Add memory format support to empty_like operator#27244
Closed
VitalyFedyunin wants to merge 17 commits intogh/VitalyFedyunin/5/basefrom
Closed
Add memory format support to empty_like operator#27244VitalyFedyunin wants to merge 17 commits intogh/VitalyFedyunin/5/basefrom
empty_like operator#27244VitalyFedyunin wants to merge 17 commits intogh/VitalyFedyunin/5/basefrom
Conversation
[ghstack-poisoned]
[ghstack-poisoned]
Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. [ghstack-poisoned]
Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. [ghstack-poisoned]
This was referenced Oct 8, 2019
Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. [ghstack-poisoned]
Closed
Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. [ghstack-poisoned]
This was referenced Oct 14, 2019
…ator" Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. [ghstack-poisoned]
…operator" Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. [ghstack-poisoned]
… operator" Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. [ghstack-poisoned]
This was referenced Oct 16, 2019
houseroad
reviewed
Oct 17, 2019
| supports_named_tensor: True | ||
|
|
||
| - func: empty_like.dtype(Tensor self, *, ScalarType dtype, Layout layout, Device device, bool pin_memory=False, MemoryFormat? memory_format=contiguous_format) -> Tensor | ||
| - func: empty_like.dtype(Tensor self, *, ScalarType dtype, Layout layout, Device device, bool pin_memory=False, MemoryFormat? memory_format=None) -> Tensor |
Member
There was a problem hiding this comment.
None and contigous_format (i.e., 0) have the same behavior?
Contributor
Author
There was a problem hiding this comment.
Yes, but it is exceptional case. No need to handle it in the test.
Contributor
Author
|
@ailzhang hi |
ezyang
reviewed
Oct 18, 2019
| if (self.is_quantized()) { | ||
|
|
||
| auto memory_format = | ||
| optional_memory_format.value_or(MemoryFormat::Contiguous); |
Contributor
There was a problem hiding this comment.
To be changed later right :>
Contributor
Author
There was a problem hiding this comment.
Yes, right now we need to introduce this option, after we check all current code for expected behaviour and change default.
Contributor
There was a problem hiding this comment.
If it's to be changed later, can we add a comment linking some issue or with some information ?
ezyang
reviewed
Oct 18, 2019
| if decl['name'].endswith('_like') and not has_tensor_options: | ||
| # it's a tensor | ||
| actuals.append('{}.options().is_variable(false)'.format(actuals[0])) | ||
| if decl['name'] == 'empty_like': |
Contributor
There was a problem hiding this comment.
Umm... this looks very non-general. (I assume it gets fixed later in the stack but it is very shocking here)
Contributor
Author
There was a problem hiding this comment.
Yes, need to list by function until I convert all of them
ezyang
reviewed
Oct 18, 2019
| check_sum_all(torch.tensor([True, False, True], dtype=torch.bool, device=device)) | ||
|
|
||
| def _test_memory_format_transformations(self, device, input_generator_fn, transformation_fn): | ||
| def _test_memory_format_transformations(self, device, input_generator_fn, transformation_fn, compare_data=True): |
ezyang
approved these changes
Oct 18, 2019
This was referenced Oct 18, 2019
…ator" Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. Differential Revision: [D17980310](https://our.internmc.facebook.com/intern/diff/D17980310) [ghstack-poisoned]
…perator" Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. Differential Revision: [D17980310](https://our.internmc.facebook.com/intern/diff/D17980310) [ghstack-poisoned]
Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. Differential Revision: [D17980310](https://our.internmc.facebook.com/intern/diff/D17980310) [ghstack-poisoned]
…to `empty_like` operator" Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. Differential Revision: [D17980310](https://our.internmc.facebook.com/intern/diff/D17980310) [ghstack-poisoned]
…ke` operator" Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. Differential Revision: [D17980310](https://our.internmc.facebook.com/intern/diff/D17980310) [ghstack-poisoned]
Closed
zdevito
pushed a commit
to zdevito/ATen
that referenced
this pull request
Oct 25, 2019
Summary: Pull Request resolved: pytorch/pytorch#27244 Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. Test Plan: Imported from OSS Reviewed By: ezyang Differential Revision: D17980310 Pulled By: VitalyFedyunin fbshipit-source-id: 00a39b40daa4b8ee63c32e60d920222f8be2d6a1
Contributor
|
@VitalyFedyunin merged this pull request in baf8488. |
xxtEchjovs44
pushed a commit
to xxtEchjovs44/pytorch
that referenced
this pull request
Jan 29, 2020
ghstack-source-id: 3b495ce Pull Request resolved: pytorch/pytorch#27244
thiagocrepaldi
pushed a commit
to thiagocrepaldi/pytorch
that referenced
this pull request
Feb 4, 2020
Summary: Pull Request resolved: pytorch#27244 Adds memory_format keyword argument (positional for cpp). 'Preserve' behavior now follows next rules: 1) If tensor is non-overlapping and dense - output tensor will have the same strides as input tensor. 2) If not (1) and tensor is stored in the channels last format, output tensor going to have channels last format. 3) Output tensor is going to be contiguous in all other cases. --- Dense tensor is the tensor that store values in a contiguous block of memory. Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory. Test Plan: Imported from OSS Reviewed By: ezyang Differential Revision: D17980310 Pulled By: VitalyFedyunin fbshipit-source-id: 00a39b40daa4b8ee63c32e60d920222f8be2d6a1
ezyang
reviewed
Feb 19, 2020
|
|
||
| - func: empty_like(Tensor self) -> Tensor | ||
| use_c10_dispatcher: full | ||
| - func: empty_like(Tensor self, *, MemoryFormat? memory_format=None) -> Tensor |
Contributor
There was a problem hiding this comment.
With the benefit of hindsight, I should have asked: why do we have two overloads of empty_like in the first place?!
Contributor
Author
There was a problem hiding this comment.
Do you mean this line and line 1123?
Contributor
There was a problem hiding this comment.
yes. But see my most recent PR.
Contributor
Author
There was a problem hiding this comment.
As far as I remember combining them together was never considered by me.
karansachdev-1012
pushed a commit
to karansachdev-1012/pytorch
that referenced
this pull request
Feb 17, 2026
ghstack-source-id: b9323d5 Pull Request resolved: pytorch/pytorch#27244
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 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.
Stack from ghstack:
resize_op. #28292 Add memory format support to theresize_op.operator==of TensorOptions as confusing one #28076 Killoperator==of TensorOptions as confusing oneresize_as_operator #27979 Add memory format support toresize_as_operatorrandn_likeoperator #27890 Add memory format support torandn_likeoperatorrandint_likeoperator #27889 Add memory format support torandint_likeoperatorzeros_likeoperator #27562 Add memory format support tozeros_likeoperatorrand_likeoperator #27561 Add memory format support torand_likeoperatorones_likeoperator #27270 Add memory format support toones_likeoperatorfull_likeoperator #27262 Add memory format support tofull_likeoperatorempty_likeoperator #27244 Add memory format support toempty_likeoperatorAdds memory_format keyword argument (positional for cpp).
'Preserve' behavior now follows next rules:
Dense tensor is the tensor that store values in a contiguous block of memory.
Non-overlapping tensor is the tensor in which elements occupy individual non-repetitive memory.
Differential Revision: D17980310