KEMBAR78
Add Half support for range, logspace, logit, median, nanmedian, kthvalue, poisson, cummax, cummin, prod, cumprod, histc, logcumsumexp, vander, cross, aten2, logaddexp, logaddexp2, hypot, and nextafter on CPU by CaoE · Pull Request #107764 · pytorch/pytorch · GitHub
Skip to content

Conversation

@CaoE
Copy link
Collaborator

@CaoE CaoE commented Aug 23, 2023

Add Half support for range, logspace, logit, median, nanmedian, kthvalue, poisson, cummax, cummin, prod, cumprod, histc, logcumsumexp, vander, cross, aten2, logaddexp, logaddexp2, hypot, and nextafter on CPU.

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @voznesenskym @penguinwu @EikanWang @Guobing-Chen @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @aakhundov @ColinPeppler @Xia-Weiwen @ngimel

@pytorch-bot
Copy link

pytorch-bot bot commented Aug 23, 2023

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/107764

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

✅ You can merge normally! (1 Unrelated Failure)

As of commit 6fe7bf7 with merge base 64fd027 (image):

FLAKY - The following job failed but was likely due to flakiness present on trunk:

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

@github-actions github-actions bot added the module: cpu CPU specific problem (e.g., perf, algorithm) label Aug 23, 2023
@CaoE CaoE added module: half Related to float16 half-precision floats topic: not user facing topic category ciflow/inductor ciflow/trunk Trigger trunk jobs on your pull request ciflow/mps Run MPS tests (subset of trunk) labels Aug 23, 2023
@CaoE CaoE force-pushed the fp16_ops branch 5 times, most recently from 12e9b2e to 212f29a Compare August 30, 2023 08:35
@CaoE CaoE force-pushed the fp16_ops branch 9 times, most recently from 352e7b4 to badb645 Compare September 7, 2023 06:25
@CaoE CaoE force-pushed the fp16_ops branch 7 times, most recently from db6daf0 to 0096ea0 Compare September 13, 2023 01:17
@CaoE CaoE force-pushed the fp16_ops branch 2 times, most recently from 619bf41 to facd4df Compare September 15, 2023 06:47
@CaoE CaoE force-pushed the fp16_ops branch 3 times, most recently from b1cc5aa to 38abff4 Compare September 26, 2023 06:40
@CaoE CaoE force-pushed the fp16_ops branch 2 times, most recently from a8fe409 to d5ae1be Compare October 17, 2023 07:10
@CaoE CaoE requested review from jgong5 and mingfeima October 19, 2023 01:42
@CaoE CaoE added the ciflow/periodic Trigger jobs ran periodically on master (periodic.yml) on the PR label Oct 19, 2023
b_16 = b.to(dtype=lowp_dtype)
actual_16 = a_16.atan2(b_16)
self.assertEqual(actual_16, actual.to(dtype=lowp_dtype))
self.assertEqual(expected, actual_16.view(-1), exact_dtype=False, rtol=0, atol=0.02)
Copy link
Collaborator

Choose a reason for hiding this comment

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

you are not using the rtol and atol variables set previously?

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

Thanks for your comments. Fixed it.

@CaoE CaoE marked this pull request as ready for review October 24, 2023 00:43
@CaoE CaoE requested a review from cpuhrsch October 25, 2023 05:56
@CaoE
Copy link
Collaborator Author

CaoE commented Oct 25, 2023

@cpuhrsch Could you please review this PR ? Thank you.

@cpuhrsch
Copy link
Contributor

@CaoE - I know this might be annoying, but can we split this up a bit more?

While the number of lines changed isn't high, the number of ops is. That'll make it harder to bisect if there's compounding issues.

…lue, poisson, cummax, cummin, prod, cumprod, histc, logcumsumexp, vander, cross, aten2, logaddexp, logaddexp2, hypot, and nextafter on CPU
@CaoE
Copy link
Collaborator Author

CaoE commented Oct 30, 2023

@cpuhrsch Thanks for your comments. I split this PR into #112124, #112131, #112132, #112135, and #112138. Could you please review them ? Thank you.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ciflow/inductor ciflow/mps Run MPS tests (subset of trunk) ciflow/periodic Trigger jobs ran periodically on master (periodic.yml) on the PR ciflow/trunk Trigger trunk jobs on your pull request module: cpu CPU specific problem (e.g., perf, algorithm) module: half Related to float16 half-precision floats module: inductor open source topic: not user facing topic category

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants