-
Notifications
You must be signed in to change notification settings - Fork 304
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
[BUG]: build libcugraph/pylibcugraph error #4056
Comments
Hi @leo4183 , sorry for the delay in responding. I suspect some of the missing files @robertmaynard may be able to provide more information about that as well as the other observations you made above. |
@vyasr Have you seen these issues before with wheel builds? Are our wheels builds ever done on non Debain/Ubuntu machines? |
We build wheels on Rockylinux 8, but that's not materially different here I don't think. I'm a bit confused by some of the output above. Are you running build.sh, and then running the pip install command? If so, is the build.sh command completing successfully? You say that libcugraph didn't build successfully for you in a couple of scenarios, is the output of the above what you get if libcugraph does build but then you try to build pylibcugraph with pip? I notice that you have a number of nvidia wheels installed. Do you also have the CTK installed on your system, or just the compiler? Your |
@leo4183 I'll close this issue, but please re-open or file a new issue if you're still having problems. |
sorry for the confusing. I tried both build.sh and pip build. both returned same error mentioned above. I do have a customized LD_LIBRARY_PATH which points to some pip build cudnn lib. |
Version
23.12.00
Which installation method(s) does this occur on?
Pip, Source
Describe the bug.
cannot build
libcugraph
/pylibcugraph
from source (nor pip)several observations:
libcugraph
from source simply returns "ptxas error: Value of threads per SM for entry xxxxxx is out of range" (legacy release such as 22.12.00 would mis-identify sm_8x cards as sm_5x)/usr/include/gdeflate
. this might be a bug on Redhat/Centos/Rocky machine given that dnf/yum wont havenvcomp
installed (repository only containsnvidia-driver
,cuda-1x-x
and etc)nvcomp
(includinggdeflate
) to the system dev environment, thegdeflate
missing complain was gone. however thelibcugraph
building process doesnt seem to successfully have those dependencies installed (check the log output below. basically the cmake configure file couldnt locate those dependencies.)ps: no
$CONDA_PREFIX
,$PREFIX
,$CUGRAPH_HOME
or any related env parameters defined under the user sessionMinimum reproducible example
Relevant log output
Environment details
Other/Misc.
No response
Code of Conduct
The text was updated successfully, but these errors were encountered: