Vectorize erf #89837
Vectorize erf #89837
Conversation
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/89837
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 1fcea57: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
EikanWang
added
topic: not user facing
| return f"{x}.exp()" | ||
|
|
||
| @staticmethod | ||
| def erf(x): |
There was a problem hiding this comment.
I see #89388 didn't add erf here. This raises a general question: people changing the cpp backend may easily miss here, or not sure about if vectorization applies for the new op. Can we have a better way to avoid this kind of problem in future (I think Jason asked about using inheritance in one of the initial vectorization PRs)?
There was a problem hiding this comment.
That's why @EikanWang added a UT to make sure the op list is the same for explicit vec and omp vec here:
https://github.com/pytorch/pytorch/pull/89837/files#diff-809c39aeafb3acc92289f42a63e670a8719d4ce5627d5f88820142d80edf8d2aR4950
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
Pull Request resolved: pytorch#89837 Approved by: https://github.com/jgong5, https://github.com/desertfire, https://github.com/jansel
6 participants
Stack from ghstack (oldest at bottom):
cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @mlazos @soumith @voznesenskym @yanboliang @penguinwu @anijain2305 @Guobing-Chen @chunyuan-w @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @peterbell10 @desertfire