Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[opengl] Cache loaded arr elements. #3901

Closed
wants to merge 3 commits into from

Conversation

ailzhang
Copy link
Contributor

@ailzhang ailzhang commented Dec 29, 2021

Stack from ghstack:

WIP: implementation is a bit hacky, for testing only...

Noticed that we're reloading the same elements in ndarrays again and
again in the opengl generated code. This doesn't affect JIT perf on NV
GPUs but I'm suspicious that it's one source of bottleneck on snapdragon
android.

Unlike read-only shape information saved in args buffer, cached ndarray
elements should be invalidated once it's overwritten.

WIP: implementation is a bit hacky, for testing only...

Noticed that we're reloading the same elements in ndarrays again and
again in the opengl generated code. This doesn't affect JIT perf on NV
GPUs but I'm suspicious that it's one source of bottleneck on snapdragon
android.

Unlike read-only shape information saved in args buffer, cached ndarray
elements should be invalidated once it's overwritten.

[ghstack-poisoned]
ailzhang pushed a commit that referenced this pull request Dec 29, 2021
WIP: implementation is a bit hacky, for testing only...

Noticed that we're reloading the same elements in ndarrays again and
again in the opengl generated code. This doesn't affect JIT perf on NV
GPUs but I'm suspicious that it's one source of bottleneck on snapdragon
android.

Unlike read-only shape information saved in args buffer, cached ndarray
elements should be invalidated once it's overwritten.

ghstack-source-id: b0421029dce09397b8b0f2a71b77221f43705d3d
Pull Request resolved: #3901
WIP: implementation is a bit hacky, for testing only...

Noticed that we're reloading the same elements in ndarrays again and
again in the opengl generated code. This doesn't affect JIT perf on NV
GPUs but I'm suspicious that it's one source of bottleneck on snapdragon
android.

Unlike read-only shape information saved in args buffer, cached ndarray
elements should be invalidated once it's overwritten.

[ghstack-poisoned]
ailzhang pushed a commit that referenced this pull request Dec 29, 2021
WIP: implementation is a bit hacky, for testing only...

Noticed that we're reloading the same elements in ndarrays again and
again in the opengl generated code. This doesn't affect JIT perf on NV
GPUs but I'm suspicious that it's one source of bottleneck on snapdragon
android.

Unlike read-only shape information saved in args buffer, cached ndarray
elements should be invalidated once it's overwritten.

ghstack-source-id: b0421029dce09397b8b0f2a71b77221f43705d3d
Pull Request resolved: #3901
WIP: implementation is a bit hacky, for testing only...

Noticed that we're reloading the same elements in ndarrays again and
again in the opengl generated code. This doesn't affect JIT perf on NV
GPUs but I'm suspicious that it's one source of bottleneck on snapdragon
android.

Unlike read-only shape information saved in args buffer, cached ndarray
elements should be invalidated once it's overwritten.

[ghstack-poisoned]
ailzhang pushed a commit that referenced this pull request Dec 29, 2021
WIP: implementation is a bit hacky, for testing only...

Noticed that we're reloading the same elements in ndarrays again and
again in the opengl generated code. This doesn't affect JIT perf on NV
GPUs but I'm suspicious that it's one source of bottleneck on snapdragon
android.

Unlike read-only shape information saved in args buffer, cached ndarray
elements should be invalidated once it's overwritten.

ghstack-source-id: 7ae7a18286a8ac6fcc8f1ed3e8f22132e1441e70
Pull Request resolved: #3901
@ailzhang
Copy link
Contributor Author

superseded by #3992

@ailzhang ailzhang closed this Jan 11, 2022
@ailzhang ailzhang deleted the gh/ailzhang/50/head branch February 21, 2022 09:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant