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

[dnn] better specification of recompilation key

Browse Source
This commit is contained in:
Philippe Tillet
2019-08-02 17:42:48 -07:00
parent 3b92ddf7e6
commit d9945692a9
31 changed files with 418 additions and 428 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
examples/python/tensorflow/dot.cpp
View File

@@ -49,7 +49,7 @@ class DotOp : public OpKernel {
triton::driver::cu_buffer db(ctx, b.tensor_data().size(), (CUdeviceptr)b.tensor_data().data(), false);
triton::driver::cu_buffer dc(ctx, c->tensor_data().size(), (CUdeviceptr)c->tensor_data().data(), false);
// template
triton::dnn::dot dot(M, N, K, false, false, "fp16", "fp16", 8, 8);
triton::dnn::dot dot(M, N, K, false, false, "half", "half", 8, 8, 8);
dot.enqueue(stream, {&da, &db, &dc});
}