_foreach_norm: Align with PyTorch operator semantics on allocation scheme of return tensors #709
Conversation
This file contains 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
fengyuan14
changed the title
|
The error is exposed by -WError in preci. Please fix it, |
fengyuan14
reviewed
Aug 8, 2024
fengyuan14
reviewed
Aug 8, 2024
fengyuan14
approved these changes
Aug 12, 2024
2 tasks
|
fengyuan14
reviewed
Aug 16, 2024
| ret_per_tensor.push_back(at::empty({}, res_option)); | ||
| } | ||
| sycl::queue q{sycl::property::queue::in_order()}; | ||
| void** tensor_list_addresses = sycl::malloc_shared<void*>((ntensors), q); |
There was a problem hiding this comment.
Do not use raw runtime malloc, but helpers provided by PyTorch XPU backend. The usage here, should be,
- Allocating ping memory
- Initing meta on ping memory
- Allocating device memory
- memcpy from ping to dev
fengyuan14
approved these changes
Aug 19, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
PyTorch requires separate copies returned in foreach_norm. The existing XPU implementation follows an out-of-date allocation scheme, to share storage among returned tensor. In latest PyTorch unit test, the behavior is not allowed.
related case: