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
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
|
// DEFINE: %{option} = enable-runtime-library=true
// DEFINE: %{compile} = mlir-opt %s --sparse-compiler=%{option}
// DEFINE: %{run} = mlir-cpu-runner \
// DEFINE: -e entry -entry-point-result=void \
// DEFINE: -shared-libs=%mlir_c_runner_utils | \
// DEFINE: FileCheck %s
//
// RUN: %{compile} | %{run}
//
// Do the same run, but now with direct IR generation.
// REDEFINE: %{option} = "enable-runtime-library=false enable-buffer-initialization=true"
// RUN: %{compile} | %{run}
//
// Do the same run, but now with direct IR generation and vectorization.
// REDEFINE: %{option} = "enable-runtime-library=false enable-buffer-initialization=true vl=2 reassociate-fp-reductions=true enable-index-optimizations=true"
// RUN: %{compile} | %{run}
// Do the same run, but now with direct IR generation and, if available, VLA
// vectorization.
// REDEFINE: %{option} = "enable-runtime-library=false enable-buffer-initialization=true vl=4 enable-arm-sve=%ENABLE_VLA"
// REDEFINE: %{run} = %lli_host_or_aarch64_cmd \
// REDEFINE: --entry-function=entry_lli \
// REDEFINE: --extra-module=%S/Inputs/main_for_lli.ll \
// REDEFINE: %VLA_ARCH_ATTR_OPTIONS \
// REDEFINE: --dlopen=%mlir_native_utils_lib_dir/libmlir_c_runner_utils%shlibext | \
// REDEFINE: FileCheck %s
// RUN: %{compile} | mlir-translate -mlir-to-llvmir | %{run}
#DCSR = #sparse_tensor.encoding<{
dimLevelType = [ "compressed", "compressed" ]
}>
#trait_mult_elt = {
indexing_maps = [
affine_map<(i,j) -> (i,j)>, // A
affine_map<(i,j) -> (i,j)>, // B
affine_map<(i,j) -> (i,j)> // X (out)
],
iterator_types = ["parallel", "parallel"],
doc = "X(i,j) = A(i,j) * B(i,j)"
}
module {
// Sparse kernel.
func.func @sparse_mult_elt(
%arga: tensor<32x16xf32, #DCSR>, %argb: tensor<32x16xf32, #DCSR>) -> tensor<32x16xf32, #DCSR> {
%argx = bufferization.alloc_tensor() : tensor<32x16xf32, #DCSR>
%0 = linalg.generic #trait_mult_elt
ins(%arga, %argb: tensor<32x16xf32, #DCSR>, tensor<32x16xf32, #DCSR>)
outs(%argx: tensor<32x16xf32, #DCSR>) {
^bb(%a: f32, %b: f32, %x: f32):
%1 = arith.mulf %a, %b : f32
linalg.yield %1 : f32
} -> tensor<32x16xf32, #DCSR>
return %0 : tensor<32x16xf32, #DCSR>
}
// Driver method to call and verify kernel.
func.func @entry() {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
// Setup very sparse matrices.
%ta = arith.constant sparse<
[ [2,2], [15,15], [31,0], [31,14] ], [ 2.0, 3.0, -2.0, 4.0 ]
> : tensor<32x16xf32>
%tb = arith.constant sparse<
[ [1,1], [2,0], [2,2], [2,15], [31,0], [31,15] ], [ 5.0, 6.0, 7.0, 8.0, -10.0, 9.0 ]
> : tensor<32x16xf32>
%sta = sparse_tensor.convert %ta
: tensor<32x16xf32> to tensor<32x16xf32, #DCSR>
%stb = sparse_tensor.convert %tb
: tensor<32x16xf32> to tensor<32x16xf32, #DCSR>
// Call kernel.
%0 = call @sparse_mult_elt(%sta, %stb)
: (tensor<32x16xf32, #DCSR>,
tensor<32x16xf32, #DCSR>) -> tensor<32x16xf32, #DCSR>
//
// Verify results. Only two entries stored in result!
//
// CHECK: ( 14, 20, 0, 0 )
//
%val = sparse_tensor.values %0 : tensor<32x16xf32, #DCSR> to memref<?xf32>
%vv = vector.transfer_read %val[%c0], %f0: memref<?xf32>, vector<4xf32>
vector.print %vv : vector<4xf32>
// Release the resources.
bufferization.dealloc_tensor %sta : tensor<32x16xf32, #DCSR>
bufferization.dealloc_tensor %stb : tensor<32x16xf32, #DCSR>
bufferization.dealloc_tensor %0 : tensor<32x16xf32, #DCSR>
return
}
}
|
