-
-
Notifications
You must be signed in to change notification settings - Fork 25
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
Rebuild for CUDA 12 w/arch + Windows support #227
Rebuild for CUDA 12 w/arch + Windows support #227
Conversation
The transition to CUDA 12 SDK includes new packages for all CUDA libraries and build tools. Notably, the cudatoolkit package no longer exists, and packages should depend directly on the specific CUDA libraries (libcublas, libcusolver, etc) as needed. For an in-depth overview of the changes and to report problems [see this issue]( conda-forge/conda-forge.github.io#1963 ). Please feel free to raise any issues encountered there. Thank you! 🙏
…nda-forge-pinning 2024.02.13.22.01.17
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
- cudnn # [cuda_compiler_version != "None"] | ||
- nccl # [cuda_compiler_version != "None"] | ||
- cuda-cupti-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- cuda-cudart-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- cuda-nvml-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- cuda-nvtx-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- libcublas-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- libcusolver-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- libcurand-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- libcufft-dev # [(cuda_compiler_version or "").startswith("12")] | ||
- libcusparse-dev # [(cuda_compiler_version or "").startswith("12")] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Does anyone know if any of these has the ptxas
binary in it?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
in the nvidia channel, ptxas
used to be in cuda-nvcc
. Either ptxas or nvlink is needed for jaxlib to actually function ...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess it is still in cuda-nvcc
in conda-forge (see https://github.com/conda-forge/cuda-nvcc-feedstock), via some transitive dependency that ends in https://conda-metadata-app.streamlit.app/?q=conda-forge%2Fnoarch%2Fcuda-nvcc-impl_linux-64-12.0.76-h59595ed_2.conda .
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Well, alright, that's good to know, but also note I think we only have cuda-nvcc for 12+, meaning one would need to get cuda-nvcc from the nvda channel. Surprised not many people are complaining...
if [[ ${cuda_compiler_version} == 11.8 ]]; then | ||
export TF_CUDA_COMPUTE_CAPABILITIES=sm_35,sm_50,sm_60,sm_62,sm_70,sm_72,sm_75,sm_80,sm_86,sm_87,sm_89,sm_90,compute_90 | ||
export TF_CUDA_PATHS="${CUDA_HOME},${PREFIX}" | ||
elif [[ ${cuda_compiler_version} == 12* ]]; then | ||
export TF_CUDA_COMPUTE_CAPABILITIES=sm_60,sm_70,sm_75,sm_80,sm_86,sm_89,sm_90,compute_90 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
should we reduce the target arches here to save a little of time? it likely won't be enough though ...
@conda-forge/jaxlib, shall I set up a bigger runner for this repo using https://github.com/Quansight/open-gpu-server/ ? |
Yes, that would be interesting. It probably doesn't solve all the issues as also |
…nda-forge-pinning 2024.03.13.18.01.29
Hi! This is the friendly automated conda-forge-linting service. I was trying to look for recipes to lint for you, but it appears we have a merge conflict. Please ping the 'conda-forge/core' team (using the @ notation in a comment) if you believe this is a bug. |
…into rebuild-cuda120-0-3_hd740a2
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( I do have some suggestions for making it better though... For recipe:
This was used for disabling testing for cross-compiling.
|
This should be good to go. I can resolve the conflicts and merge. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Go for it!
Hi! This is the friendly automated conda-forge-linting service. I just wanted to let you know that I linted all conda-recipes in your PR ( |
We should add cuda-nvcc for cuda120 (can address in a follow-up pr; jaxlib cuda isn't functional without ptxas which comes from cuda-nvcc) |
See also https://github.com/google/jax/blob/aaeeaf5f0caa497d2f6e33d995cdd88a07ee523a/jaxlib/setup.py#L81-L88 for a reference for this runtime dependency. |
This PR has been triggered in an effort to update cuda120.
Notes and instructions for merging this PR:
Please note that if you close this PR we presume that the feedstock has been rebuilt, so if you are going to perform the rebuild yourself don't close this PR until the your rebuild has been merged.
Here are some more details about this specific migrator:
If this PR was opened in error or needs to be updated please add the
bot-rerun
label to this PR. The bot will close this PR and schedule another one. If you do not have permissions to add this label, you can use the phrase@conda-forge-admin, please rerun bot
in a PR comment to have theconda-forge-admin
add it for you.This PR was created by the regro-cf-autotick-bot. The regro-cf-autotick-bot is a service to automatically track the dependency graph, migrate packages, and propose package version updates for conda-forge. Feel free to drop us a line if there are any issues! This PR was generated by https://github.com/regro/cf-scripts/actions/runs/7893793235, please use this URL for debugging.