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
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
|
// 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
// RUN: %{compile} | %{run}
//
// Do the same run, but now with direct IR generation and vectorization.
// REDEFINE: %{option} = "enable-runtime-library=false 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 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}
#ST = #sparse_tensor.encoding<{lvlTypes = ["compressed", "compressed", "compressed"]}>
//
// Trait for 3-d tensor element wise multiplication.
//
#trait_mul = {
indexing_maps = [
affine_map<(i,j,k) -> (i,j,k)>, // A (in)
affine_map<(i,j,k) -> (i,j,k)>, // B (in)
affine_map<(i,j,k) -> (i,j,k)> // X (out)
],
iterator_types = ["parallel", "parallel", "parallel"],
doc = "X(i,j,k) = A(i,j,k) * B(i,j,k)"
}
module {
// Multiplies two 3-d sparse tensors element-wise into a new sparse tensor.
func.func @tensor_mul(%arga: tensor<?x?x?xf64, #ST>,
%argb: tensor<?x?x?xf64, #ST>) -> tensor<?x?x?xf64, #ST> {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
%c2 = arith.constant 2 : index
%d0 = tensor.dim %arga, %c0 : tensor<?x?x?xf64, #ST>
%d1 = tensor.dim %arga, %c1 : tensor<?x?x?xf64, #ST>
%d2 = tensor.dim %arga, %c2 : tensor<?x?x?xf64, #ST>
%xt = bufferization.alloc_tensor(%d0, %d1, %d2) : tensor<?x?x?xf64, #ST>
%0 = linalg.generic #trait_mul
ins(%arga, %argb: tensor<?x?x?xf64, #ST>, tensor<?x?x?xf64, #ST>)
outs(%xt: tensor<?x?x?xf64, #ST>) {
^bb(%a: f64, %b: f64, %x: f64):
%1 = arith.mulf %a, %b : f64
linalg.yield %1 : f64
} -> tensor<?x?x?xf64, #ST>
return %0 : tensor<?x?x?xf64, #ST>
}
// Driver method to call and verify tensor multiplication kernel.
func.func @entry() {
%c0 = arith.constant 0 : index
%default_val = arith.constant -1.0 : f64
// Setup sparse tensor A
%ta = arith.constant dense<
[ [ [1.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 0.0, 0.0, 0.0, 0.0 ],
[1.2, 0.0, 3.5, 0.0, 0.0 ] ],
[ [0.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 0.0, 0.0, 0.0, 0.0 ] ],
[ [2.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 0.0, 4.0, 0.0, 0.0 ]] ]> : tensor<3x3x5xf64>
// Setup sparse tensor B
%tb = arith.constant dense<
[ [ [0.0, 0.0, 0.0, 0.0, 4.0 ],
[0.0, 0.0, 0.0, 0.0, 0.0 ],
[2.0, 0.0, 1.0, 0.0, 0.0 ] ],
[ [0.0, 0.0, 0.0, 0.0, 9.0 ],
[0.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 7.0, 0.0, 0.0, 0.0 ] ],
[ [1.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 0.0, 0.0, 0.0, 0.0 ],
[0.0, 0.0, 2.0, 0.0, 0.0 ]] ]> : tensor<3x3x5xf64>
%sta = sparse_tensor.convert %ta : tensor<3x3x5xf64> to tensor<?x?x?xf64, #ST>
%stb = sparse_tensor.convert %tb : tensor<3x3x5xf64> to tensor<?x?x?xf64, #ST>
// Call sparse tensor multiplication kernel.
%0 = call @tensor_mul(%sta, %stb)
: (tensor<?x?x?xf64, #ST>, tensor<?x?x?xf64, #ST>) -> tensor<?x?x?xf64, #ST>
// Verify results
//
// CHECK: 4
// CHECK-NEXT: ( 2.4, 3.5, 2, 8 )
// CHECK-NEXT: ( ( ( 0, 0, 0, 0, 0 ), ( 0, 0, 0, 0, 0 ), ( 2.4, 0, 3.5, 0, 0 ) ),
// CHECK-SAME: ( ( 0, 0, 0, 0, 0 ), ( 0, 0, 0, 0, 0 ), ( 0, 0, 0, 0, 0 ) ),
// CHECK-SAME: ( ( 2, 0, 0, 0, 0 ), ( 0, 0, 0, 0, 0 ), ( 0, 0, 8, 0, 0 ) ) )
//
%n = sparse_tensor.number_of_entries %0 : tensor<?x?x?xf64, #ST>
vector.print %n : index
%m1 = sparse_tensor.values %0 : tensor<?x?x?xf64, #ST> to memref<?xf64>
%v1 = vector.transfer_read %m1[%c0], %default_val: memref<?xf64>, vector<4xf64>
vector.print %v1 : vector<4xf64>
// Print %0 in dense form.
%dt = sparse_tensor.convert %0 : tensor<?x?x?xf64, #ST> to tensor<?x?x?xf64>
%v2 = vector.transfer_read %dt[%c0, %c0, %c0], %default_val: tensor<?x?x?xf64>, vector<3x3x5xf64>
vector.print %v2 : vector<3x3x5xf64>
// Release the resources.
bufferization.dealloc_tensor %sta : tensor<?x?x?xf64, #ST>
bufferization.dealloc_tensor %stb : tensor<?x?x?xf64, #ST>
bufferization.dealloc_tensor %0 : tensor<?x?x?xf64, #ST>
return
}
}
|
