[BUG]: Unable to use MultiGPU to build Graph #4753
Comments
kellie19
added
? - Needs Triage
|
Hi @kellie19, Terribly sorry to hear about your issue. I'd be happy to help look into this for you. You said you're working with a 24.12 conda environment. What sort of commands did you use before building cugraph-24.12? Were these the steps you took?
Let's try and make sure everything works at the build stage first before looking into the multi-GPU dask example you included. |
|
Hi @kellie19 just following up here. How are things going? |
|
Hi Ralph,
Thanks for the provided information.I used rapids to set up CuGraph via
conda create -n rapids-24.12 -c rapidsai -c conda-forge -c nvidia \
rapids=24.12 python=3.12 'cuda-version>=12.0,<=12.5' not by using
the ./build.sh libcugraph pylibcugraph cugraph because it had error
for not able to find rapids library.
…On Wed, Jan 29, 2025 at 7:15 AM Ralph Liu ***@***.***> wrote:
Hi @kellie19 <https://github.com/kellie19> just following up here. How
are things going?
—
Reply to this email directly, view it on GitHub
<#4753 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AL3W55JWFMN7ZBOWEAGVWTT2NDWCJAVCNFSM6AAAAABRVFF7L6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMRRHEZTOOJZGY>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
Hi @kellie19 I've created a similar environment and am currently looking into the multi-GPU issue. I don't think this is related to the build issues you're seeing AFAIK. (explained in the next thread) What did you use to generate the |
Regarding the Build IssueI believe the conda environment you created using the command you provided conda create -n rapids-24.12 -c rapidsai -c conda-forge -c nvidia rapids=24.12 python=3.12 'cuda-version>=12.0,<=12.5Might not contain all the packages needed to support building from source. It looks like from the logs, there are a rapids CMake files that are missing. If you'd like to build from source, you may need to update your environment using the from source steps (I also listed them below for your case). Checkout branch-24.12 of cugraphgit clone https://github.com/rapidsai/cugraph.git
cd cugraph
git checkout branch-24.12Update Your Conda Environment Using the YAML File# for CUDA 11.x
conda env update --name rapids-24.12 --file conda/environments/all_cuda-118_arch-x86_64.yaml
# for CUDA 12.x
conda env update --name rapids-24.12 --file conda/environments/all_cuda-125_arch-x86_64.yaml
# hit [y]Activate and Buildconda activate rapids-24.12
./build.sh clean # Just in case to reset any CMake variables that have been set before
./build.sh libcugraph pylibcugraph cugraph --skip_cpp_testsGive this a go and let me know if you run into any troubles! |
|
Thank you that is very helpful! #resolve
…On Thu, Jan 30, 2025 at 7:15 AM Ralph Liu ***@***.***> wrote:
Regarding the Build Issue
I believe the conda environment you created using the command you provided
conda create -n rapids-24.12 -c rapidsai -c conda-forge -c nvidia rapids=24.12 python=3.12 'cuda-version>=12.0,<=12.5
Might not contain all the packages needed to support building from source.
It looks like from the logs, there are a rapids CMake files that are
missing. If you'd like to build from source, you may need to update your
environment using the from source steps
<https://github.com/rapidsai/cugraph/blob/branch-24.12/docs/cugraph/source/installation/source_build.md#clone-the-repository>
.
Checkout branch-24.12 of cugraph
git clone https://github.com/rapidsai/cugraph.git
git checkout branch-24.12cd cugraph
Update Your Conda Environment Using the YAML File
# for CUDA 11.x
conda env update --name rapids-24.12 --file conda/environments/all_cuda-118_arch-x86_64.yaml
# for CUDA 12.x
conda env update --name rapids-24.12 --file conda/environments/all_cuda-125_arch-x86_64.yaml
# hit [y]
Activate and Build
conda activate rapids-24.12
./build.sh clean # Just in case to reset any CMake variables that have been set before
./build.sh libcugraph pylibcugraph cugraph --skip_cpp_tests
—
Reply to this email directly, view it on GitHub
<#4753 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AL3W55P2KJIARRPBF6IK5Q32NI6ZDAVCNFSM6AAAAABRVFF7L6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMMRUG43TMNBSGE>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
If you need additional help with your workflow example feel free to reach out again. |
Version
24.12
Which installation method(s) does this occur on?
Conda
Describe the bug.
I downloaded rapids-24.12 via conda and start a conda environment to running Cugraph.
I was able to run single GPU algorithm, build graph and generate graph. However, when I tried to explore multi-gpu usage in cugraph, I receive keyError when I tried to build multi-gpu graph by using
from_dask_cudf_edgelistandcugraph.generators.rmatby setting mg = True and set up multi gpu cluster and clientBesides multi-gpu in cugraph, I was not able to use
cudf.read_csv()with internal error described in https://community.deeplearning.ai/t/has-anyone-gotten-cudf-read-csv-to-work/558425 so I use pandas to convert to cudf for input offrom_dask_cudf_edgelist.I guess maybe some GPU setup was incorrect so I ran
./build.shand shows error as well.Could you please tell me how to fix the multi-gpu error and if that is relates to my build fail?
Thanks
Minimum reproducible example
Relevant log output
Environment details
Other/Misc.
No response
Code of Conduct
The text was updated successfully, but these errors were encountered: