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

Raise "NVRTC version too high for GPU (compute_7)" on RTX2060 when "Sort_By_Key" #25

Open
LCM1999 opened this issue Jun 16, 2022 · 2 comments

Comments

@LCM1999
Copy link

LCM1999 commented Jun 16, 2022

My code:
image
Error happens in "trtc.Sort_By_Key"
Std::out:
image
My GPU information:
RTX 2060 12GB; CUDA version: 11.6.124; Compute Capability: (7, 5): 64
My ThrustRTC version: 0.3.19

What other information should I provide?
Thanks for you help.

@LCM1999 LCM1999 closed this as completed Jun 17, 2022
@LCM1999 LCM1999 reopened this Jun 17, 2022
@LCM1999
Copy link
Author

LCM1999 commented Jun 17, 2022

I found this could happen when I use both ThrustRTC and PyCuda.
In my code, 'Hash List' and 'Index List' are computed by pyCuda, and when I use ThrustRTC to sort them, I got this error.
But if I only use ThrustRTC to sort same arrays, it works well.
So i wonder if there is some conflict between ThrustRTC and pyCuda?

@fynv
Copy link
Owner

fynv commented Aug 13, 2022

Sorry for late reply. This was a bug.

ThrustRTC simply calculated an incorrect compute capability when CUDA is initialized by a different library.
(should be compute_75 instead of compute_7)

I fixed this just now in version 0.3.20.

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