Resolve int[]? arguments to new OptionalIntArrayRef class
#70864
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
CI Flow Status⚛️ CI FlowRuleset - Version:
You can add a comment to the PR and tag @pytorchbot with the following commands: # ciflow rerun, "ciflow/default" will always be added automatically
@pytorchbot ciflow rerun
# ciflow rerun with additional labels "-l <ciflow/label_name>", which is equivalent to adding these labels manually and trigger the rerun
@pytorchbot ciflow rerun -l ciflow/scheduled -l ciflow/slowFor more information, please take a look at the CI Flow Wiki. |
Contributor
🔗 Helpful links
💊 CI failures summary and remediationsAs of commit 005ad76 (more details on the Dr. CI page): 💚 💚 Looks good so far! There are no failures yet. 💚 💚 This comment was automatically generated by Dr. CI (expand for details).Please report bugs/suggestions to the (internal) Dr. CI Users group. |
facebook-github-bot
added
the
oncall: jit
kurtamohler
changed the title
int[]? dim arguments to new OptionalIntArrayRef class
peterbell10
reviewed
Jan 6, 2022
tools/codegen/api/python.py
Outdated
Comment on lines
605
to
608
Collaborator
There was a problem hiding this comment.
I think OptionalArrayRef<T> makes sense for all types, not just int. e.g. a float[]? called with an int would likely have the exact same dangling pointer issue.
Collaborator
Author
There was a problem hiding this comment.
That's true. Alright, I'll focus on fixing the remaining issues with int[]? --> OptionalIntArrayRef, and after that's fully working, I'll generalize for all OptionalArrayRef<T>
peterbell10
reviewed
Jan 12, 2022
kurtamohler
force-pushed
the
dim-list-schema
branch
2 times, most recently
from
3ec8494 to
c4f5afd
Compare
kurtamohler
commented
Jan 18, 2022
aten/src/ATen/core/boxing/impl/make_boxed_from_unboxed_functor.h
Outdated
Show resolved
Hide resolved
peterbell10
reviewed
Jan 18, 2022
aten/src/ATen/core/boxing/impl/make_boxed_from_unboxed_functor.h
Outdated
Show resolved
Hide resolved
kurtamohler
force-pushed
the
dim-list-schema
branch
4 times, most recently
from
ce8d647 to
6db2234
Compare
kurtamohler
commented
Jan 21, 2022
torch/backends/_nnapi/serializer.py
Outdated
Collaborator
Author
There was a problem hiding this comment.
The changes I made to this file are kind of a hacky workaround. The upsample functions, not just upsample_nearest2d, have two different overloads that the parser checks for in the corresponding THPVariable_* function in torch/csrc/autograd/generated/python_nn_functions.cpp. For upsample_nearest2d, this used to be how the parser was defined before this PR:
static PythonArgParser parser({
"upsample_nearest2d(Tensor input, IntArrayRef? output_size, ArrayRef<double>? scale_factors)",
"upsample_nearest2d(Tensor input, IntArrayRef[2] output_size, double? scales_h=None, double? scales_w=None, *, Tensor out=None)",
}, /*traceable=*/true);If upsample_nearest2d is called with just the input and output_size, and no other arguments, both of the function schemas are a match, but the parser checks them in order, so the first one would be chosen. That worked fine until this PR changed IntArrayRef? to OptionalIntArrayRef. These entries are sorted in alphabetical order, so now they are switched:
static PythonArgParser parser({
"upsample_nearest2d(Tensor input, IntArrayRef[2] output_size, double? scales_h=None, double? scales_w=None, *, Tensor out=None)",
"upsample_nearest2d(Tensor input, OptionalIntArrayRef output_size, ArrayRef<double>? scale_factors)",
}, /*traceable=*/true); And now the wrong overload is chosen. The one with scales_h and scales_w is an old overload that still exists for BC, and it doesn't have all the features that the newer overload has.
One of the missing features was support for this NNAPI serializer. The change I made in this file adds support for that, and it seems to be working. It looks like upsample_nearest2d was the only upsample function that was supported here.
However, there's at least one more missing feature that won't be so simple (nor advisable, probably?) to add for these old overloads--forward AD. At the moment, all of the upsample functions are failing forward AD tests because of the overload reordering. If I could just change the order back to what it used to be, the issue should be fixed, but at the moment, I'm not completely sure how to do that. We could add specialized logic in codegen to reorder the upsample functions, but I'm not too sure if that's the best way to go or not. I haven't thought of any better idea though.
We could create THPVariable_ functions manually in tools/autograd/templates/python_nn_functions.cpp, but there are quite a few upsample functions affected by this:
THPVariable_upsample_bicubic2dTHPVariable_upsample_bilinear2dTHPVariable_upsample_linear1dTHPVariable_upsample_nearest1dTHPVariable_upsample_nearest2dTHPVariable_upsample_nearest3dTHPVariable_upsample_trilinear3d
@peterbell10 or @mruberry, do you have any thoughts?
Collaborator
Author
There was a problem hiding this comment.
I just tried the codegen reordering idea, and it seems to work. I just reversed the grouped_overloads within the method_impl() function in tools/autograd/gen_python_functions.py any time it sees an upsample function:
It looks like there already are some special-cased comparisons in the sorting function here:
So I guess it might not be a bad idea to just add some logic there to prioritize OptionalIntArrayRef over IntArrayRef, or something like that.
Collaborator
There was a problem hiding this comment.
I think all ambiguous overloads should behave equivalently; so I would consider this a bug in the nnapi code. The scales_h overload even looks to be the main overload in ATen, the other one just wraps it.
Also, while I just said it shouldn't matter. IntArrayRef is a more specific overload than OptionalIntArrayRef so it seems backwards to sort the Optional version first.
Collaborator
Author
There was a problem hiding this comment.
That's a fair point. Then I guess we should enable forward AD for theIntArrayRef overloads?
Edit: I didn't realize how simple it is to add forward AD support for these--just needed to add result: auto_linear to each of them in derivatives.yaml. So I'll go ahead and push it
kurtamohler
force-pushed
the
dim-list-schema
branch
2 times, most recently
from
d4ef2fb to
ad61f38
Compare
kurtamohler
changed the title
int[]? dim arguments to new OptionalIntArrayRef classint[]? arguments to new OptionalIntArrayRef class
kurtamohler
force-pushed
the
dim-list-schema
branch
2 times, most recently
from
b96cc6e to
8f636f5
Compare
kurtamohler
requested review from
IvanYashchuk,
albanD,
lezcano,
nikitaved and
soulitzer
as code owners
pytorchmergebot
pushed a commit
that referenced
this pull request
Sep 14, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
pytorchmergebot
pushed a commit
that referenced
this pull request
Sep 14, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
pytorchmergebot
pushed a commit
that referenced
this pull request
Sep 14, 2023
Pull Request resolved: #101995 We have C++17 now! I am intentionally dropping the c10::optional<c10::ArrayRef> size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use optional<ArrayRef> in function arguments anymore anyway. ghstack-source-id: 30231b8 Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/)
swolchok
added a commit
that referenced
this pull request
Nov 11, 2023
Pull Request resolved: #101995 We have C++17 now! I am intentionally dropping the c10::optional<c10::ArrayRef> size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use optional<ArrayRef> in function arguments anymore anyway. ghstack-source-id: 207227009 Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/)
swolchok
added a commit
that referenced
this pull request
Nov 11, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 11, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 13, 2023
…:optional to std::optional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 13, 2023
…::optional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
Pull Request resolved: #101995 We have C++17 now! I am intentionally dropping the c10::optional<c10::ArrayRef> size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use optional<ArrayRef> in function arguments anymore anyway. ghstack-source-id: 207417170 Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/)
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
Pull Request resolved: #101995 We have C++17 now! I am intentionally dropping the c10::optional<c10::ArrayRef> size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use optional<ArrayRef> in function arguments anymore anyway. ghstack-source-id: 207509669 Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/)
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 14, 2023
Pull Request resolved: #101995 We have C++17 now! I am intentionally dropping the c10::optional<c10::ArrayRef> size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use optional<ArrayRef> in function arguments anymore anyway. ghstack-source-id: 207558686 Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/)
swolchok
added a commit
that referenced
this pull request
Nov 28, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 28, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 29, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab mrshenli zhaojuanmao rohan-varma kiukchung d4l3k lucasllc XilunWu tianyu-l [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 29, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab mrshenli zhaojuanmao rohan-varma kiukchung d4l3k lucasllc XilunWu tianyu-l [ghstack-poisoned]
swolchok
added a commit
that referenced
this pull request
Nov 29, 2023
Pull Request resolved: #101995 We have C++17 now! I am intentionally dropping the c10::optional<c10::ArrayRef> size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use optional<ArrayRef> in function arguments anymore anyway. ghstack-source-id: 208581191 Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/)
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 30, 2023
Pull Request resolved: #101995 We have C++17 now! I am intentionally dropping the c10::optional<c10::ArrayRef> size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use optional<ArrayRef> in function arguments anymore anyway. ghstack-source-id: be773c9 Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/)
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 30, 2023
…ptional" We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab mrshenli zhaojuanmao rohan-varma kiukchung d4l3k lucasllc XilunWu tianyu-l [ghstack-poisoned]
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 30, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) cc ezyang gchanan jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10 EikanWang H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab mrshenli zhaojuanmao rohan-varma kiukchung d4l3k lucasllc XilunWu tianyu-l [ghstack-poisoned]
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 30, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / #70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) Pull Request resolved: #101995 Approved by: https://github.com/malfet, https://github.com/Skylion007, https://github.com/ezyang
dmenig
pushed a commit
to dmenig/pytorch
that referenced
this pull request
Dec 21, 2023
We have C++17 now! I am intentionally dropping the `c10::optional<c10::ArrayRef>` size optimization. It was intended to improve dispatch, but thanks to D34602980 / pytorch#70864 we don't use `optional<ArrayRef>` in function arguments anymore anyway. Differential Revision: [D46079028](https://our.internmc.facebook.com/intern/diff/D46079028/) Pull Request resolved: pytorch#101995 Approved by: https://github.com/malfet, https://github.com/Skylion007, https://github.com/ezyang
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
9 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.
This PR uses the
OptionalArrayReftemplate class that was drafted in #64084.Fixes #44409