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
|
// RUN: mlir-opt %s -split-input-file | FileCheck %s
// -----
// Test with loop IVs.
func.func @test0(%arg0 : index, %arg1 : index) {
%0 = memref.alloc() : memref<100x100xf32>
%1 = memref.alloc() : memref<100x100xf32, affine_map<(d0, d1) -> (d0, d1)>, 2>
%2 = memref.alloc() : memref<1xi32>
%c0 = arith.constant 0 : index
%c64 = arith.constant 64 : index
affine.for %i0 = 0 to 10 {
affine.for %i1 = 0 to 10 {
affine.dma_start %0[%i0, %i1], %1[%i0, %i1], %2[%c0], %c64
: memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
affine.dma_wait %2[%c0], %c64 : memref<1xi32>
// CHECK: affine.dma_start %{{.*}}[%{{.*}}, %{{.*}}], %{{.*}}[%{{.*}}, %{{.*}}], %{{.*}}[%{{.*}}], %{{.*}} : memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
// CHECK: affine.dma_wait %{{.*}}[%{{.*}}], %{{.*}} : memref<1xi32>
}
}
return
}
// -----
// Test with loop IVs and optional stride arguments.
func.func @test1(%arg0 : index, %arg1 : index) {
%0 = memref.alloc() : memref<100x100xf32>
%1 = memref.alloc() : memref<100x100xf32, affine_map<(d0, d1) -> (d0, d1)>, 2>
%2 = memref.alloc() : memref<1xi32>
%c0 = arith.constant 0 : index
%c64 = arith.constant 64 : index
%c128 = arith.constant 128 : index
%c256 = arith.constant 256 : index
affine.for %i0 = 0 to 10 {
affine.for %i1 = 0 to 10 {
affine.dma_start %0[%i0, %i1], %1[%i0, %i1], %2[%c0], %c64, %c128, %c256
: memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
affine.dma_wait %2[%c0], %c64 : memref<1xi32>
// CHECK: affine.dma_start %{{.*}}[%{{.*}}, %{{.*}}], %{{.*}}[%{{.*}}, %{{.*}}], %{{.*}}[%{{.*}}], %{{.*}}, %{{.*}}, %{{.*}} : memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
// CHECK: affine.dma_wait %{{.*}}[%{{.*}}], %{{.*}} : memref<1xi32>
}
}
return
}
// -----
// Test with loop IVs and symbols (without symbol keyword).
func.func @test2(%arg0 : index, %arg1 : index) {
%0 = memref.alloc() : memref<100x100xf32>
%1 = memref.alloc() : memref<100x100xf32, affine_map<(d0, d1) -> (d0, d1)>, 2>
%2 = memref.alloc() : memref<1xi32>
%c0 = arith.constant 0 : index
%c64 = arith.constant 64 : index
affine.for %i0 = 0 to 10 {
affine.for %i1 = 0 to 10 {
affine.dma_start %0[%i0 + %arg0, %i1], %1[%i0, %i1 + %arg1 + 5],
%2[%c0], %c64
: memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
affine.dma_wait %2[%c0], %c64 : memref<1xi32>
// CHECK: affine.dma_start %{{.*}}[%{{.*}} + %{{.*}}, %{{.*}}], %{{.*}}[%{{.*}}, %{{.*}} + %{{.*}} + 5], %{{.*}}[%{{.*}}], %{{.*}} : memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
// CHECK: affine.dma_wait %{{.*}}[%{{.*}}], %{{.*}} : memref<1xi32>
}
}
return
}
// -----
// Test with loop IVs and symbols (with symbol keyword).
func.func @test3(%arg0 : index, %arg1 : index) {
%0 = memref.alloc() : memref<100x100xf32>
%1 = memref.alloc() : memref<100x100xf32, affine_map<(d0, d1) -> (d0, d1)>, 2>
%2 = memref.alloc() : memref<1xi32>
%c0 = arith.constant 0 : index
%c64 = arith.constant 64 : index
affine.for %i0 = 0 to 10 {
affine.for %i1 = 0 to 10 {
affine.dma_start %0[%i0 + symbol(%arg0), %i1],
%1[%i0, %i1 + symbol(%arg1) + 7],
%2[%i0 + %i1 + 11], %c64
: memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
affine.dma_wait %2[%c0], %c64 : memref<1xi32>
// CHECK: affine.dma_start %{{.*}}[%{{.*}} + symbol(%{{.*}}), %{{.*}}], %{{.*}}[%{{.*}}, %{{.*}} + symbol(%{{.*}}) + 7], %{{.*}}[%{{.*}} + %{{.*}} + 11], %{{.*}} : memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
// CHECK: affine.dma_wait %{{.*}}[%{{.*}}], %{{.*}} : memref<1xi32>
}
}
return
}
// -----
// Test with loop IVs, symbols and constants in nested affine expressions.
func.func @test4(%arg0 : index, %arg1 : index) {
%0 = memref.alloc() : memref<100x100xf32>
%1 = memref.alloc() : memref<100x100xf32, 2>
%2 = memref.alloc() : memref<1xi32>
%c64 = arith.constant 64 : index
affine.for %i0 = 0 to 10 {
affine.for %i1 = 0 to 10 {
affine.dma_start %0[(%i0 + symbol(%arg0)) floordiv 3, %i1],
%1[%i0, (%i1 + symbol(%arg1)) mod 9 + 7],
%2[%i0 + %i1 + 11], %c64
: memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
affine.dma_wait %2[%i0 + %i1 + 11], %c64 : memref<1xi32>
// CHECK: affine.dma_start %{{.*}}[(%{{.*}} + symbol(%{{.*}})) floordiv 3, %{{.*}}], %{{.*}}[%{{.*}}, (%{{.*}} + symbol(%{{.*}})) mod 9 + 7], %{{.*}}[%{{.*}} + %{{.*}} + 11], %{{.*}} : memref<100x100xf32>, memref<100x100xf32, 2>, memref<1xi32>
// CHECK: affine.dma_wait %{{.*}}[%{{.*}} + %{{.*}} + 11], %{{.*}} : memref<1xi32>
}
}
return
}
|
