Skip to content

CUDA out of memory with batchsize = 1 in one 1080Ti #34

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.

Sign up for GitHub

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

Jump to bottom
Closed
jmt330 opened this issue Oct 13, 2018 · 5 comments
Closed

CUDA out of memory with batchsize = 1 in one 1080Ti #34

jmt330 opened this issue Oct 13, 2018 · 5 comments

Comments

@jmt330
Copy link

jmt330 commented Oct 13, 2018

No description provided.

@jmt330 jmt330 changed the title CUDA out of memory with batchsize = 1 in one Titan X CUDA out of memory with batchsize = 1 in one 1080Ti Oct 13, 2018
@jmt330
Copy link
Author

jmt330 commented Oct 14, 2018

@heilaw

@heilaw
Copy link
Collaborator

heilaw commented Oct 14, 2018

I also tried running CornerNet on a 1080 Ti. I was able to train the network with a batch size of 3. Can you give more details about your settings?

@jmt330
Copy link
Author

jmt330 commented Oct 16, 2018

I also tried running CornerNet on a 1080 Ti. I was able to train the network with a batch size of 3. Can you give more details about your settings?

I also tried running CornerNet on a 1080 Ti. I was able to train the network with a batch size of 3. Can you give more details about your settings?

Thank you for your reply, and we finally find it's due to our Arch system made some compile errors to pyTorch(didn't open the cudnn). After we recompile the pytorch, it works!

@heilaw heilaw closed this as completed Oct 16, 2018
@njustymk njustymk mentioned this issue Apr 3, 2019
Closed
@mrlaiii mrlaiii mentioned this issue Apr 13, 2019
Closed
@AlphaGoMK
Copy link

I also tried running CornerNet on a 1080 Ti. I was able to train the network with a batch size of 3. Can you give more details about your settings?

I also tried running CornerNet on a 1080 Ti. I was able to train the network with a batch size of 3. Can you give more details about your settings?

Thank you for your reply, and we finally find it's due to our Arch system made some compile errors to pyTorch(didn't open the cudnn). After we recompile the pytorch, it works!

I got the same question with you, but I correctly installed cudnn and it works fine in other code.
Would you like to offer more details about you solution? Did you just reinstall cudnn?
Thanks in advance :)

@GalyaZalesskaya GalyaZalesskaya mentioned this issue Nov 22, 2019
Open
@ltwwwww
Copy link

ltwwwww commented Jan 10, 2021

I also tried running CornerNet on a 1080 Ti. I was able to train the network with a batch size of 3. Can you give more details about your settings?

I also tried running CornerNet on a 1080 Ti. I was able to train the network with a batch size of 3. Can you give more details about your settings?

Thank you for your reply, and we finally find it's due to our Arch system made some compile errors to pyTorch(didn't open the cudnn). After we recompile the pytorch, it works!

I got the same question with you, but I correctly installed cudnn and it works fine in other code.
Would you like to offer more details about you solution? Did you just reinstall cudnn?
Thanks in advance :)

Have you solve the problem? I also installed cudnn correctly but it doesn't work and still out of memory with the batchsize = 1. Why??? Thank you!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
4 participants
@heilaw @AlphaGoMK @jmt330 @ltwwwww