Use pip install cu117 by atalman · Pull Request #85097 · pytorch/pytorch

atalman · 2022-09-15T18:41:06Z

Creates new wheel workflow specific to CUDA 11.7 that does not bundle the cudnn and cublas.

Workflow:
https://github.com/pytorch/pytorch/actions/runs/3094622781

New Package:
manywheel-py3_10-cuda11_7-with-pypi-cudnn | 843 MB

Old Package:
manywheel-py3_10-cuda11_7 | 1.65 GB

Testing workflow:

manywheel-py3_7-cuda11_7-with-pypi-cudnn-build / build:

Bundling without cudnn and cublas.
+ DEPS_LIST=("/usr/local/cuda/lib64/libcudart.so.11.0" "/usr/local/cuda/lib64/libnvToolsExt.so.1" "/usr/local/cuda/lib64/libnvrtc.so.11.2" "/usr/local/cuda/lib64/libnvrtc-builtins.so.11.7" "$LIBGOMP_PATH")
+ DEPS_SONAME=("libcudart.so.11.0" "libnvToolsExt.so.1" "libnvrtc.so.11.2" "libnvrtc-builtins.so.11.7" "libgomp.so.1")
.....
pytorch_extra_install_requirements: nvidia-cuda-runtime-cu11, nvidia-cudnn-cu11, nvidia-cublas-cu11

manywheel-py3_7-cuda11_7-build / build

Bundling with cudnn and cublas.
+ DEPS_LIST=("/usr/local/cuda/lib64/libcudart.so.11.0" "/usr/local/cuda/lib64/libnvToolsExt.so.1" "/usr/local/cuda/lib64/libnvrtc.so.11.2" "/usr/local/cuda/lib64/libnvrtc-builtins.so.11.7" "/usr/local/cuda/lib64/libcudnn_adv_infer.so.8" "/usr/local/cuda/lib64/libcudnn_adv_train.so.8" "/usr/local/cuda/lib64/libcudnn_cnn_infer.so.8" "/usr/local/cuda/lib64/libcudnn_cnn_train.so.8" "/usr/local/cuda/lib64/libcudnn_ops_infer.so.8" "/usr/local/cuda/lib64/libcudnn_ops_train.so.8" "/usr/local/cuda/lib64/libcudnn.so.8" "/usr/local/cuda/lib64/libcublas.so.11" "/usr/local/cuda/lib64/libcublasLt.so.11" "$LIBGOMP_PATH")
+ DEPS_SONAME=("libcudart.so.11.0" "libnvToolsExt.so.1" "libnvrtc.so.11.2" "libnvrtc-builtins.so.11.7" "libcudnn_adv_infer.so.8" "libcudnn_adv_train.so.8" "libcudnn_cnn_infer.so.8" "libcudnn_cnn_train.so.8" "libcudnn_ops_infer.so.8" "libcudnn_ops_train.so.8" "libcudnn.so.8" "libcublas.so.11" "libcublasLt.so.11" "libgomp.so.1")

cc: @malfet @ptrblck

pytorch-bot · 2022-09-15T18:41:08Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/85097

📄 Preview Python docs built from this PR
📄 Preview C++ docs built from this PR
❓ Need help or want to give feedback on the CI? Visit our office hours

Note: Links to docs will display an error until the docs builds have been completed.

❌ 5 Failures

As of commit d60d7e4:

The following jobs have failed:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

setup.py

.github/scripts/generate_binary_build_matrix.py

setup.py

Testing Testing Testing Testing Trying to build whl with pip New workflow Adding build name as environment Using build name to determine how manywheel should be packaged Fix typo Fix typo fix log Adding new environment variable 1 Adding new env variable Extra install req Adding new environment More new var setting Adding binary env populate Test Fix if condition testing Testing Refactor the code Testing Testing Address comments Fix lint issue

Refactoring code Fix builder folder Testing variables Testing

atalman · 2022-09-21T01:56:28Z

.github/scripts/generate_binary_build_matrix.py

+                        "package_type": package_type,
+                        "pytorch_extra_install_requirements":
+                        "nvidia-cuda-runtime-cu11;"
+                        "nvidia-cudnn-cu11==8.5.0.96;"


Please note had to specify full version for now. Partial version did not work.

setup.py

malfet · 2022-09-21T15:06:25Z

setup.py

            extra_link_args += ['-g']

+    # special CUDA 11.7 package that requires installation of cuda runtime, cudnn and cublas
+    pytorch_extra_install_requirements = os.getenv('PYTORCH_EXTRA_INSTALL_REQUIREMENTS', '')


Nit: use double quotes. Also, why do you need default value here?

Suggested change

pytorch_extra_install_requirements = os.getenv('PYTORCH_EXTRA_INSTALL_REQUIREMENTS', '')

pytorch_extra_install_requirements = os.getenv("PYTORCH_EXTRA_INSTALL_REQUIREMENTS",)

Done, converted to double quotes. default values I think good idea to keep in case if environment variable is not set.

atalman · 2022-09-21T16:26:38Z

@pytorchmergebot -f "Already run through all tests"

pytorch-bot · 2022-09-21T16:26:40Z

❌ 🤖 pytorchbot command failed:

@pytorchbot: error: argument command: invalid choice: 'Already run through all tests' (choose from 'merge', 'revert', 'rebase', 'label')

usage: @pytorchbot [-h] {merge,revert,rebase,label} ...

Try @pytorchbot --help for more info.

atalman · 2022-09-21T16:27:49Z

@pytorchbot --help

pytorch-bot · 2022-09-21T16:27:51Z

PyTorchBot Help

usage: @pytorchbot [-h] {merge,revert,rebase,label} ...

In order to invoke the bot on your PR, include a line that starts with
@pytorchbot anywhere in a comment. That line will form the command; no
multi-line commands are allowed. 

Example:
    Some extra context, blah blah, wow this PR looks awesome

    @pytorchbot merge

optional arguments:
  -h, --help            Show this help message and exit.

command:
  {merge,revert,rebase,label}
    merge               Merge a PR
    revert              Revert a PR
    rebase              Rebase a PR
    label               Add label to a PR

Merge

usage: @pytorchbot merge [-g | -f MESSAGE | -l | -r [REBASE]]

Merge an accepted PR, subject to the rules in .github/merge_rules.json.
By default, this will wait for all required checks (lint, pull) to succeed before merging.

optional arguments:
  -g, --green           Merge when all status checks running on the PR pass. To add status checks, use labels like `ciflow/trunk`.
  -f MESSAGE, --force MESSAGE
                        Merge without checking anything. This requires a reason for auditting purpose, for example:
                        @pytorchbot merge -f 'Minor update to fix lint. Expecting all PR tests to pass'
  -l, --land-checks     Merge with land time checks. This will create a new branch with your changes rebased on viable/strict and run a majority of trunk tests _before_ landing to increase trunk reliability and decrease risk of revert. The tests added are: pull, Lint and trunk. Note that periodic is excluded. (EXPERIMENTAL)
  -r [REBASE], --rebase [REBASE]
                        Rebase the PR to re run checks before merging.  It will accept a branch name and will default to master if not specified.

Revert

usage: @pytorchbot revert -m MESSAGE -c
                          {nosignal,ignoredsignal,landrace,weird,ghfirst}

Revert a merged PR. This requires that you are a Meta employee.

Example:
  @pytorchbot revert -m="This is breaking tests on trunk. hud.pytorch.org/" -c=nosignal

optional arguments:
  -m MESSAGE, --message MESSAGE
                        The reason you are reverting, will be put in the commit message. Must be longer than 3 words.
  -c {nosignal,ignoredsignal,landrace,weird,ghfirst}, --classification {nosignal,ignoredsignal,landrace,weird,ghfirst}
                        A machine-friendly classification of the revert reason.

Rebase

usage: @pytorchbot rebase [-s | -b BRANCH]

Rebase a PR. Rebasing defaults to the default branch of pytorch (master).
You, along with any member of the pytorch organization, can rebase your PR.

optional arguments:
  -s, --stable          Rebase to viable/strict
  -b BRANCH, --branch BRANCH
                        Branch you would like to rebase to

Label

usage: @pytorchbot label labels [labels ...]

Adds label to a PR

positional arguments:
  labels  Labels to add to given Pull Request

atalman · 2022-09-21T16:29:00Z

@pytorchbot merge -f "Full set of tests was executed, this one will fail anyway because of wheels 3.11"

pytorchmergebot · 2022-09-21T16:30:21Z

@pytorchbot successfully started a merge job. Check the current status here.
The merge job was triggered with the force (-f) flag. This means your change will be merged immediately, bypassing any CI checks (ETA: 1-5 minutes). If this is not the intended behavior, feel free to use some of the other merge options in the wiki.
Please reach out to the PyTorch DevX Team with feedback or questions!

@malfet

Creates new wheel workflow specific to CUDA 11.7 that does not bundle the cudnn and cublas. Workflow: https://github.com/pytorch/pytorch/actions/runs/3094622781 New Package: manywheel-py3_10-cuda11_7-with-pypi-cudnn | 843 MB Old Package: manywheel-py3_10-cuda11_7 | 1.65 GB Testing workflow: [manywheel-py3_7-cuda11_7-with-pypi-cudnn-build / build](https://github.com/pytorch/pytorch/actions/runs/3091145546/jobs/5000867662#logs): ``` Bundling without cudnn and cublas. + DEPS_LIST=("/usr/local/cuda/lib64/libcudart.so.11.0" "/usr/local/cuda/lib64/libnvToolsExt.so.1" "/usr/local/cuda/lib64/libnvrtc.so.11.2" "/usr/local/cuda/lib64/libnvrtc-builtins.so.11.7" "$LIBGOMP_PATH") + DEPS_SONAME=("libcudart.so.11.0" "libnvToolsExt.so.1" "libnvrtc.so.11.2" "libnvrtc-builtins.so.11.7" "libgomp.so.1") ..... pytorch_extra_install_requirements: nvidia-cuda-runtime-cu11, nvidia-cudnn-cu11, nvidia-cublas-cu11 ``` [manywheel-py3_7-cuda11_7-build / build](https://github.com/pytorch/pytorch/actions/runs/3091145546/jobs/5000863250#logs) ``` Bundling with cudnn and cublas. + DEPS_LIST=("/usr/local/cuda/lib64/libcudart.so.11.0" "/usr/local/cuda/lib64/libnvToolsExt.so.1" "/usr/local/cuda/lib64/libnvrtc.so.11.2" "/usr/local/cuda/lib64/libnvrtc-builtins.so.11.7" "/usr/local/cuda/lib64/libcudnn_adv_infer.so.8" "/usr/local/cuda/lib64/libcudnn_adv_train.so.8" "/usr/local/cuda/lib64/libcudnn_cnn_infer.so.8" "/usr/local/cuda/lib64/libcudnn_cnn_train.so.8" "/usr/local/cuda/lib64/libcudnn_ops_infer.so.8" "/usr/local/cuda/lib64/libcudnn_ops_train.so.8" "/usr/local/cuda/lib64/libcudnn.so.8" "/usr/local/cuda/lib64/libcublas.so.11" "/usr/local/cuda/lib64/libcublasLt.so.11" "$LIBGOMP_PATH") + DEPS_SONAME=("libcudart.so.11.0" "libnvToolsExt.so.1" "libnvrtc.so.11.2" "libnvrtc-builtins.so.11.7" "libcudnn_adv_infer.so.8" "libcudnn_adv_train.so.8" "libcudnn_cnn_infer.so.8" "libcudnn_cnn_train.so.8" "libcudnn_ops_infer.so.8" "libcudnn_ops_train.so.8" "libcudnn.so.8" "libcublas.so.11" "libcublasLt.so.11" "libgomp.so.1") ``` cc: @malfet @ptrblck Pull Request resolved: #85097 Approved by: https://github.com/malfet

Fixes #88049 #85097 added new extra dependencies on `nvidia-*`. They are linux (GPU) only packages, but were not marked as such, causing issues installing pytorch 1.13 via Poetry (and possibly other tools that follow PyPI's metadata API) on non-Linux systems. This "fixes" the issue by adding the `; platform_system = 'Linux'` marker on these dependencies, but the main problem of different metadata for different wheels is a [somewhat larger issue](#88049 (comment)). #85097 used `;` as a delimiter for splitting the different deps, but that is the delimiter used in markers, so I changed to split on `|`. Pull Request resolved: #88826 Approved by: https://github.com/neersighted, https://github.com/lalmei, https://github.com/malfet

…8826) Fixes pytorch#88049 pytorch#85097 added new extra dependencies on `nvidia-*`. They are linux (GPU) only packages, but were not marked as such, causing issues installing pytorch 1.13 via Poetry (and possibly other tools that follow PyPI's metadata API) on non-Linux systems. This "fixes" the issue by adding the `; platform_system = 'Linux'` marker on these dependencies, but the main problem of different metadata for different wheels is a [somewhat larger issue](pytorch#88049 (comment)). pytorch#85097 used `;` as a delimiter for splitting the different deps, but that is the delimiter used in markers, so I changed to split on `|`. Pull Request resolved: pytorch#88826 Approved by: https://github.com/neersighted, https://github.com/lalmei, https://github.com/malfet

…89924) Fixes #88049 #85097 added new extra dependencies on `nvidia-*`. They are linux (GPU) only packages, but were not marked as such, causing issues installing pytorch 1.13 via Poetry (and possibly other tools that follow PyPI's metadata API) on non-Linux systems. This "fixes" the issue by adding the `; platform_system = 'Linux'` marker on these dependencies, but the main problem of different metadata for different wheels is a [somewhat larger issue](#88049 (comment)). #85097 used `;` as a delimiter for splitting the different deps, but that is the delimiter used in markers, so I changed to split on `|`. Pull Request resolved: #88826 Approved by: https://github.com/neersighted, https://github.com/lalmei, https://github.com/malfet Co-authored-by: Jacob Hayes <jacob.r.hayes@gmail.com>

…8826) Fixes pytorch#88049 pytorch#85097 added new extra dependencies on `nvidia-*`. They are linux (GPU) only packages, but were not marked as such, causing issues installing pytorch 1.13 via Poetry (and possibly other tools that follow PyPI's metadata API) on non-Linux systems. This "fixes" the issue by adding the `; platform_system = 'Linux'` marker on these dependencies, but the main problem of different metadata for different wheels is a [somewhat larger issue](pytorch#88049 (comment)). pytorch#85097 used `;` as a delimiter for splitting the different deps, but that is the delimiter used in markers, so I changed to split on `|`. Pull Request resolved: pytorch#88826 Approved by: https://github.com/neersighted, https://github.com/lalmei, https://github.com/malfet

…3066) Like #89924 #91083 #85097 added new extra dependencies on nvidia-*. They are linux x86_64 (GPU) only packages, but were not marked as such, causing issues installing pytorch 1.13 via Poetry (and possibly other tools that follow PyPI's metadata API) on Linux aarch64 systems. This "fixes" the issue by adding the `and platform_machine == 'x86_64'` marker on these dependencies. Pull Request resolved: #93066 Approved by: https://github.com/malfet

atalman requested a review from jeffdaily as a code owner September 15, 2022 18:41

pytorch-bot bot added the topic: not user facing topic category label Sep 15, 2022

facebook-github-bot added the cla signed label Sep 15, 2022

atalman added ciflow/binaries_wheel Trigger binary build and upload jobs for wheel on the PR and removed cla signed labels Sep 15, 2022

atalman force-pushed the use_pip_install_cu117 branch from 4e8a1a4 to 7d92f92 Compare September 15, 2022 18:43

facebook-github-bot added the cla signed label Sep 15, 2022

atalman requested review from malfet and removed request for jeffdaily September 15, 2022 18:43

atalman marked this pull request as draft September 15, 2022 18:44

atalman force-pushed the use_pip_install_cu117 branch from a69724c to 73b212c Compare September 19, 2022 15:55

malfet reviewed Sep 19, 2022

View reviewed changes

setup.py Outdated Show resolved Hide resolved

atalman force-pushed the use_pip_install_cu117 branch from 34962e3 to 2c73691 Compare September 19, 2022 19:52

malfet reviewed Sep 20, 2022

View reviewed changes

.github/scripts/generate_binary_build_matrix.py Outdated Show resolved Hide resolved

setup.py Outdated Show resolved Hide resolved

atalman force-pushed the use_pip_install_cu117 branch from 801a314 to 931bd11 Compare September 20, 2022 19:03

atalman force-pushed the use_pip_install_cu117 branch from 07978d1 to b631a9e Compare September 20, 2022 19:35

Define pytorch version for new package

56b1a0e

Refactoring code Fix builder folder Testing variables Testing

atalman force-pushed the use_pip_install_cu117 branch from 9391c16 to 56b1a0e Compare September 20, 2022 21:59

Specifying full versions

6c51125

atalman changed the title ~~[WIP] Use pip install cu117~~ Use pip install cu117 Sep 21, 2022

atalman marked this pull request as ready for review September 21, 2022 01:55

atalman requested a review from a team as a code owner September 21, 2022 01:55

atalman commented Sep 21, 2022

View reviewed changes

atalman requested a review from malfet September 21, 2022 11:18

malfet approved these changes Sep 21, 2022

View reviewed changes

Addressing comments

d60d7e4

pytorchmergebot added the Merged label Sep 21, 2022

pytorchmergebot closed this in eb94df2 Sep 21, 2022

syed-ahmed mentioned this pull request Sep 22, 2022

Integrates CUDA pip wheels pytorch/builder#1136

Merged

atalman mentioned this pull request Nov 8, 2022

Uses cuda pypi nccl #88125

Closed

JacobHayes mentioned this pull request Nov 10, 2022

Add platform markers for linux only extra_install_requires #88826

Closed

atalman mentioned this pull request Nov 30, 2022

Add platform markers for linux only extra_install_requires (#88826) #89924

Merged

TimeRainStarSky mentioned this pull request Dec 19, 2022

Add platform markers for linux x86_64 only extra_install_requires #91083

Closed

TimeRainStarSky mentioned this pull request Jan 26, 2023

Add platform markers for linux x86_64 only extra_install_requires #93066

Closed

yukw777 mentioned this pull request Feb 22, 2023

Cannot install PyTorch 1.13.x with PDM pdm-project/pdm#1732

Closed

1 task

	pytorch_extra_install_requirements = os.getenv('PYTORCH_EXTRA_INSTALL_REQUIREMENTS', '')
	pytorch_extra_install_requirements = os.getenv("PYTORCH_EXTRA_INSTALL_REQUIREMENTS",)

Conversation

atalman commented Sep 15, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot bot commented Sep 15, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/85097

❌ 5 Failures

Uh oh!

Uh oh!

Uh oh!

Uh oh!

atalman Sep 21, 2022

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

malfet Sep 21, 2022

Choose a reason for hiding this comment

Uh oh!

atalman Sep 21, 2022

Choose a reason for hiding this comment

Uh oh!

atalman commented Sep 21, 2022

Uh oh!

pytorch-bot bot commented Sep 21, 2022

Uh oh!

atalman commented Sep 21, 2022

Uh oh!

pytorch-bot bot commented Sep 21, 2022

PyTorchBot Help

Merge

Revert

Rebase

Label

Uh oh!

atalman commented Sep 21, 2022

Uh oh!

pytorchmergebot commented Sep 21, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

atalman commented Sep 15, 2022 •

edited

Loading

pytorch-bot bot commented Sep 15, 2022 •

edited

Loading