-
Notifications
You must be signed in to change notification settings - Fork 22.2k
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
[dtensor] skip pytree when not necessary #110132
Commits on Sep 27, 2023
-
[dtensor] skip pytree when not necessary
pytree is a great tool, but it sometimes considers to be evil for tensor subclasses, it's useful to implement subclass quickly, but it: * exposes non-trival CPU overhead * many ops don't need pytree, only the one with list/dict ops needs * blindly use pytree to re-wrap have semantic issues for inplace/out ops This PR avoid using pytree for most ops during torch_dispatch and only enable it for certain ops [ghstack-poisoned]
Configuration menu - View commit details
-
Copy full SHA for 56b3d3f - Browse repository at this point
Copy the full SHA 56b3d3fView commit details
Commits on Sep 28, 2023
-
Update on "[dtensor] skip pytree when not necessary"
pytree is a great tool, but it sometimes considers to be evil for tensor subclasses, it's useful to implement subclass quickly, but it: * exposes non-trival CPU overhead * many ops don't need pytree, only the one with list/dict ops needs * blindly use pytree to re-wrap have semantic issues for inplace/out ops This PR avoid using pytree for most ops during torch_dispatch and only enable it for certain ops [ghstack-poisoned]
Configuration menu - View commit details
-
Copy full SHA for da0080c - Browse repository at this point
Copy the full SHA da0080cView commit details -
Update on "[dtensor] skip pytree when not necessary"
pytree is a great tool, but it sometimes considers to be evil for tensor subclasses, it's useful to implement subclass quickly, but it: * exposes non-trival CPU overhead * many ops don't need pytree, only the one with list/dict ops needs * blindly use pytree to re-wrap have semantic issues for inplace/out ops This PR avoid using pytree for most ops during torch_dispatch and only enable it for certain ops [ghstack-poisoned]
Configuration menu - View commit details
-
Copy full SHA for 2c674c0 - Browse repository at this point
Copy the full SHA 2c674c0View commit details -
Update on "[dtensor] skip pytree when not necessary"
pytree is a great tool, but it sometimes considers to be evil for tensor subclasses, it's useful to implement subclass quickly, but it: * exposes non-trival CPU overhead * many ops don't need pytree, only the one with list/dict ops needs * blindly use pytree to re-wrap have semantic issues for inplace/out ops This PR avoid using pytree for most ops during torch_dispatch and only enable it for certain ops [ghstack-poisoned]
Configuration menu - View commit details
-
Copy full SHA for 6322903 - Browse repository at this point
Copy the full SHA 6322903View commit details -
Update on "[dtensor] skip pytree when not necessary"
pytree is a great tool, but it sometimes considers to be evil for tensor subclasses, it's useful to implement subclass quickly, but it: * exposes non-trival CPU overhead * many ops don't need pytree, only the one with list/dict ops needs * blindly use pytree to re-wrap have semantic issues for inplace/out ops This PR avoid using pytree for most ops during torch_dispatch and only enable it for certain ops [ghstack-poisoned]
Configuration menu - View commit details
-
Copy full SHA for ed322f4 - Browse repository at this point
Copy the full SHA ed322f4View commit details -
Update on "[dtensor] skip pytree when not necessary"
pytree is a great tool, but it sometimes considers to be evil for tensor subclasses, it's useful to implement subclass quickly, but it: * exposes non-trival CPU overhead * many ops don't need pytree, only the one with list/dict ops needs * blindly use pytree to re-wrap have semantic issues for inplace/out ops This PR avoid using pytree for most ops during torch_dispatch and only enable it for certain ops [ghstack-poisoned]
Configuration menu - View commit details
-
Copy full SHA for e661f9f - Browse repository at this point
Copy the full SHA e661f9fView commit details -
Update on "[dtensor] skip pytree when not necessary"
pytree is a great tool, but it sometimes considers to be evil for tensor subclasses, it's useful to implement subclass quickly, but it: * exposes non-trival CPU overhead * many ops don't need pytree, only the one with list/dict ops needs * blindly use pytree to re-wrap have semantic issues for inplace/out ops This PR avoid using pytree for most ops during torch_dispatch and only enable it for certain ops [ghstack-poisoned]
Configuration menu - View commit details
-
Copy full SHA for 9be485d - Browse repository at this point
Copy the full SHA 9be485dView commit details