gaspersic/triton
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[EXAMPLES][TUTORIAL] Changed to new triton.kernel API

Browse Source
This commit is contained in:
Philippe Tillet
2020-07-08 13:39:19 -04:00
committed by Philippe Tillet
parent c33d6d15f5
commit 4ccd78f1a6
1 changed files with 21 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

41
python/examples/tutorials/mat_mul.py
View File

@@ -16,16 +16,14 @@ class _dot(torch.autograd.Function):
// accumulator
float c[TM, TN] = 0;
//pointers
// pointers
TYPE* pa[TM, TK] = A + rk[newaxis, :] * 1 + rm[:, newaxis] * lda;
TYPE* pb[TK, TN] = B + rk[:, newaxis] * ldb + rn[newaxis, :] * 1;
for(int k=K; k>0; k-=TK) {
TYPE a[TM, TK] = *pa;
TYPE b[TK, TN] = *pb;
c += a @ b;
pa = pa + TK * 1;
pb = pb + TK * ldb;
}
@@ -40,32 +38,35 @@ class _dot(torch.autograd.Function):
c = _dot._call(a,b)
return c
kernel = dict()
@staticmethod
def _call(a, b):
# shapes
M, K = a.shape
K, N = b.shape
# leading dimension
lda = K
ldb = N
ldc = N
dtype = a.dtype
# create kernel if necessary
if dtype not in _dot.kernel:
defines = {
'TYPE' : dtype,
'TM' : [64, 128],
'TN' : [64, 128],
'TK' : [8, 16],
}
_dot.kernel[dtype] = triton.kernel(_dot.src, num_warps=[2, 4], defines=defines)
kernel = _dot.kernel[dtype]
# allocate output
c = triton.empty([M,N], dtype=dtype)
grid = lambda opt: [triton.cdiv(M, opt.d('TM')), triton.cdiv(N, opt.d('TN'))]
defines= {
'TYPE' : dtype,
'TM' : [32,64,128],
'TN' : [32,64,128],
'TK' : [8],
}
_dot.kernel = triton.kernel(_dot.src, defines=defines)
_dot.kernel(a, b, c, M, N, K, lda, ldb, ldc,
grid=grid, num_warps=4, defines=defines)
# enqueue
grid = lambda opt: [triton.cdiv(M, opt.d('TM')),
triton.cdiv(N, opt.d('TN'))]
kernel(a, b, c, M, N, K, lda, ldb, ldc, grid=grid)
return c
@@ -81,4 +82,4 @@ b = torch.rand((K, N)).cuda()
zc = torch.matmul(a,b)
zc_ = dot(a,b)
print(torch.allclose(zc, zc_))
print(torch.allclose(zc, zc_))