test-maskedstore.mlir
2.92 KB
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
// RUN: mlir-opt %s -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
// RUN: mlir-cpu-runner -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
func @maskedstore16(%base: memref<?xf32>,
%mask: vector<16xi1>, %value: vector<16xf32>) {
vector.maskedstore %base, %mask, %value
: vector<16xi1>, vector<16xf32> into memref<?xf32>
return
}
func @printmem16(%A: memref<?xf32>) {
%c0 = constant 0: index
%c1 = constant 1: index
%c16 = constant 16: index
%z = constant 0.0: f32
%m = vector.broadcast %z : f32 to vector<16xf32>
%mem = scf.for %i = %c0 to %c16 step %c1
iter_args(%m_iter = %m) -> (vector<16xf32>) {
%c = load %A[%i] : memref<?xf32>
%i32 = index_cast %i : index to i32
%m_new = vector.insertelement %c, %m_iter[%i32 : i32] : vector<16xf32>
scf.yield %m_new : vector<16xf32>
}
vector.print %mem : vector<16xf32>
return
}
func @entry() {
// Set up memory.
%f0 = constant 0.0: f32
%c0 = constant 0: index
%c1 = constant 1: index
%c16 = constant 16: index
%A = alloc(%c16) : memref<?xf32>
scf.for %i = %c0 to %c16 step %c1 {
store %f0, %A[%i] : memref<?xf32>
}
// Set up value vector.
%v = vector.broadcast %f0 : f32 to vector<16xf32>
%val = scf.for %i = %c0 to %c16 step %c1
iter_args(%v_iter = %v) -> (vector<16xf32>) {
%i32 = index_cast %i : index to i32
%fi = sitofp %i32 : i32 to f32
%v_new = vector.insertelement %fi, %v_iter[%i32 : i32] : vector<16xf32>
scf.yield %v_new : vector<16xf32>
}
// Set up masks.
%t = constant 1: i1
%none = vector.constant_mask [0] : vector<16xi1>
%some = vector.constant_mask [8] : vector<16xi1>
%more = vector.insert %t, %some[13] : i1 into vector<16xi1>
%all = vector.constant_mask [16] : vector<16xi1>
//
// Masked store tests.
//
vector.print %val : vector<16xf32>
// CHECK: ( 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 )
call @printmem16(%A) : (memref<?xf32>) -> ()
// CHECK: ( 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 )
call @maskedstore16(%A, %none, %val)
: (memref<?xf32>, vector<16xi1>, vector<16xf32>) -> ()
call @printmem16(%A) : (memref<?xf32>) -> ()
// CHECK: ( 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 )
call @maskedstore16(%A, %some, %val)
: (memref<?xf32>, vector<16xi1>, vector<16xf32>) -> ()
call @printmem16(%A) : (memref<?xf32>) -> ()
// CHECK: ( 0, 1, 2, 3, 4, 5, 6, 7, 0, 0, 0, 0, 0, 0, 0, 0 )
call @maskedstore16(%A, %more, %val)
: (memref<?xf32>, vector<16xi1>, vector<16xf32>) -> ()
call @printmem16(%A) : (memref<?xf32>) -> ()
// CHECK: ( 0, 1, 2, 3, 4, 5, 6, 7, 0, 0, 0, 0, 0, 13, 0, 0 )
call @maskedstore16(%A, %all, %val)
: (memref<?xf32>, vector<16xi1>, vector<16xf32>) -> ()
call @printmem16(%A) : (memref<?xf32>) -> ()
// CHECK: ( 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 )
return
}