[pytorch] introduce INTERN_DISABLE_AUTOGRAD flag to create inference only library for mobile #25697
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
…only library for mobile Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. Test Plan: - will check CI; - test mobile build in sample app;
pytorchbot
added
caffe2
module: autograd
This was referenced Sep 5, 2019
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. Test Plan: - will check CI; - test mobile build in sample app; Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 5, 2019
…only library for mobile Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; ghstack-source-id: 31317fc Pull Request resolved: #25697
gchanan
reviewed
Sep 5, 2019
torch/csrc/autograd/variable.cpp
Outdated
| if (!diff_view_meta->grad_fn_ && !diff_view_meta->base_.requires_grad()) { | ||
| return diff_view_meta->grad_fn_; | ||
| } | ||
| #ifndef C10_MOBILE |
Contributor
There was a problem hiding this comment.
as mentioned over chat, this part seems difficult to maintain.
The rest doesn't seem to be adding that much complexity, because we already expose AutoNonVariableType as part of the pseudo-public interface and you are just slicing along that border.
Contributor
There was a problem hiding this comment.
The reason this is commented out is, I assume, because of the direct reference to generated::AsStridedBackward? That doesn't seem too bad to me.
Contributor
Author
There was a problem hiding this comment.
Yes, generated::AsStridedBackward is the main blocker.
Ok, I updated the PR to keep the Functions.h/cpp for now to keep generated::AsStridedBackward and remove #if/#endif here.
In next PR I can update the codegen script to only generate this single function for mobile build - Does it sound better than the gating? @gchanan
Contributor
I wanted to claim that it should (morally) be possible to reduce the size of a static library by specifying what symbols you want to keep, and then doing a DCE with those symbols as roots. But I spent a while googling and it doesn't seem like this is a thing people actually do. I want to confirm that the static library size for iOS is a cosmetic issue: in the end, you'll still end up with an executable of the same size--it's just that it looks bad if your static library looks big, is that right? |
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
This was referenced Sep 6, 2019
Contributor
Author
Correct |
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 6, 2019
…D flag to create inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 6, 2019
…only library for mobile Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; ghstack-source-id: 8f45c06 Pull Request resolved: #25697
ljk53
added a commit
that referenced
this pull request
Sep 7, 2019
…D flag to create inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ezyang
reviewed
Sep 9, 2019
|
|
||
| # Load deprecated signatures | ||
| deprecated = load_deprecated_signatures( | ||
| aten_decls, os.path.join(autograd_dir, 'deprecated.yaml')) |
Contributor
There was a problem hiding this comment.
Oh? Is this dead? Joyous day!
Contributor
Author
There was a problem hiding this comment.
It's still used by gen_autograd_python code path :)
load_deprecated_signatures() doesn't seem to have side-effect so I assume it's safe to delete since the result is not used here.
ezyang
approved these changes
Sep 9, 2019
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…D flag to create inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…D flag to create inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
This was referenced Sep 10, 2019
… inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…D flag to create inference only library for mobile" Summary: This is the first of a series of changes to reduce build size by cutting autograd functions from mobile build. When INTERN_DISABLE_AUTOGRAD is set: * On CMake side we exclude Functions.h/cpp, VariableType*.h/cpp, VariableTypeManual.cpp from the build process. Still keep variable_factories.h as we rely on it to create variables instead of tensors. * In source code we gate a couple autograd references (in autograd/variable.cpp) with C10_MOBILE (technically we should use a dedicated c macro but its maintenance cost is higher than cmake macro as we have several build systems to change). * Pass --disable-autograd flag to codegen script, which will stop generating Functions/VariableType code. And for variable_factories.h it will stop generating tracing code. Edit: in this diff we will keep Functions.h/cpp to avoid changing source code. Why we need this change if it's already not calling VariableType and autograd stuff with USE_STATIC_DISPATCH=ON for mobile? It's trying to reduce static library size for iOS build, for which it's relatively harder to strip size with linker approach. Why we need make involved change into codegen script? There isn't a global config system in codegen - autograd/env.py provides similar functionality but it says not adding anything there. Test Plan: - will check CI; - test mobile build in sample app; Pull Request resolved: #25697 Differential Revision: [D17202733](https://our.internmc.facebook.com/intern/diff/D17202733)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…n codegen if disable_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 10, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 11, 2019
…n codegen if disable_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 11, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 11, 2019
…n codegen if disable_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 11, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 11, 2019
…n codegen if disable_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
ljk53
added a commit
that referenced
this pull request
Sep 11, 2019
…able_autograd is set" Summary: As discussed on PR #25697 - to remove autograd/functions.* without using intrusive #if/#else we need whitelist "AsStridedBackward" used by variable.cpp. Test Plan: - builds and runs with stacked diffs Pull Request resolved: #25817 Differential Revision: [D17247235](https://our.internmc.facebook.com/intern/diff/D17247235)
Contributor
Author
|
This is already landed: 8485710 Don't know why it's not closed automatically... |
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.
Stack from ghstack:
Summary:
This is the first of a series of changes to reduce build size by cutting
autograd functions from mobile build.
When INTERN_DISABLE_AUTOGRAD is set:
VariableTypeManual.cpp from the build process. Still keep variable_factories.h
as we rely on it to create variables instead of tensors.
with C10_MOBILE (technically we should use a dedicated c macro but its
maintenance cost is higher than cmake macro as we have several build systems
to change).
Functions/VariableType code. And for variable_factories.h it will stop
generating tracing code.
Edit: in this diff we will keep Functions.h/cpp to avoid changing source code.
Why we need this change if it's already not calling VariableType and autograd
stuff with USE_STATIC_DISPATCH=ON for mobile?
It's trying to reduce static library size for iOS build, for which it's
relatively harder to strip size with linker approach.
Why we need make involved change into codegen script?
There isn't a global config system in codegen - autograd/env.py provides similar
functionality but it says not adding anything there.
Test Plan:
Pull Request resolved: #25697
Differential Revision: D17202733