[Mlir-commits] [mlir] 750799b - [mlir][NFC] Don't outline kernel in MMA integration tests
llvmlistbot at llvm.org
llvmlistbot at llvm.org
Thu May 27 09:45:02 PDT 2021
Author: thomasraoux
Date: 2021-05-27T09:43:54-07:00
New Revision: 750799b7bc3faeda0d4a14e556ce788e0452152e
URL: https://github.com/llvm/llvm-project/commit/750799b7bc3faeda0d4a14e556ce788e0452152e
DIFF: https://github.com/llvm/llvm-project/commit/750799b7bc3faeda0d4a14e556ce788e0452152e.diff
LOG: [mlir][NFC] Don't outline kernel in MMA integration tests
This matches better how other gpu integration tests are done.
Differential Revision: https://reviews.llvm.org/D103099
Added:
Modified:
mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f16.mlir
mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f32.mlir
Removed:
################################################################################
diff --git a/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f16.mlir b/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f16.mlir
index 3dc911b32b69b..8c067c4502caa 100644
--- a/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f16.mlir
+++ b/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f16.mlir
@@ -10,85 +10,77 @@
// Test case to check the working of Tensor cores on Nvidia GPUs. The kernel has already
// been outlined to prevent crashing due to introduction of an empty basic block by --gpu-
// kernel-outling.
-module attributes {gpu.container_module} {
- func @main() {
- %0 = memref.alloc() : memref<16x16xf16>
- %22 = memref.alloc() : memref<16x16xf16>
- %1 = memref.alloc() : memref<16x16xf32>
+func @main() {
+ %0 = memref.alloc() : memref<16x16xf16>
+ %22 = memref.alloc() : memref<16x16xf16>
+ %1 = memref.alloc() : memref<16x16xf32>
- %f1 = constant 1.0e+00 : f16
- %f0 = constant 0.0e+00 : f16
- %c0 = constant 0 : index
- %c16 = constant 16 : index
- %c32 = constant 32 : index
- %c1 = constant 1 : index
+ %f1 = constant 1.0e+00 : f16
+ %f0 = constant 0.0e+00 : f16
+ %c0 = constant 0 : index
+ %c16 = constant 16 : index
+ %c32 = constant 32 : index
+ %c1 = constant 1 : index
- // Intialize the Input matrix with ones.
- scf.for %arg0 = %c0 to %c16 step %c1 {
- scf.for %arg1 = %c0 to %c16 step %c1 {
- memref.store %f1, %0[%arg0, %arg1] : memref<16x16xf16>
- }
+ // Intialize the Input matrix with ones.
+ scf.for %arg0 = %c0 to %c16 step %c1 {
+ scf.for %arg1 = %c0 to %c16 step %c1 {
+ memref.store %f1, %0[%arg0, %arg1] : memref<16x16xf16>
}
- // Intialize the accumulator matrix with zeros.
- scf.for %arg0 = %c0 to %c16 step %c1 {
- scf.for %arg1 = %c0 to %c16 step %c1 {
- memref.store %f0, %22[%arg0, %arg1] : memref<16x16xf16>
- }
- }
-
- %2 = memref.cast %0 : memref<16x16xf16> to memref<*xf16>
- %33 = memref.cast %22 : memref<16x16xf16> to memref<*xf16>
- %3 = memref.cast %1 : memref<16x16xf32> to memref<*xf32>
- gpu.host_register %2 : memref<*xf16>
- gpu.host_register %33 : memref<*xf16>
-
- gpu.launch_func @main_kernel::@main_kernel blocks in (%c1, %c1, %c1) threads in (%c32, %c1, %c1) args(%0 : memref<16x16xf16>, %22 : memref<16x16xf16>)
-
- // Convert the results from f16 to f32 for printing.
- scf.for %arg0 = %c0 to %c16 step %c1 {
- scf.for %arg1 = %c0 to %c16 step %c1 {
- %6 = memref.load %0[%arg0, %arg1] : memref<16x16xf16>
- %7 = fpext %6 : f16 to f32
- memref.store %7, %1[%arg0, %arg1] : memref<16x16xf32>
- }
+ }
+ // Intialize the accumulator matrix with zeros.
+ scf.for %arg0 = %c0 to %c16 step %c1 {
+ scf.for %arg1 = %c0 to %c16 step %c1 {
+ memref.store %f0, %22[%arg0, %arg1] : memref<16x16xf16>
}
-
- // Print the memref after computation.
- call @print_memref_f32(%3) : (memref<*xf32>) -> ()
- // CHECK: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16]
- return
}
- gpu.module @main_kernel {
- gpu.func @main_kernel(%arg0: memref<16x16xf16>, %arg22 : memref<16x16xf16>) kernel {
- %c0 = constant 0 : index
+ %2 = memref.cast %0 : memref<16x16xf16> to memref<*xf16>
+ %33 = memref.cast %22 : memref<16x16xf16> to memref<*xf16>
+ %3 = memref.cast %1 : memref<16x16xf32> to memref<*xf32>
+ gpu.host_register %2 : memref<*xf16>
+ gpu.host_register %33 : memref<*xf16>
- %0 = gpu.subgroup_mma_load_matrix %arg0[%c0, %c0] {operand = "AOp", leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "AOp">
- %1 = gpu.subgroup_mma_load_matrix %arg0[%c0, %c0] {operand = "BOp", leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "BOp">
- %2 = gpu.subgroup_mma_load_matrix %arg22[%c0, %c0] {operand = "COp", leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "COp">
+ gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c1, %grid_y = %c1, %grid_z = %c1)
+ threads(%tx, %ty, %tz) in (%block_x = %c32, %block_y = %c1, %block_z = %c1) {
+ %A = gpu.subgroup_mma_load_matrix %0[%c0, %c0] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "AOp">
+ %B = gpu.subgroup_mma_load_matrix %0[%c0, %c0] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "BOp">
+ %C = gpu.subgroup_mma_load_matrix %22[%c0, %c0] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "COp">
- %3 = gpu.subgroup_mma_compute %0, %1, %2 : !gpu.mma_matrix<16x16xf16, "AOp">, !gpu.mma_matrix<16x16xf16, "BOp"> -> !gpu.mma_matrix<16x16xf16, "COp">
+ %R = gpu.subgroup_mma_compute %A, %B, %C : !gpu.mma_matrix<16x16xf16, "AOp">, !gpu.mma_matrix<16x16xf16, "BOp"> -> !gpu.mma_matrix<16x16xf16, "COp">
- gpu.subgroup_mma_store_matrix %3, %arg0[%c0, %c0] {leadDimension = 16 : index}: !gpu.mma_matrix<16x16xf16, "COp">, memref<16x16xf16>
+ gpu.subgroup_mma_store_matrix %R, %0[%c0, %c0] {leadDimension = 16 : index}: !gpu.mma_matrix<16x16xf16, "COp">, memref<16x16xf16>
+ gpu.terminator
+ }
- gpu.return
+ // Convert the results from f16 to f32 for printing.
+ scf.for %arg0 = %c0 to %c16 step %c1 {
+ scf.for %arg1 = %c0 to %c16 step %c1 {
+ %6 = memref.load %0[%arg0, %arg1] : memref<16x16xf16>
+ %7 = fpext %6 : f16 to f32
+ memref.store %7, %1[%arg0, %arg1] : memref<16x16xf32>
}
}
- func private @print_memref_f32(memref<*xf32>)
+ // Print the memref after computation.
+ call @print_memref_f32(%3) : (memref<*xf32>) -> ()
+ // CHECK: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16]
+ return
}
+
+func private @print_memref_f32(memref<*xf32>)
diff --git a/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f32.mlir b/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f32.mlir
index ba948ea8997b1..478c8662c3808 100644
--- a/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f32.mlir
+++ b/mlir/test/Integration/GPU/CUDA/TensorCore/wmma-matmul-f32.mlir
@@ -7,79 +7,68 @@
// RUN: --shared-libs=%linalg_test_lib_dir/libmlir_runner_utils%shlibext \
// RUN: --entry-point-result=void \
// RUN: | FileCheck %s
-// Test case to check the working of Tensor cores on Nvidia GPUs. The kernel has already
-// been outlined to prevent crashing due to introduction of an empty basic block by --gpu-
-// kernel-outling.
-module attributes {gpu.container_module} {
- func @main() {
- %0 = memref.alloc() : memref<16x16xf16>
- %22 = memref.alloc() : memref<16x16xf32>
- %1 = memref.alloc() : memref<16x16xf32>
- %f1 = constant 1.0e+00 : f16
- %f0 = constant 0.0e+00 : f32
- %c0 = constant 0 : index
- %c16 = constant 16 : index
- %c32 = constant 32 : index
- %c1 = constant 1 : index
+func @main() {
+ %0 = memref.alloc() : memref<16x16xf16>
+ %22 = memref.alloc() : memref<16x16xf32>
+ %1 = memref.alloc() : memref<16x16xf32>
- // Intialize the Input matrix with ones.
- scf.for %arg0 = %c0 to %c16 step %c1 {
- scf.for %arg1 = %c0 to %c16 step %c1 {
- memref.store %f1, %0[%arg0, %arg1] : memref<16x16xf16>
- }
+ %f1 = constant 1.0e+00 : f16
+ %f0 = constant 0.0e+00 : f32
+ %c0 = constant 0 : index
+ %c16 = constant 16 : index
+ %c32 = constant 32 : index
+ %c1 = constant 1 : index
+
+ // Intialize the Input matrix with ones.
+ scf.for %arg0 = %c0 to %c16 step %c1 {
+ scf.for %arg1 = %c0 to %c16 step %c1 {
+ memref.store %f1, %0[%arg0, %arg1] : memref<16x16xf16>
}
- // Intialize the accumulator matrix with zeros.
- scf.for %arg0 = %c0 to %c16 step %c1 {
- scf.for %arg1 = %c0 to %c16 step %c1 {
- memref.store %f0, %22[%arg0, %arg1] : memref<16x16xf32>
- }
+ }
+ // Intialize the accumulator matrix with zeros.
+ scf.for %arg0 = %c0 to %c16 step %c1 {
+ scf.for %arg1 = %c0 to %c16 step %c1 {
+ memref.store %f0, %22[%arg0, %arg1] : memref<16x16xf32>
}
-
- %2 = memref.cast %0 : memref<16x16xf16> to memref<*xf16>
- %33 = memref.cast %22 : memref<16x16xf32> to memref<*xf32>
- %3 = memref.cast %1 : memref<16x16xf32> to memref<*xf32>
- gpu.host_register %2 : memref<*xf16>
- gpu.host_register %33 : memref<*xf32>
-
- gpu.launch_func @main_kernel::@main_kernel blocks in (%c1, %c1, %c1) threads in (%c32, %c1, %c1) args(%0 : memref<16x16xf16>, %22 : memref<16x16xf32>)
-
- // Print the memref after computation.
- call @print_memref_f32(%33) : (memref<*xf32>) -> ()
- // CHECK: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
- // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16]
- return
}
- gpu.module @main_kernel {
- gpu.func @main_kernel(%arg0: memref<16x16xf16>, %arg22 : memref<16x16xf32>) kernel {
- %c0 = constant 0 : index
+ %2 = memref.cast %0 : memref<16x16xf16> to memref<*xf16>
+ %33 = memref.cast %22 : memref<16x16xf32> to memref<*xf32>
+ %3 = memref.cast %1 : memref<16x16xf32> to memref<*xf32>
+ gpu.host_register %2 : memref<*xf16>
+ gpu.host_register %33 : memref<*xf32>
- %0 = gpu.subgroup_mma_load_matrix %arg0[%c0, %c0] {operand = "AOp", leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "AOp">
- %1 = gpu.subgroup_mma_load_matrix %arg0[%c0, %c0] {operand = "BOp", leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "BOp">
- %2 = gpu.subgroup_mma_load_matrix %arg22[%c0, %c0] {operand = "COp", leadDimension = 16 : index} : memref<16x16xf32> -> !gpu.mma_matrix<16x16xf32, "COp">
+ gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c1, %grid_y = %c1, %grid_z = %c1)
+ threads(%tx, %ty, %tz) in (%block_x = %c32, %block_y = %c1, %block_z = %c1) {
+ %A = gpu.subgroup_mma_load_matrix %0[%c0, %c0] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "AOp">
+ %B = gpu.subgroup_mma_load_matrix %0[%c0, %c0] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "BOp">
+ %C = gpu.subgroup_mma_load_matrix %22[%c0, %c0] {leadDimension = 16 : index} : memref<16x16xf32> -> !gpu.mma_matrix<16x16xf32, "COp">
- %3 = gpu.subgroup_mma_compute %0, %1, %2 : !gpu.mma_matrix<16x16xf16, "AOp">, !gpu.mma_matrix<16x16xf16, "BOp"> -> !gpu.mma_matrix<16x16xf32, "COp">
+ %R = gpu.subgroup_mma_compute %A, %B, %C : !gpu.mma_matrix<16x16xf16, "AOp">, !gpu.mma_matrix<16x16xf16, "BOp"> -> !gpu.mma_matrix<16x16xf32, "COp">
- gpu.subgroup_mma_store_matrix %3, %arg22[%c0, %c0] {leadDimension = 16 : index}: !gpu.mma_matrix<16x16xf32, "COp">, memref<16x16xf32>
-
- gpu.return
- }
+ gpu.subgroup_mma_store_matrix %R, %22[%c0, %c0] {leadDimension = 16 : index}: !gpu.mma_matrix<16x16xf32, "COp">, memref<16x16xf32>
+ gpu.terminator
}
-
- func private @print_memref_f32(memref<*xf32>)
+ // Print the memref after computation.
+ call @print_memref_f32(%33) : (memref<*xf32>) -> ()
+ // CHECK: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16],
+ // CHECK-NEXT: [16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16]
+ return
}
+
+func private @print_memref_f32(memref<*xf32>)
More information about the Mlir-commits
mailing list