1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
|
// RUN: mlir-opt %s \
// RUN: | mlir-opt -gpu-kernel-outlining \
// RUN: | mlir-opt -pass-pipeline='builtin.module(gpu.module(strip-debuginfo,convert-gpu-to-nvvm,gpu-to-cubin))' \
// RUN: | mlir-opt -gpu-to-llvm \
// RUN: | mlir-cpu-runner \
// RUN: --shared-libs=%mlir_cuda_runtime \
// RUN: --shared-libs=%mlir_runner_utils \
// RUN: --entry-point-result=void \
// RUN: | FileCheck %s
// CHECK: [{{(35, ){34}35}}]
func.func @main() {
%arg = memref.alloc() : memref<35xf32>
%dst = memref.cast %arg : memref<35xf32> to memref<?xf32>
%one = arith.constant 1 : index
%c0 = arith.constant 0 : index
%sx = memref.dim %dst, %c0 : memref<?xf32>
%cast_dst = memref.cast %dst : memref<?xf32> to memref<*xf32>
gpu.host_register %cast_dst : memref<*xf32>
gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %one, %grid_y = %one, %grid_z = %one)
threads(%tx, %ty, %tz) in (%block_x = %sx, %block_y = %one, %block_z = %one) {
%val = arith.index_cast %tx : index to i32
%xor = gpu.all_reduce %val uniform {
^bb(%lhs : i32, %rhs : i32):
%xor = arith.xori %lhs, %rhs : i32
"gpu.yield"(%xor) : (i32) -> ()
} : (i32) -> (i32)
%res = arith.sitofp %xor : i32 to f32
memref.store %res, %dst[%tx] : memref<?xf32>
gpu.terminator
}
call @printMemrefF32(%cast_dst) : (memref<*xf32>) -> ()
return
}
func.func private @printMemrefF32(memref<*xf32>)
|
