Skip to content

[WIP] Dispatch Friendly collectives#93379

Closed
wconstab wants to merge 10 commits intopytorch:masterfrom
kumpera:land_will_pr
Closed

[WIP] Dispatch Friendly collectives#93379
wconstab wants to merge 10 commits intopytorch:masterfrom
kumpera:land_will_pr

Conversation

@wconstab
Copy link
Copy Markdown
Contributor

@wconstab wconstab commented Jan 31, 2023

@pytorch-bot
Copy link
Copy Markdown

pytorch-bot bot commented Jan 31, 2023

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/93379

Note: Links to docs will display an error until the docs builds have been completed.

❌ 6 Failures

As of commit d008410:

NEW FAILURES - The following jobs have failed:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@kumpera kumpera changed the title Land will pr [WIP] Dispatch Friendly collectives Jan 31, 2023
self: grad.reshape_symint(self.sym_sizes())
result: auto_linear

- name: all_reduce(Tensor self, str reduceOp, str tag, int[] ranks, int stride) -> Tensor
Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

have you thought about how to handle str reduceOp?

i think it is OK to leave it as str instead of a new C++ type.

however, we need a way to translate between str/enum, and I didn't think about where to put that translator yet.

currently, i think we only need this in python (since inductor also calls into py api), so maybe we can keep it together with the pg mapping utils.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it should be in the same place as the op implementation.

@albanD albanD removed their request for review January 31, 2023 18:28
Rodrigo Kumpera added 2 commits January 31, 2023 21:12
wconstab added a commit that referenced this pull request Feb 8, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 8, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 9, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
@kumpera kumpera closed this Feb 10, 2023
@kumpera kumpera deleted the land_will_pr branch February 10, 2023 18:38
wconstab added a commit that referenced this pull request Feb 10, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 10, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 13, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 13, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 15, 2023
(updated 2/10)

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

ghstack-source-id: 2030643
Pull Request resolved: #94059
wconstab added a commit that referenced this pull request Feb 15, 2023
…ce Support"

NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
wconstab added a commit that referenced this pull request Feb 15, 2023
NOTE: this is a hacked squashed version of Rodrigo's PR (#93379) designed to unblock inductor impl work.  Wait for real PR to land then rebase the inductor stuff on top of that.

[ghstack-poisoned]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants