Add high-level operator interface #708
Merged
Conversation
This file contains 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
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/708
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 72cfb21 with merge base ac8ce4c ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D60321449 |
|
This pull request was exported from Phabricator. Differential Revision: D60321449 |
metascroy
added a commit
to metascroy/ao
that referenced
this pull request
Aug 19, 2024
Summary: Pull Request resolved: pytorch#708 Here is a draft of the high-level kernel operator. This is something I quickly put together to get early feedback, and it will be iterated. It closely follows what I had in the design doc, but I think it's easier to see how things fit together in the code. The operator uses parallel_for and get_num_threads. At build time, we need to link these to ET/PT implementations. In this diff, they are single-threaded or OMP-backed implementations that I threw together. The operator (inside channelwise_8bit_activation_groupwise_lowbit_weight.h) depends on a ukernel_config and tiling configs. But in the wrapper, these can be hidden if desired (see examples/ for how). Reviewed By: digantdesai Differential Revision: D60321449
metascroy
force-pushed
the
export-D60321449
branch
from
a446838
to
c617411
Compare
Summary: Pull Request resolved: pytorch#708 Here is a draft of the high-level kernel operator. This is something I quickly put together to get early feedback, and it will be iterated. It closely follows what I had in the design doc, but I think it's easier to see how things fit together in the code. The operator uses parallel_for and get_num_threads. At build time, we need to link these to ET/PT implementations. In this diff, they are single-threaded or OMP-backed implementations that I threw together. The operator (inside channelwise_8bit_activation_groupwise_lowbit_weight.h) depends on a ukernel_config and tiling configs. But in the wrapper, these can be hidden if desired (see examples/ for how). Reviewed By: digantdesai Differential Revision: D60321449
|
This pull request was exported from Phabricator. Differential Revision: D60321449 |
metascroy
force-pushed
the
export-D60321449
branch
from
c617411
to
72cfb21
Compare
jerryzh168
approved these changes
Aug 20, 2024
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.
Summary:
Here is a draft of the high-level kernel operator. This is something I quickly put together to get early feedback, and it will be iterated. It closely follows what I had in the design doc, but I think it's easier to see how things fit together in the code.
The operator uses parallel_for and get_num_threads. At build time, we need to link these to ET/PT implementations. In this diff, they are single-threaded or OMP-backed implementations that I threw together.
The operator (inside channelwise_8bit_activation_groupwise_lowbit_weight.h) depends on a ukernel_config and tiling configs. But in the wrapper, these can be hidden if desired (see examples/ for how).
Reviewed By: digantdesai
Differential Revision: D60321449