-
Notifications
You must be signed in to change notification settings - Fork 295
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
Provide kernels with true reference implementations for quantized ops #4108
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/4108
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 150d4ab with merge base 7047162 (): This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This pull request was exported from Phabricator. Differential Revision: D59238748 |
This pull request was exported from Phabricator. Differential Revision: D59238748 |
…#4108) Summary: Pull Request resolved: #4108 We want to be able to run the reference implementations on x86, so we don't want any intrinsics or anything like that in the reference kernels. In the end, this change has a lot of things: - introduce a `reference` folder for reference implementations - moved the primary cmake flow from HiFi to reference, so that the default mode can run on x86 - that means we will need a proper flag to use HiFi optimized ops, which we will add later - add a `quantized_matmul` reference kernel Differential Revision: D59238748
a9f0c4d
to
beb0137
Compare
This pull request was exported from Phabricator. Differential Revision: D59238748 |
…#4108) Summary: Pull Request resolved: #4108 We want to be able to run the reference implementations on x86, so we don't want any intrinsics or anything like that in the reference kernels. In the end, this change has a lot of things: - introduce a `reference` folder for reference implementations - moved the primary cmake flow from HiFi to reference, so that the default mode can run on x86 - that means we will need a proper flag to use HiFi optimized ops, which we will add later - add a `quantized_matmul` reference kernel Differential Revision: D59238748
beb0137
to
9215276
Compare
…#4108) Summary: Pull Request resolved: #4108 We want to be able to run the reference implementations on x86, so we don't want any intrinsics or anything like that in the reference kernels. In the end, this change has a lot of things: - introduce a `reference` folder for reference implementations - moved the primary cmake flow from HiFi to reference, so that the default mode can run on x86 - that means we will need a proper flag to use HiFi optimized ops, which we will add later - add a `quantized_matmul` reference kernel Differential Revision: D59238748
This pull request was exported from Phabricator. Differential Revision: D59238748 |
9215276
to
150d4ab
Compare
This pull request has been merged in ef640bf. |
Summary:
We want to be able to run the reference implementations on x86, so we don't want any intrinsics or anything like that in the reference kernels.
In the end, this change has a lot of things:
reference
folder for reference implementationsquantized_matmul
reference kernelDifferential Revision: D59238748