[MPS] Fix sliced cast #138314
[MPS] Fix sliced cast #138314
Conversation
This fixes internal crash due to the invalid bufer size computation if sliced API is used
Not sure what was the purpose of
```c++
IntArrayRef baseShape;
if (src.is_view()) {
baseShape = src._base().sizes();
} else {
baseShape = getIMPSAllocator()->getBufferShape(src.storage().data());
}
int flattenedShaped = 1;
for (const auto i : c10::irange(baseShape.size())) {
flattenedShaped *= baseShape[i];
}
```
As flattenShaped could be much easier computed as `[srcBuf
lengh]/src.element_size()`
When someone allocated buffer to hold say uint8 and that view-casted it
to float16, attempt to compute `baseShape` returned sizes of original
tensor in its data type, rather than size in new dtypes
Fixes #137800
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/138314
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 451e406 with merge base a9014d2 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
ciflow/mps
albanD
left a comment
albanD
left a comment
There was a problem hiding this comment.
Sounds good thanks!
We can use storage().size() as well if we don't want to query the srcBuf.
|
I think the main idea was that the storage buffer might be bigger than the actual |
|
@pytorchbot merge -f "MPS changes are green" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
|
@pytorchbot cherry-pick --onto release/2.5 -c regression |
This fixes internal crash due to the invalid bufer size computation if sliced API is used
Not sure what was the purpose of
```c++
IntArrayRef baseShape;
if (src.is_view()) {
baseShape = src._base().sizes();
} else {
baseShape = getIMPSAllocator()->getBufferShape(src.storage().data());
}
int flattenedShaped = 1;
for (const auto i : c10::irange(baseShape.size())) {
flattenedShaped *= baseShape[i];
}
```
As flattenShaped could be much easier computed as `[srcBuf
lengh]/src.element_size()`, and even if `srcBuf` is padded it's a safe thing to do.
When someone allocated buffer to hold say uint8 and that view-casted it
to float16, attempt to compute `baseShape` returned sizes of original
tensor in its data type, rather than size in new dtypes
Fixes #137800
Pull Request resolved: #138314
Approved by: https://github.com/albanD, https://github.com/DenisVieriu97
(cherry picked from commit de16159)
Cherry picking #138314The cherry pick PR is at #138535 and it is recommended to link a regression cherry pick PR with an issue. Details for Dev Infra teamRaised by workflow job |
[MPS] Fix sliced cast (#138314) This fixes internal crash due to the invalid bufer size computation if sliced API is used Not sure what was the purpose of ```c++ IntArrayRef baseShape; if (src.is_view()) { baseShape = src._base().sizes(); } else { baseShape = getIMPSAllocator()->getBufferShape(src.storage().data()); } int flattenedShaped = 1; for (const auto i : c10::irange(baseShape.size())) { flattenedShaped *= baseShape[i]; } ``` As flattenShaped could be much easier computed as `[srcBuf lengh]/src.element_size()`, and even if `srcBuf` is padded it's a safe thing to do. When someone allocated buffer to hold say uint8 and that view-casted it to float16, attempt to compute `baseShape` returned sizes of original tensor in its data type, rather than size in new dtypes Fixes #137800 Pull Request resolved: #138314 Approved by: https://github.com/albanD, https://github.com/DenisVieriu97 (cherry picked from commit de16159) Co-authored-by: Nikita Shulga <[email protected]>
This fixes internal crash due to the invalid bufer size computation if sliced API is used
Not sure what was the purpose of
As flattenShaped could be much easier computed as
[srcBuf lengh]/src.element_size(), and even ifsrcBufis padded it's a safe thing to do.When someone allocated buffer to hold say uint8 and that view-casted it
to float16, attempt to compute
baseShapereturned sizes of originaltensor in its data type, rather than size in new dtypes
Fixes #137800