu]add AffineScope Trait to gpu.launch. #121058

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 5 commits into from
Feb 4, 2025
Merged

[mlir][gpu]add AffineScope Trait to gpu.launch. #121058

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
ca23030
add AffineScope Trait to gpu.launch.
linuxlonelyeagle Dec 24, 2024
d1d55d8
rebase main.
linuxlonelyeagle Jan 18, 2025
989e3d6
rebase main.
linuxlonelyeagle Jan 18, 2025
12e2e4a
update test and update op define,.
linuxlonelyeagle Feb 4, 2025
c5327cb
Merge branch 'main' into add-affinescope-to-gpulaunch
linuxlonelyeagle Feb 4, 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: 1 addition & 1 deletion mlir/include/mlir/Dialect/GPU/IR/GPUOps.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -796,7 +796,7 @@ def GPU_LaunchFuncOp :GPU_Op<"launch_func", [
def GPU_LaunchOp : GPU_Op<"launch", [
AutomaticAllocationScope, AttrSizedOperandSegments, GPU_AsyncOpInterface,
DeclareOpInterfaceMethods<InferIntRangeInterface, ["inferResultRanges"]>,
RecursiveMemoryEffects]>,
RecursiveMemoryEffects, AffineScope]>,
Arguments<(ins Variadic<GPU_AsyncToken>:$asyncDependencies,
Index:$gridSizeX, Index:$gridSizeY, Index:$gridSizeZ,
Index:$blockSizeX, Index:$blockSizeY, Index:$blockSizeZ,
Expand Down
43 changes: 36 additions & 7 deletions mlir/test/Dialect/Affine/ops.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -317,13 +317,42 @@ module attributes {gpu.container_module} {
}
}
}
// CHECK-SAME: (%[[VAL_0:.*]]: memref<?x?xf32>) kernel {
// CHECK: %[[VAL_1:.*]] = arith.constant 1 : index
// CHECK: %[[VAL_2:.*]] = memref.dim %[[VAL_0]], %[[VAL_1]] : memref<?x?xf32>
// CHECK: %[[VAL_3:.*]] = arith.constant 0 : index
// CHECK: affine.for %[[VAL_4:.*]] = %[[VAL_3]] to %[[VAL_2]] step 32 {
// CHECK: }
// CHECK: gpu.return
// CHECK-SAME: (%[[VAL_0:.*]]: memref<?x?xf32>) kernel {
// CHECK: %[[VAL_1:.*]] = arith.constant 1 : index
// CHECK: %[[VAL_2:.*]] = memref.dim %[[VAL_0]], %[[VAL_1]] : memref<?x?xf32>
// CHECK: %[[VAL_3:.*]] = arith.constant 0 : index
// CHECK: affine.for %{{.*}} = %[[VAL_3]] to %[[VAL_2]] step 32 {
// CHECK: }
// CHECK: gpu.return

// -----

// CHECK-LABEL: @gpu_launch_affine

module {
func.func @gpu_launch_affine() {
%c1 = arith.constant 1 : index
gpu.launch blocks(%arg0, %arg1, %arg2) in (%arg6 = %c1, %arg7 = %c1, %arg8 = %c1)
threads(%arg3, %arg4, %arg5) in (%arg9 = %c1, %arg10 = %c1, %arg11 = %c1) {
%thread_id_x = gpu.thread_id x
%c128 = arith.constant 128 : index
affine.for %arg12 = %thread_id_x to %c128 step 8 {
}
gpu.terminator
}
return
}
}

// CHECK-NEXT: %[[VAL_0:.*]] = arith.constant 1 : index
// CHECK-NEXT: gpu.launch blocks(%[[VAL_1:.*]], %[[VAL_2:.*]], %[[VAL_3:.*]]) in (%[[VAL_4:.*]] = %[[VAL_0]], %[[VAL_5:.*]] = %[[VAL_0]], %[[VAL_6:.*]] = %[[VAL_0]]) threads(%[[VAL_7:.*]], %[[VAL_8:.*]], %[[VAL_9:.*]]) in (%[[VAL_10:.*]] = %[[VAL_0]], %[[VAL_11:.*]] = %[[VAL_0]], %[[VAL_12:.*]] = %[[VAL_0]]) {
// CHECK-NEXT: %[[VAL_13:.*]] = gpu.thread_id x
// CHECK-NEXT: %[[VAL_14:.*]] = arith.constant 128 : index
// CHECK-NEXT: affine.for %{{.*}} = %[[VAL_13]] to %[[VAL_14]] step 8 {
// CHECK-NEXT: }
// CHECK-NEXT: gpu.terminator
// CHECK-NEXT: }
// CHECK-NEXT: return

// -----

Expand Down
4 changes: 2 additions & 2 deletions mlir/test/Dialect/GPU/transform-gpu.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,7 +43,7 @@ module attributes {transform.with_named_sequence} {
!type = memref<2 x 32 x f32>
!type1d = memref<32 x f32>

// CHECK-DAG: #[[$MAP:.*]] = affine_map<()[s0] -> (s0 floordiv 128)>
// CHECK-DAG: #[[$MAP:.*]] = affine_map<()[s0] -> (s0 floordiv 128)>

// CHECK-LABEL: func.func @warpgroup_3d(
// CHECK-SAME: %[[ARGX:[0-9a-z]+]]: memref<2x32xf32>
Expand Down Expand Up @@ -647,7 +647,7 @@ module attributes {transform.with_named_sequence} {
#map = affine_map<(d0) -> (d0 * 128)>
#map1 = affine_map<(d0) -> (d0 * 32)>

// CHECK-DAG: #[[$MAPB:.*]] = affine_map<()[s0] -> (s0 * 128)>
// CHECK-DAG: #[[$MAPB:.*]] = affine_map<()[s0] -> (s0 * 128)>
// CHECK-DAG: #[[$MAPW:.*]] = affine_map<()[s0, s1, s2] -> (s2 * 32 + ((s0 + s1 * 4) floordiv 32) * 32)>

// CHECK-LABEL: func.func @simple_fill(
Expand Down
Loading