1// RUN: mlir-opt %s -convert-scf-to-std -convert-vector-to-llvm -convert-std-to-llvm | \
2// RUN: mlir-cpu-runner -e entry -entry-point-result=void  \
3// RUN:   -shared-libs=%mlir_integration_test_dir/libmlir_c_runner_utils%shlibext | \
4// RUN: FileCheck %s
5
6func @entry() {
7  %f0 = constant 0.0: f32
8  %f1 = constant 1.0: f32
9  %f2 = constant 2.0: f32
10  %f3 = constant 3.0: f32
11  %f4 = constant 4.0: f32
12  %f5 = constant 5.0: f32
13  %f6 = constant 6.0: f32
14  %f7 = constant 7.0: f32
15  %f8 = constant 8.0: f32
16  %f9 = constant 9.0: f32
17  %f10 = constant 10.0: f32
18  %f11 = constant 11.0: f32
19  %f12 = constant 12.0: f32
20  %f13 = constant 13.0: f32
21  %f14 = constant 14.0: f32
22  %f15 = constant 15.0: f32
23
24  %a0 = vector.broadcast %f0 : f32 to vector<4x4xf32>
25  %a1 = vector.insert %f0, %a0[0, 0] : f32 into vector<4x4xf32>
26  %a2 = vector.insert %f1, %a1[0, 1] : f32 into vector<4x4xf32>
27  %a3 = vector.insert %f2, %a2[0, 2] : f32 into vector<4x4xf32>
28  %a4 = vector.insert %f3, %a3[0, 3] : f32 into vector<4x4xf32>
29  %a5 = vector.insert %f4, %a4[1, 0] : f32 into vector<4x4xf32>
30  %a6 = vector.insert %f5, %a5[1, 1] : f32 into vector<4x4xf32>
31  %a7 = vector.insert %f6, %a6[1, 2] : f32 into vector<4x4xf32>
32  %a8 = vector.insert %f7, %a7[1, 3] : f32 into vector<4x4xf32>
33  %a9 = vector.insert %f8, %a8[2, 0] : f32 into vector<4x4xf32>
34  %a10 = vector.insert %f9, %a9[2, 1] : f32 into vector<4x4xf32>
35  %a11 = vector.insert %f10, %a10[2, 2] : f32 into vector<4x4xf32>
36  %a12 = vector.insert %f11, %a11[2, 3] : f32 into vector<4x4xf32>
37  %a13 = vector.insert %f12, %a12[3, 0] : f32 into vector<4x4xf32>
38  %a14 = vector.insert %f13, %a13[3, 1] : f32 into vector<4x4xf32>
39  %a15 = vector.insert %f14, %a14[3, 2] : f32 into vector<4x4xf32>
40  %a16 = vector.insert %f15, %a15[3, 3] : f32 into vector<4x4xf32>
41
42  vector.print %a16 : vector<4x4xf32>
43  //
44  // test matrix:
45  //
46  // CHECK: ( ( 0, 1, 2, 3 ), ( 4, 5, 6, 7 ), ( 8, 9, 10, 11 ), ( 12, 13, 14, 15 ) )
47
48  // Tile 4x4 with 3x3 as follows:
49  //
50  //   +--------+--+
51  //   +0   1  2| 3|
52  //   |4   5  6| 7|
53  //   |8   9 10|11|
54  //   +--------+--+
55  //   |12 13 14|15|
56  //   +--------+--+
57  //
58  %es = vector.extract_slices %a16, [3, 3], [1, 1] :
59     vector<4x4xf32> into tuple<vector<3x3xf32>, vector<3x1xf32>, vector<1x3xf32>, vector<1x1xf32>>
60
61  %0 = vector.tuple_get %es, 0 : tuple<vector<3x3xf32>, vector<3x1xf32>, vector<1x3xf32>, vector<1x1xf32>>
62  %1 = vector.tuple_get %es, 1 : tuple<vector<3x3xf32>, vector<3x1xf32>, vector<1x3xf32>, vector<1x1xf32>>
63  %2 = vector.tuple_get %es, 2 : tuple<vector<3x3xf32>, vector<3x1xf32>, vector<1x3xf32>, vector<1x1xf32>>
64  %3 = vector.tuple_get %es, 3 : tuple<vector<3x3xf32>, vector<3x1xf32>, vector<1x3xf32>, vector<1x1xf32>>
65
66  vector.print %0 : vector<3x3xf32>
67  vector.print %1 : vector<3x1xf32>
68  vector.print %2 : vector<1x3xf32>
69  vector.print %3 : vector<1x1xf32>
70  //
71  // extract slices:
72  //
73  // CHECK: ( ( 0, 1, 2 ), ( 4, 5, 6 ), ( 8, 9, 10 ) )
74  // CHECK: ( ( 3 ), ( 7 ), ( 11 ) )
75  // CHECK: ( ( 12, 13, 14 ) )
76  // CHECK: ( ( 15 ) )
77
78  return
79}
80