Fix cuBLAS arguments for fp16 dot #3660
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
soumith
requested changes
Nov 12, 2017
|
|
||
| #ifdef CUDA_HALF_TENSOR | ||
| float THCudaBlas_Hdot(THCState *state, int64_t n, half *x, int64_t incx, half *y, int64_t incy) | ||
| half THCudaBlas_Hdot(THCState *state, int64_t n, half *x, int64_t incx, half *y, int64_t incy) |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
apaszke
force-pushed
the
half_dot_fix
branch
from
122f681 to
2e1f834
Compare
apaszke
force-pushed
the
half_dot_fix
branch
from
2e1f834 to
f17f2bf
Compare
|
@soumith can you please review again? I added tests for CUDA half <-> CPU float comparison |
|
looks good! |
colesbury
added a commit
to colesbury/pytorch
that referenced
this pull request
Feb 1, 2018
The test_cuda.py setup purports to test half tensors, but actually just re-tests FloatTensors because the keys in type_map were str instead of type. Testing HalfTensors is more complicated, requiring changes to precision and requires excluding some unimplemented methods. We should fully test half CUDA tensors. This change just deletes the duplicate tests of FloatTensor.
colesbury
added a commit
that referenced
this pull request
Feb 7, 2018
The test_cuda.py setup purports to test half tensors, but actually just re-tests FloatTensors because the keys in type_map were str instead of type. Testing HalfTensors is more complicated, requiring changes to precision and requires excluding some unimplemented methods. We should fully test half CUDA tensors. This change just deletes the duplicate tests of FloatTensor.
soumith
pushed a commit
that referenced
this pull request
Feb 7, 2018
* Fix cuBLAS arguments for fp16 dot * Enable FloatTensor <-> CUDA HalfTensor checks in test_cuda.py
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Result type has to be fp16 for fp16 dot. See docs of
cublasDotEx(look for "datatypes combinations currrently supported").