Skip to content

Fix topk work size computation #5053

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
soumith merged 2 commits into from
Feb 6, 2018
Merged

Fix topk work size computation #5053

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c5f4868
fix grid computation for topk kernel
albanD Feb 5, 2018
dbaca59
backslash alignment, no change in code
albanD Feb 5, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 9 additions & 7 deletions aten/src/THC/generic/THCTensorTopK.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,7 @@ THC_API void THCTensor_(topk)(THCState* state,
THLongStorage_free(topKSize);

#define RUN_K(INDEX_T, DIM, DIR) \
gatherTopK<real, INDEX_T, DIM, DIR> \
gatherTopK<real, INDEX_T, DIM, DIR> \
<<<grid, block, 0, THCState_getCurrentStream(state)>>>( \
inputInfo, \
sliceSize, \
Expand Down Expand Up @@ -63,10 +63,10 @@ THC_API void THCTensor_(topk)(THCState* state,
}

#define RUN_T(INDEX_T) \
TensorInfo<real, INDEX_T> inputInfo = \
getTensorInfo<THCTensor, INDEX_T>(state, input); \
TensorInfo<real, INDEX_T> topKInfo = \
getTensorInfo<THCTensor, INDEX_T>(state, topK); \
TensorInfo<real, INDEX_T> inputInfo = \
getTensorInfo<THCTensor, INDEX_T>(state, input); \
TensorInfo<real, INDEX_T> topKInfo = \
getTensorInfo<THCTensor, INDEX_T>(state, topK); \
TensorInfo<int64_t, INDEX_T> indicesInfo = \
getTensorInfo<THCudaLongTensor, INDEX_T>(state, indices); \
\
Expand All @@ -82,9 +82,11 @@ THC_API void THCTensor_(topk)(THCState* state,
int collapseIndicesDim = indicesInfo.collapseDims(dim); \
\
int64_t inputSlices = 1; \
int64_t topKSlices = 1; \
for (int i = 0; i < numDims; ++i) { \
for (int i = 0; i < inputInfo.dims; ++i) { \
inputSlices *= inputInfo.sizes[i]; \
} \
int64_t topKSlices = 1; \
for (int i = 0; i < topKInfo.dims; ++i) { \
topKSlices *= topKInfo.sizes[i]; \
} \
\
Expand Down