-
Notifications
You must be signed in to change notification settings - Fork 114
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
Compile error on recent Arch #533
Comments
It's difficult to solve from the log, full compile log might help. Which version of gcc and CUDA have you used? I've tested that NVEnc did compile fine with gcc 12.3 + CUDA 12.3, but not with gcc 13, as CUDA 12.3 support is up to gcc12. |
Thanks for your answer, I'll send log and system details when I'm back at the machine. |
gcc (GCC) 13.2.1 20230801 I checked that I correctly build with gcc-12 I think we can close this issue, sorry for the noise. For completeness, here the build log error with gcc 13.2:
|
Nice to hear it works fine with gcc 12. I'll try adding gcc13 support when future CUDA version supports it. |
Unfortunately, for now I have to put nvenc updates in Arch on ice: the current nvenc requires highway/libhwy via ffmpeg, and that is now compiled with gcc13 on Arch, and when compiling nvenc with gcc12 the link stage fails. Do you have any idea when a gcc13 compatible cuda comes out? |
I've understood the situation.
I think that only NVIDIA knows, but might be possible after GTC 2024 on Mar19-22(JST), next CUDA 12.4 might come out. |
Thanks! |
CUDA 12.4 seems to come out. Although I haven't tested myself as I'm testing on Ubuntu 22.04, the document states that it supports up to gcc13.2. |
Thanks @rigaya
|
Ahh, passing Compiling now works, bug gives lots of warnings
Final link step also worked besides a bunch of errors, and compiled binary at least works for Thanks |
Thank you for testing, and also sharing to set the macro. I’ll change to add the macro by default in the next version. I’ve now also tested with Ubuntu 22.04 + CUDA12.4 + gcc13.2 and was able to build successfully. I didn’t get the warnings you’ve had, so I’m not sure of the warnings you’ve had. Anyway, nice to hear compile works with CUDA 12.4. |
Thanks for testing. I saw in your last release the following change
but that only duplicated an already present flag, and still fails in my case. The reason is that these
which shows that the check in configure fails. Without the check the compilation would already succeed, because the |
Thank you, that was a wrong commit. 396ca3f shall work without --extra-cxxflags. |
Hi
thanks for your work on nvencc! I am the maintainer of nvenc in Arch.
A recent change (my guess is the update of CUDA) breaks compiling nvencc:
Do you have any suggestion to resolved this?
Thanks
The text was updated successfully, but these errors were encountered: