Reland #161649, vectorize stored in cat for all dtypes#162440
Reland #161649, vectorize stored in cat for all dtypes#162440
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/162440
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 82805d4 with merge base 5fd6b6a ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
ngimel
added
ciflow/trunk
ngimel
force-pushed
the
ngimel/cat_perf2
branch
from
f6c4787 to
82805d4
Compare
| // which requires the input tensor addresses to be aligned to a | ||
| // 16 Byte boundary. | ||
|
|
||
| constexpr bool isContig = stride_size == 1; |
There was a problem hiding this comment.
Do we care about stride 0 or is that considered not contiguous?
There was a problem hiding this comment.
stride_size is the dimension of the strides, stride_size template argument is set to 1 if all input and output tensors return true for .is_contiguous
| for (int i = nDims - 1; i >= 0; --i) { | ||
| outputParam.tensorSize[i] = out.size(i); | ||
| outputParam.tensorStride[i] = out.stride(i); | ||
| if (isContig) { |
There was a problem hiding this comment.
Should be if constexpr
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
…h#162440) Per title Pull Request resolved: pytorch#162440 Approved by: https://github.com/Skylion007
…h#162440) Per title Pull Request resolved: pytorch#162440 Approved by: https://github.com/Skylion007
…h#162440) Per title Pull Request resolved: pytorch#162440 Approved by: https://github.com/Skylion007
5 participants
Per title