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

Improve the performance of bevformer #224

Merged
merged 29 commits into from
Nov 23, 2022
Merged

Improve the performance of bevformer #224

merged 29 commits into from
Nov 23, 2022

Conversation

Cathy0908
Copy link
Collaborator

@Cathy0908 Cathy0908 commented Nov 2, 2022

训练性能:

单机单卡,fp16:

Official吞吐 EasyCV吞吐 性能提升 环境
bevformer-base 0.727 0.8 10% A100, batch_size=5, num_worker=8
bevformer-tiny 3.55 9.84 177% A100, batch_size=32, num_worker=8

推理性能

fp32 (精度已对齐)

性能提升:42%

Name Backend Median(FPS) Mean(FPS) Median(ms) Mean(ms) 99th_p std_dev
0 easycv TensorRT 3.68697 3.68651 0.271226 0.271259 0.272165 0.000197132
1 easycv script TensorRT 3.8131 3.79859 0.262254 0.26337 0.295066 0.00577846
2 blade TensorRT 5.40248 5.23383 0.1851 0.192212 0.261654 0.0163032

@Cathy0908 Cathy0908 changed the title Optimize bevformer train speed [WIP] Optimize bevformer train speed Nov 2, 2022
@Cathy0908 Cathy0908 changed the title [WIP] Optimize bevformer train speed [WIP] Optimize bevformer train speed and inferrence speed Nov 18, 2022
@Cathy0908 Cathy0908 changed the title [WIP] Optimize bevformer train speed and inferrence speed Optimize bevformer train speed and inferrence speed Nov 21, 2022
@Cathy0908 Cathy0908 changed the title Optimize bevformer train speed and inferrence speed Improve the performance of bevformer Nov 23, 2022
@Cathy0908 Cathy0908 merged commit f8c9a9a into alibaba:master Nov 23, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants