-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest.mlir
64 lines (64 loc) · 5.89 KB
/
test.mlir
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
// -------------- kernel0
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [4, 1], instrShape = [16, 128, 16]}>
#blocked2 = #triton_gpu.blocked<{sizePerThread = [8, 1], threadsPerWarp = [16, 2], warpsPerCTA = [1, 4], order = [0, 1]}>
%396 = triton_gpu.convert_layout %395 : tensor<128x128xf16, #mma> -> tensor<128x128xf16, #blocked2> loc(#loc139)
// -------------- kernel1
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 32]}>
#blocked3 = #triton_gpu.blocked<{sizePerThread = [1, 16], threadsPerWarp = [2, 16], warpsPerCTA = [8, 1], order = [1, 0]}>
%412 = triton_gpu.convert_layout %411 : tensor<128x256xf8E5M2, #mma> -> tensor<128x256xf8E5M2, #blocked3> loc(#loc345)
// -------------- kernel2
#blocked = #triton_gpu.blocked<{sizePerThread = [1, 16], threadsPerWarp = [8, 4], warpsPerCTA = [4, 1], order = [1, 0]}>
#blocked1 = #triton_gpu.blocked<{sizePerThread = [16, 1], threadsPerWarp = [4, 8], warpsPerCTA = [1, 4], order = [0, 1]}>
%108 = triton_gpu.convert_layout %92 : tensor<64x64xi8, #blocked> -> tensor<64x64xi8, #blocked1> loc(#loc70)
// -------------- kernel3
#blocked2 = #triton_gpu.blocked<{sizePerThread = [1, 16], threadsPerWarp = [2, 16], warpsPerCTA = [8, 1], order = [1, 0]}>
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 32]}>
%409 = triton_gpu.convert_layout %408 : tensor<128x256xf8E4M3FNUZ, #mma> -> tensor<128x256xf8E4M3FNUZ, #blocked2> loc(#loc364)
// -------------- kernel4
#blocked = #triton_gpu.blocked<{sizePerThread = [8], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
#blocked1 = #triton_gpu.blocked<{sizePerThread = [4], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
%21 = triton_gpu.convert_layout %18 : tensor<1024xf8E5M2, #blocked> -> tensor<1024xf8E5M2, #blocked1> loc(#loc11)
// -------------- kernel5
#blocked2 = #triton_gpu.blocked<{sizePerThread = [1, 8], threadsPerWarp = [1, 32], warpsPerCTA = [8, 1], order = [1, 0]}>
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 16]}>
%428 = triton_gpu.convert_layout %427 : tensor<128x256xf16, #mma> -> tensor<128x256xf16, #blocked2> loc(#loc368)
// -------------- kernel6
#blocked = #triton_gpu.blocked<{sizePerThread = [8], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
#blocked1 = #triton_gpu.blocked<{sizePerThread = [4], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
%21 = triton_gpu.convert_layout %18 : tensor<1024xf16, #blocked> -> tensor<1024xf16, #blocked1> loc(#loc11)
// -------------- kernel7
#blocked2 = #triton_gpu.blocked<{sizePerThread = [8, 1], threadsPerWarp = [16, 2], warpsPerCTA = [1, 8], order = [0, 1]}>
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 16]}>
%464 = triton_gpu.convert_layout %463 : tensor<128x256xf16, #mma> -> tensor<128x256xf16, #blocked2> loc(#loc357)
// -------------- kernel8
#blocked = #triton_gpu.blocked<{sizePerThread = [8], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
#blocked1 = #triton_gpu.blocked<{sizePerThread = [4], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
%21 = triton_gpu.convert_layout %18 : tensor<1024xf16, #blocked> -> tensor<1024xf16, #blocked1> loc(#loc11)
// -------------- kernel9
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 16]}>
#blocked2 = #triton_gpu.blocked<{sizePerThread = [8, 1], threadsPerWarp = [16, 2], warpsPerCTA = [1, 8], order = [0, 1]}>
%464 = triton_gpu.convert_layout %463 : tensor<128x256xf16, #mma> -> tensor<128x256xf16, #blocked2> loc(#loc357)
// -------------- kernel10
#blocked = #triton_gpu.blocked<{sizePerThread = [1, 16], threadsPerWarp = [8, 4], warpsPerCTA = [4, 1], order = [1, 0]}>
#blocked1 = #triton_gpu.blocked<{sizePerThread = [16, 1], threadsPerWarp = [4, 8], warpsPerCTA = [1, 4], order = [0, 1]}>
%108 = triton_gpu.convert_layout %92 : tensor<64x64xf8E5M2, #blocked> -> tensor<64x64xf8E5M2, #blocked1> loc(#loc68)
// -------------- kernel11
#blocked3 = #triton_gpu.blocked<{sizePerThread = [1, 16], threadsPerWarp = [2, 16], warpsPerCTA = [8, 1], order = [1, 0]}>
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 32]}>
%412 = triton_gpu.convert_layout %411 : tensor<128x256xf8E5M2, #mma> -> tensor<128x256xf8E5M2, #blocked3> loc(#loc345)
// -------------- kernel12
#blocked2 = #triton_gpu.blocked<{sizePerThread = [1, 8], threadsPerWarp = [1, 32], warpsPerCTA = [8, 1], order = [1, 0]}>
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 16]}>
%428 = triton_gpu.convert_layout %427 : tensor<128x256xf16, #mma> -> tensor<128x256xf16, #blocked2> loc(#loc368)
// -------------- kernel13
#blocked2 = #triton_gpu.blocked<{sizePerThread = [1, 16], threadsPerWarp = [2, 16], warpsPerCTA = [8, 1], order = [1, 0]}>
#mma = #triton_gpu.nvidia_mma<{versionMajor = 3, versionMinor = 0, warpsPerCTA = [8, 1], instrShape = [16, 256, 32]}>
%409 = triton_gpu.convert_layout %408 : tensor<128x256xf8E4M3FNUZ, #mma> -> tensor<128x256xf8E4M3FNUZ, #blocked2> loc(#loc364)
// -------------- kernel14
#blocked = #triton_gpu.blocked<{sizePerThread = [8], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
#blocked1 = #triton_gpu.blocked<{sizePerThread = [4], threadsPerWarp = [32], warpsPerCTA = [4], order = [0]}>
%21 = triton_gpu.convert_layout %18 : tensor<1024xf8E5M2, #blocked> -> tensor<1024xf8E5M2, #blocked1> loc(#loc11)
// -------------- kernel15
#blocked = #triton_gpu.blocked<{sizePerThread = [1, 16], threadsPerWarp = [8, 4], warpsPerCTA = [4, 1], order = [1, 0]}>
#blocked1 = #triton_gpu.blocked<{sizePerThread = [16, 1], threadsPerWarp = [4, 8], warpsPerCTA = [1, 4], order = [0, 1]}>
%108 = triton_gpu.convert_layout %92 : tensor<64x64xi8, #blocked> -> tensor<64x64xi8, #blocked1> loc(#loc70)