Skip to content
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

Fix FP6-LLM benchmark #312

Merged
merged 3 commits into from
Jun 4, 2024
Merged

Fix FP6-LLM benchmark #312

merged 3 commits into from
Jun 4, 2024

Conversation

gau-nernst
Copy link
Collaborator

@gau-nernst gau-nernst commented Jun 3, 2024

Use the new user API for FP6-LLM benchmark.

4070 Ti SUPER

m k n fp6_latency (ms) fp16_latency (ms) speedup (d/s) correct
1 8192 8192 0.0259424 0.212709 8.1993 1
1 10240 8192 0.104522 0.265314 2.53835 1
1 57344 8192 0.558859 1.47519 2.63965 1
1 8192 28672 0.286704 0.756377 2.63818 1
2 8192 8192 0.0271974 0.225132 8.2777 1
2 10240 8192 0.105459 0.278289 2.63884 1
2 57344 8192 0.560128 1.51171 2.69886 1
2 8192 28672 0.290515 0.740313 2.54827 1
4 8192 8192 0.0341437 0.226175 6.62419 1
4 10240 8192 0.106526 0.278763 2.61686 1
4 57344 8192 0.562239 1.51169 2.6887 1
4 8192 28672 0.296588 0.741644 2.50059 1
8 8192 8192 0.0521865 0.228708 4.38251 1
8 10240 8192 0.108563 0.279899 2.57821 1
8 57344 8192 0.565344 1.51386 2.67776 1
8 8192 28672 0.305728 0.744223 2.43426 1
16 8192 8192 0.0684504 0.228147 3.33302 1
16 10240 8192 0.11321 0.281227 2.48413 1
16 57344 8192 0.571792 1.51852 2.65573 1
16 8192 28672 0.325349 0.748451 2.30046 1
32 8192 8192 0.0970511 0.226781 2.33671 1
32 10240 8192 0.125646 0.293164 2.33325 1
32 57344 8192 0.642472 1.59968 2.48989 1
32 8192 28672 0.368386 0.774405 2.10215 1
64 8192 8192 0.126184 0.230384 1.82577 1
64 10240 8192 0.155499 0.295118 1.89788 1
64 57344 8192 0.776818 1.61018 2.07278 1
64 8192 28672 0.487071 0.826973 1.69785 1
128 8192 8192 0.227225 0.246218 1.08359 1
128 10240 8192 0.281636 0.306252 1.0874 1
128 57344 8192 1.50182 1.69311 1.12737 1
128 8192 28672 0.971188 0.915551 0.942713 1
256 8192 8192 0.501718 0.396104 0.789494 1
256 10240 8192 0.603751 0.495955 0.821456 1
256 57344 8192 3.01183 2.81517 0.934703 1
256 8192 28672 1.85452 1.42665 0.769285 1
512 8192 8192 1.03497 0.793699 0.766883 1
512 10240 8192 1.24271 0.989675 0.796386 1
512 57344 8192 6.01086 5.47631 0.91107 1
512 8192 28672 3.41326 2.72313 0.797807 1

Copy link

pytorch-bot bot commented Jun 3, 2024

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/312

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

✅ No Failures

As of commit e12eb2f with merge base 12f44ab (image):
💚 Looks good so far! There are no failures yet. 💚

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

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jun 3, 2024
@gau-nernst gau-nernst marked this pull request as ready for review June 3, 2024 23:38
@gau-nernst gau-nernst requested a review from msaroufim June 3, 2024 23:39
@msaroufim msaroufim merged commit 8dbf031 into pytorch:main Jun 4, 2024
13 checks passed
@gau-nernst gau-nernst deleted the fix_fp6_bench branch June 4, 2024 12:43
dbyoung18 pushed a commit to dbyoung18/ao that referenced this pull request Jul 31, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants