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

Why the speed is very slow? #180

Open
dimuthuanuraj opened this issue Feb 18, 2024 · 3 comments
Open

Why the speed is very slow? #180

dimuthuanuraj opened this issue Feb 18, 2024 · 3 comments

Comments

@dimuthuanuraj
Copy link

I have experiencing several time speed is not going more than, 100Hz, I am using NvidiaA10 GPU with server. Could anyone please help me to figure this out?

Processing 1900 of 1076800:Loss 13.631720 TEER/TAcc 0.000% - 87.06 Hz

@mcflyhu
Copy link

mcflyhu commented Mar 18, 2024

i have encountered the same problem bro. Have you solved this issue? I nearly spent 2 hours for training one epoch on Voxceleb2 training set. And my gpu usage was utilized for one second while training single batch. I tried to debug to find out which it wastes too much time on reading/loading batch data. dont know if reading wav file stuck I/O channel or something bring the slow speed.
Plus i use RTX3090 for training.

@dimuthuanuraj
Copy link
Author

@mcflyhu Hi have now much improved the speed as follows,

Processing 14700 of 1076800:Loss 13.115565 TEER/TAcc 0.000% - 228.81 Hz

I made modifications to enable distributed training, and I noticed the previous port was used by some other process also. So I have changed the --port and --distributed', dest='distributed', action='store_false'

Note - I am using NVIDIA A10 GPU and T4 GPUS.

@mcflyhu
Copy link

mcflyhu commented Mar 18, 2024

@dimuthuanuraj Thanks for your immediate reply,
i will have a try on ditributed training as suggested setting to acclerate training process, really appreciate for your help.

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

No branches or pull requests

2 participants