Compute type_equal() without reference to backend()#53823
Closed
ezyang wants to merge 7 commits intogh/ezyang/943/basefrom
Closed
Compute type_equal() without reference to backend()#53823ezyang wants to merge 7 commits intogh/ezyang/943/basefrom
ezyang wants to merge 7 commits intogh/ezyang/943/basefrom
Conversation
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> [ghstack-poisoned]
This was referenced Mar 11, 2021
This was referenced Mar 11, 2021
Contributor
💊 CI failures summary and remediationsAs of commit 2b53b5f (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).Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions to the (internal) Dr. CI Users group. |
ezyang
added a commit
that referenced
this pull request
Mar 11, 2021
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> ghstack-source-id: e16f265 Pull Request resolved: #53823
smessmer
approved these changes
Mar 11, 2021
bhosmer
approved these changes
Mar 12, 2021
bhosmer
left a comment
There was a problem hiding this comment.
This makes sense but it seems pretty fragile.
- any idea what it saves us, perf-wise?
- can we throw in a lightweight injectivity test somewhere?
Contributor
Author
My intention wasn't really perf; I just really didn't want to put Meta in the list of Backends XD
I... guess? But honestly I'd rather spend the time just getting rid of Backend (as we have kept saying we would do, and then haven't) |
Makes sense. So maybe just a comment on |
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> [ghstack-poisoned]
This was referenced Mar 14, 2021
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> Differential Revision: [D27036575](https://our.internmc.facebook.com/intern/diff/D27036575) [ghstack-poisoned]
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> Differential Revision: [D27036575](https://our.internmc.facebook.com/intern/diff/D27036575) [ghstack-poisoned]
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> Differential Revision: [D27036575](https://our.internmc.facebook.com/intern/diff/D27036575) [ghstack-poisoned]
This was referenced Mar 15, 2021
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> Differential Revision: [D27036575](https://our.internmc.facebook.com/intern/diff/D27036575) [ghstack-poisoned]
Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> Differential Revision: [D27036575](https://our.internmc.facebook.com/intern/diff/D27036575) [ghstack-poisoned]
This was referenced Mar 15, 2021
Contributor
laurentdupin
pushed a commit
to laurentdupin/pytorch
that referenced
this pull request
Apr 24, 2026
Summary: Pull Request resolved: pytorch#53823 Argument for correctness: type_equal previous compared if backends are equal. Backend is computed by translation from dispatch key. I verified that computeDispatchKey never computed a weird dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend was effectively injective. Then it is always valid to compare the arguments of an injective function for equality, rather than the output of the injective function. Signed-off-by: Edward Z. Yang <ezyang@fb.com> Test Plan: Imported from OSS Reviewed By: mruberry Differential Revision: D27036575 Pulled By: ezyang fbshipit-source-id: 6aeafc89f287da0bc0065bd21c1adb5e272dbb81
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
4 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:
Argument for correctness: type_equal previous compared if backends
are equal. Backend is computed by translation from dispatch key.
I verified that computeDispatchKey never computed a weird
dispatch key (e.g., AutogradXLA), so that dispatchKeyToBackend
was effectively injective. Then it is always valid to compare
the arguments of an injective function for equality, rather than
the output of the injective function.
Signed-off-by: Edward Z. Yang ezyang@fb.com
Differential Revision: D27036575