Hello everyone,
I hope this email finds everyone well. We are currently packaging PyTorch for Fedora and we are actively packaging the dependencies. We’re reaching out to see if anyone is interested in joining the project. If you’re interested in more information, our meeting notes are in a GitHub repo and we have the AI/ML SIG discussion page where we’ve hosted all of our post meeting discussions. We have a WikiPage for our package assignment list if you would like to see what needs packaging or just to see the status of the project. Our next meeting is tomorrow, October 11th at 9AM EST and it’s on the FedoCal under the SIG calendar. All lists are found below. If you’re interested in joining, have any questions, or want to get a calendar invite for our reoccurring meeting, please feel free to email me at kabdo@redhat.com or Teng at tema@redhat.com.
AI/ML SIG Discussion Page: https://discussion.fedoraproject.org/tag/ai-ml-sig Meeting Notes: https://github.com/kaitlynabdo/pytorch-fedora-meeting-notes Packaging Assignment List: https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus FedoCal: https://calendar.fedoraproject.org/SIGs/
Thank you! Kaitlyn Abdo
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
Jun
On Wed, Oct 11, 2023 at 11:55 PM Kaitlyn Abdo kabdo@redhat.com wrote:
Hello everyone,
I hope this email finds everyone well. We are currently packaging PyTorch for Fedora and we are actively packaging the dependencies. We’re reaching out to see if anyone is interested in joining the project. If you’re interested in more information, our meeting notes are in a GitHub repo and we have the AI/ML SIG discussion page where we’ve hosted all of our post meeting discussions. We have a WikiPage for our package assignment list if you would like to see what needs packaging or just to see the status of the project. Our next meeting is tomorrow, October 11th at 9AM EST and it’s on the FedoCal under the SIG calendar. All lists are found below. If you’re interested in joining, have any questions, or want to get a calendar invite for our reoccurring meeting, please feel free to email me at kabdo@redhat.com or Teng at tema@redhat.com.
AI/ML SIG Discussion Page: https://discussion.fedoraproject.org/tag/ai-ml-sig Meeting Notes: https://github.com/kaitlynabdo/pytorch-fedora-meeting-notes Packaging Assignment List: https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus FedoCal: https://calendar.fedoraproject.org/SIGs/
Thank you! Kaitlyn Abdo _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Friday, December 8, 2023 12:41:59 AM EST Jun Aruga (he / him) wrote:
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
Yes, this is nice that we have pytorch in Fedora. Looking at the specfile...
USE_CUDA=OFF USE_ROCM=OFF
Which does not align with:
%description PyTorch is a Python package that provides two high-level features: * Tensor computation (like NumPy) with strong GPU acceleration
GPU acceleration? Also,
USE_OPENMP=OFF
So, no threading? What about at least enabling BLAS? Maybe it is by default. Not seeing it in the specfile. Without a CUDA version of this, it can't be used the way it was meant to be. We still need to use pip install to get an accelerated version:
pip install torch python3
import torch torch.__config__.show()
The config listed there should be compared with the config in the spec file to get as close to the expected feature set as possible so that people can just switch. This is a positive step and I would love to switch one day.
Best Regards, -Steve
On Friday, 08 December 2023 at 16:34, Steve Grubb wrote:
On Friday, December 8, 2023 12:41:59 AM EST Jun Aruga (he / him) wrote:
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
Yes, this is nice that we have pytorch in Fedora. Looking at the specfile...
USE_CUDA=OFF USE_ROCM=OFF
Which does not align with:
%description PyTorch is a Python package that provides two high-level features:
- Tensor computation (like NumPy) with strong GPU acceleration
GPU acceleration?
Indeed. We have ROCM in Fedora already: $ sudo dnf list rocm*devel Last metadata expiration check: 0:00:24 ago on Sat 09 Dec 2023 22:42:19. Available Packages rocm-comgr-devel.x86_64 16.1-2.fc38 updates rocm-hip-devel.x86_64 5.5.1-10.fc38 updates rocm-opencl-devel.x86_64 5.5.1-10.fc38 updates rocm-runtime-devel.x86_64 5.5.0-2.fc38 updates rocm-smi-devel.x86_64 5.5.1-3.fc38 updates
At least the OpenCL part works quite well for me.
[...]
pip install torch python3
import torch torch.__config__.show()
The config listed there should be compared with the config in the spec file to get as close to the expected feature set as possible so that people can just switch. This is a positive step and I would love to switch one day.
Good idea.
Regards, Dominik
On 12/9/23 14:47, Dominik 'Rathann' Mierzejewski wrote:
On Friday, 08 December 2023 at 16:34, Steve Grubb wrote:
On Friday, December 8, 2023 12:41:59 AM EST Jun Aruga (he / him) wrote:
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
Yes, this is nice that we have pytorch in Fedora. Looking at the specfile...
USE_CUDA=OFF USE_ROCM=OFF
Which does not align with:
%description PyTorch is a Python package that provides two high-level features:
- Tensor computation (like NumPy) with strong GPU acceleration
GPU acceleration?
Indeed. We have ROCM in Fedora already: $ sudo dnf list rocm*devel Last metadata expiration check: 0:00:24 ago on Sat 09 Dec 2023 22:42:19. Available Packages rocm-comgr-devel.x86_64 16.1-2.fc38 updates rocm-hip-devel.x86_64 5.5.1-10.fc38 updates rocm-opencl-devel.x86_64 5.5.1-10.fc38 updates rocm-runtime-devel.x86_64 5.5.0-2.fc38 updates rocm-smi-devel.x86_64 5.5.1-3.fc38 updates
At least the OpenCL part works quite well for me.
We have made a lot of progress towards getting ROCm packaged in Fedora but we're not quite done and I know there are a few yet-unpackaged ROCm components which are required before ROCm support can be enabled for pytorch.
The current status list of ROCm packages is in the wiki: https://fedoraproject.org/wiki/SIGs/HC#Package_status
The current plan is to have ROCm 6.0.0 in F40 before branch, assuming that the release dates line up. As far as I know, AMD has not given an exact release date for 6.0 but ROCm 6.0 was announced last week and it seems likely to release well before F40 branch.
Tim
[...]
pip install torch python3
import torch torch.__config__.show()
The config listed there should be compared with the config in the spec file to get as close to the expected feature set as possible so that people can just switch. This is a positive step and I would love to switch one day.
Good idea.
Regards, Dominik
On 12/8/23 08:34, Steve Grubb wrote:
On Friday, December 8, 2023 12:41:59 AM EST Jun Aruga (he / him) wrote:
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
Yes, this is nice that we have pytorch in Fedora. Looking at the specfile...
USE_CUDA=OFF USE_ROCM=OFF
Which does not align with:
%description PyTorch is a Python package that provides two high-level features:
- Tensor computation (like NumPy) with strong GPU acceleration
GPU acceleration? Also,
GPU acceleration is not enabled for the pytorch packages and that is intentional, for now. pytorch has a mess of third party dependencies which are managed upstream using git subrepos that point to external dependencies that may or may not be easy to package for Fedora. From the beginning, our plan has been to get pytorch packaged for CPU only first and add accelerator support as we can. Perhaps the description for pytorch needs to be changed but our intent is to enable ROCm support for F40.
I don't have the exact list of packages remaining before we can enable ROCm support for pytorch in front of me but I believe that we're down into the single digits and the biggest hurdle at the moment is ROCm's miopen due to some incompatibility with Fedora's llvm or hipcc.
USE_OPENMP=OFF
So, no threading? What about at least enabling BLAS? Maybe it is by default. Not seeing it in the specfile. Without a CUDA version of this, it can't be used the way it was meant to be. We still need to use pip install to get an accelerated version:
I'm not familiar with OpenMP or what might be required there, Tom (cc'd) would know more on that exact detail.
I doubt that a CUDA version of pytorch will ever be packagable for the Fedora repos - the licensing on CUDA would have to change before that happens and while it's possible, it doesn't seem likely in the foreseeable future.
It would be great to enable support for Intel accelerators but that is a different project for a different day. ROCm is the only accelerator support that we have scoped out at this point.
pip install torch python3
import torch torch.__config__.show()
The config listed there should be compared with the config in the spec file to get as close to the expected feature set as possible so that people can just switch. This is a positive step and I would love to switch one day.
In general, there are two reasons why a torch feature is not enabled in the Fedora package:
1. The license of a dependency for that feature is incompatible with Fedora 2. One or more dependencies are not yet packaged for Fedora
Obviously, features that fall into (1) are very difficult, if not impossible for us to work around. Features that fall into (2) will likely need more time - the first build for PyTorch was about a week ago and we still have work to do.
We are working to get the pytorch packages in Fedora to be as complete as we can make them. If anyone is interested in helping, please join us on discourse (#ai-ml-sig) or Matrix (#ai-ml:fedoraproject.org).
Tim
Best Regards, -Steve
-- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Monday, December 11, 2023 4:59:45 AM EST Tim Flink wrote:
On 12/8/23 08:34, Steve Grubb wrote:
On Friday, December 8, 2023 12:41:59 AM EST Jun Aruga (he / him) wrote:
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
Yes, this is nice that we have pytorch in Fedora. Looking at the specfile...
USE_CUDA=OFF USE_ROCM=OFF
Which does not align with:
%description PyTorch is a Python package that provides two high-level features:
- Tensor computation (like NumPy) with strong GPU acceleration
GPU acceleration? Also,
GPU acceleration is not enabled for the pytorch packages and that is intentional, for now. pytorch has a mess of third party dependencies which are managed upstream using git subrepos that point to external dependencies that may or may not be easy to package for Fedora.
Yes, I am familiar with the orignal LUA version which I had to build locally.
From the beginning, our plan has been to get pytorch packaged for CPU only first and add accelerator support as we can. Perhaps the description for pytorch needs to be changed but our intent is to enable ROCm support for F40.
If you are doing CPU only, you really should enable a BLAS backend. Fedora has flexiblas available.
I don't have the exact list of packages remaining before we can enable ROCm support for pytorch in front of me but I believe that we're down into the single digits and the biggest hurdle at the moment is ROCm's miopen due to some incompatibility with Fedora's llvm or hipcc.
USE_OPENMP=OFF
So, no threading? What about at least enabling BLAS? Maybe it is by default. Not seeing it in the specfile. Without a CUDA version of this, it can't be used the way it was meant to be. We still need to use pip install to get an accelerated version:
I'm not familiar with OpenMP or what might be required there, Tom (cc'd) would know more on that exact detail.
GCC should natively support it - unless it uses something brand new GCC hasn't adopted yet.
I doubt that a CUDA version of pytorch will ever be packagable for the Fedora repos - the licensing on CUDA would have to change before that happens and while it's possible, it doesn't seem likely in the foreseeable future.
It would be great to enable support for Intel accelerators but that is a different project for a different day. ROCm is the only accelerator support that we have scoped out at this point.
pip install torch python3
import torch torch.__config__.show()
The config listed there should be compared with the config in the spec file to get as close to the expected feature set as possible so that people can just switch. This is a positive step and I would love to switch one day.
In general, there are two reasons why a torch feature is not enabled in the Fedora package:
- The license of a dependency for that feature is incompatible with
Fedora 2. One or more dependencies are not yet packaged for Fedora
I think you can add OpenMP and BLAS support easily. That should be a small win.
-Steve
Obviously, features that fall into (1) are very difficult, if not impossible for us to work around. Features that fall into (2) will likely need more time - the first build for PyTorch was about a week ago and we still have work to do.
We are working to get the pytorch packages in Fedora to be as complete as we can make them. If anyone is interested in helping, please join us on discourse (#ai-ml-sig) or Matrix (#ai-ml:fedoraproject.org).
Tim
Best Regards, -Steve
-- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List
Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject. org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
-- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.or g Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On 12/8/23 7:34 AM, Steve Grubb wrote:
On Friday, December 8, 2023 12:41:59 AM EST Jun Aruga (he / him) wrote:
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
Yes, this is nice that we have pytorch in Fedora. Looking at the specfile...
USE_CUDA=OFF USE_ROCM=OFF
Which does not align with:
%description PyTorch is a Python package that provides two high-level features:
- Tensor computation (like NumPy) with strong GPU acceleration
GPU acceleration? Also,
USE_OPENMP=OFF
So, no threading? What about at least enabling BLAS? Maybe it is by default. Not seeing it in the specfile. Without a CUDA version of this, it can't be used the way it was meant to be. We still need to use pip install to get an accelerated version:
pip install torch python3
I made the choice to focus on doing cpu first believing it would be a good base to build out from.
To be useful, PyTorch needs to be accelerated, but acceleration is complicated. The acceleration stack as well as the hw drivers must be in Fedora. This work is tracked on the HC sig page. Unfortunately CUDA is not open source. But ROCm is. There is a parallel effort as Tim has said to enable the ROCm packages that PyTorch needs. Most of them are in Fedora, but we are missing a couple of them.
Tom
import torch torch.__config__.show()
The config listed there should be compared with the config in the spec file to get as close to the expected feature set as possible so that people can just switch. This is a positive step and I would love to switch one day.
Best Regards, -Steve
-- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On 12/7/23 22:41, Jun Aruga (he / him) wrote:
Congratulations for the PyTorch package! https://src.fedoraproject.org/rpms/python-torch
I hope someone will announce this great achievement to the Fedora community too, and update the following page too. https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus
PyTorch packages in Fedora are still a work in progress - torchvision and torchdata are still in review, for example. We are definitely planning to announce it once we've made more progress and are also planning to make a F40 change proposal for PyTorch.
I didn't realize that the dependency packaging list was out of date but I'll review it this week and make sure it has been updated.
Tim
Jun
On Wed, Oct 11, 2023 at 11:55 PM Kaitlyn Abdo kabdo@redhat.com wrote:
Hello everyone,
I hope this email finds everyone well. We are currently packaging PyTorch for Fedora and we are actively packaging the dependencies. We’re reaching out to see if anyone is interested in joining the project. If you’re interested in more information, our meeting notes are in a GitHub repo and we have the AI/ML SIG discussion page where we’ve hosted all of our post meeting discussions. We have a WikiPage for our package assignment list if you would like to see what needs packaging or just to see the status of the project. Our next meeting is tomorrow, October 11th at 9AM EST and it’s on the FedoCal under the SIG calendar. All lists are found below. If you’re interested in joining, have any questions, or want to get a calendar invite for our reoccurring meeting, please feel free to email me at kabdo@redhat.com or Teng at tema@redhat.com.
AI/ML SIG Discussion Page: https://discussion.fedoraproject.org/tag/ai-ml-sig Meeting Notes: https://github.com/kaitlynabdo/pytorch-fedora-meeting-notes Packaging Assignment List: https://fedoraproject.org/wiki/SIGs/PyTorch/packagingStatus FedoCal: https://calendar.fedoraproject.org/SIGs/
Thank you! Kaitlyn Abdo _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
-
Dominik 'Rathann' Mierzejewski -
Jun Aruga (he / him) -
Kaitlyn Abdo -
Steve Grubb -
Tim Flink -
Tom Rix