-
Notifications
You must be signed in to change notification settings - Fork 797
Implement dual LUT scheme for int16 reciprocal/rsqrt operations #16396
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
base: main
Are you sure you want to change the base?
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16396
Note: Links to docs will display an error until the docs builds have been completed. ❌ 7 New Failures, 1 Cancelled Job, 1 Unrelated FailureAs of commit 20401cc with merge base dbf3c37 ( NEW FAILURES - The following jobs have failed:
CANCELLED JOB - The following job was cancelled. Please retry:
UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
digantdesai
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Review automatically exported from Phabricator review in Meta.
|
@pytorchbot label "release notes: none" |
|
NOTE: I have pending changes before merging !! |
Summary: Dual LUT Strategy (from Bolt reference) 1. **Range LUT**: Covers full int16 range, LOW resolution at small values 2. **Resolution LUT**: Scaled input (x*128), HIGH resolution for codes 0-256 Interpolation Weights - **x <= 255**: Use resolution_lut (high accuracy for small values) - **x >= 256**: Use range_lut (works fine for large values) - **Transition at 255/256** to avoid interpolation artifacts Buck Build Fixes - Created TARGETS file in `/fb` subdirectory to enable Buck discovery - Restructured parent TARGETS to break circular dependency (core → fb, passes → core + fb) - Fixed Python circular import by importing directly from module files Next Steps - Full test suite validation - Performance benchmarking - Reduce number of rescale ops Reviewed By: digantdesai Differential Revision: D87120925
5060a1c to
20401cc
Compare
Summary:
Dual LUT Strategy
Reviewed By: digantdesai
Differential Revision: D87120925