PyTorch on ARC - Revision history

Dmitri: /* Testing */

2024-03-07T22:52:54Z

Testing

← Older revision		Revision as of 22:52, 7 March 2024
Line 66:		Line 66:
	You can test with the <code>torch-gpu-test.py</code> script shown below.		You can test with the <code>torch-gpu-test.py</code> script shown below.
	Copy and paste the text into a file and run if from the command line:		Copy and paste the text into a file and run if from the command line:
	(pytorch) $ python torch-gpu-test.py		(pytorch) $ python torch-gpu-test.py

	If you try this on the '''login node''', it should tell you that '''GPUs are not available'''.		If you try this on the '''login node''', it should tell you that '''GPUs are not available'''.

Dmitri: /* Installing PyTorch */

2024-03-07T22:51:54Z

Installing PyTorch

← Older revision		Revision as of 22:51, 7 March 2024
Line 53:		Line 53:
	Once it is done, your '''PyTorch''' environment is ready.		Once it is done, your '''PyTorch''' environment is ready.

	~~You can test with the <code>torch-gpu-test.py</code> script shown below.~~
	~~Copy and paste the text into a file and run if from the command line:~~
	~~(pytorch) $ python torch-gpu-test.py~~

	~~If you try~~ this ~~on the '''login node''', it should tell you that '''GPUs are not available'''.~~		At this point you can add more packages to the environment using '''conda'''.
	~~It is normal, as the login node does not have any. You will need a GPU node to test the GPUs.~~

	~~Once you know that your '''pytorch''' environment is working properly,~~ you can add more packages to the environment using '''conda'''.
	<pre>		<pre>
	(pytorch) $ CONDA_OVERRIDE_CUDA="12.2" conda install -c conda-forge torchvision cuda-nvcc		(pytorch) $ CONDA_OVERRIDE_CUDA="12.2" conda install -c conda-forge torchvision cuda-nvcc
Line 68:		Line 62:
	'''Note''' that you will have to use the '''override''' variable again, if the additional packages depend on the '''GPU presence'''.		'''Note''' that you will have to use the '''override''' variable again, if the additional packages depend on the '''GPU presence'''.

			== Testing ==

			You can test with the <code>torch-gpu-test.py</code> script shown below.
			Copy and paste the text into a file and run if from the command line:
			(pytorch) $ python torch-gpu-test.py

			If you try this on the '''login node''', it should tell you that '''GPUs are not available'''.
			It is normal, as the login node does not have any. You will need a GPU node to test the GPUs.

	To deactivate the environment (and conda itself) the		To deactivate the environment (and conda itself) the

Dmitri: /* Background */

2024-03-07T22:36:18Z

Background

← Older revision		Revision as of 22:36, 7 March 2024
Line 1:		Line 1:
	= Background =		= Background =

			* https://github.com/pytorch/pytorch - PyTorch project page on GitHub.
			* https://docs.computecanada.ca/wiki/PyTorch - Compute Canada article is not directly applicable on ARC but contains a lot of good information.

	'''PyTorch''' is a Python package that provides two high-level features:		'''PyTorch''' is a Python package that provides two high-level features:

Dmitri: /* See also */

2024-03-07T22:36:05Z

Dmitri: /* See also */

2024-03-07T22:35:32Z

Dmitri: /* Checkpointing */

2024-03-07T22:34:47Z

Checkpointing

← Older revision		Revision as of 22:34, 7 March 2024
Line 155:		Line 155:
	</syntaxhighlight>		</syntaxhighlight>

	=== Checkpointing ===		== Checkpointing ==
	Refer to the checkpointing tutorial at https://pytorch.org/tutorials/recipes/recipes/saving_and_loading_a_general_checkpoint.html.		Refer to the checkpointing tutorial at https://pytorch.org/tutorials/recipes/recipes/saving_and_loading_a_general_checkpoint.html.

Dmitri at 22:33, 7 March 2024

2024-03-07T22:33:50Z

← Older revision		Revision as of 22:33, 7 March 2024
Line 8:		Line 8:
	You can reuse your favorite Python packages such as NumPy, SciPy, and Cython to extend PyTorch when needed.		You can reuse your favorite Python packages such as NumPy, SciPy, and Cython to extend PyTorch when needed.

	== Installing PyTorch ==		= Installing PyTorch =
	You will need a working local '''Conda''' install in your home directory first. If you do not have it yet, please follow [[Conda_on_ARC#Installing_Conda \| these instructions]] to have it installed.		You will need a working local '''Conda''' install in your home directory first. If you do not have it yet, please follow [[Conda_on_ARC#Installing_Conda \| these instructions]] to have it installed.

Line 130:		Line 130:
	</syntaxhighlight>		</syntaxhighlight>

	== Using PyTorch on ARC ==		= Using PyTorch on ARC =

	=== Requesting GPU Resources for PyTorch Jobs ===		=== Requesting GPU Resources for PyTorch Jobs ===

Dmitri at 22:33, 7 March 2024

2024-03-07T22:33:19Z

← Older revision		Revision as of 22:33, 7 March 2024
Line 1:		Line 1:
			= Background =

	'''PyTorch''' is a Python package that provides two high-level features:		'''PyTorch''' is a Python package that provides two high-level features:

Dmitri: /* Installing PyTorch */

2024-03-07T22:32:05Z

Installing PyTorch

← Older revision		Revision as of 22:32, 7 March 2024
Line 63:		Line 63:
	</pre>		</pre>
	'''Note''' that you will have to use the '''override''' variable again, if the additional packages depend on the '''GPU presence'''.		'''Note''' that you will have to use the '''override''' variable again, if the additional packages depend on the '''GPU presence'''.


	To deactivate the environment (and conda itself) the		To deactivate the environment (and conda itself) the
Line 71:		Line 72:
	</pre>		</pre>
	commands.		commands.


	=== Test script ===		=== Test script ===

Dmitri: /* Installing PyTorch */

2024-03-07T22:31:32Z

Installing PyTorch

← Older revision		Revision as of 22:31, 7 March 2024
Line 58:		Line 58:
	Once you know that your '''pytorch''' environment is working properly, you can add more packages to the environment using '''conda'''.		Once you know that your '''pytorch''' environment is working properly, you can add more packages to the environment using '''conda'''.
	<pre>		<pre>
	(pytorch) $ conda install -c conda-forge torchvision cuda-nvcc		(pytorch) $ CONDA_OVERRIDE_CUDA="12.2" conda install -c conda-forge torchvision cuda-nvcc
	....		....
	....		....
	</pre>		</pre>
			'''Note''' that you will have to use the '''override''' variable again, if the additional packages depend on the '''GPU presence'''.

	To deactivate the environment (and conda itself) the		To deactivate the environment (and conda itself) the

← Older revision		Revision as of 22:35, 7 March 2024
Line 162:		Line 162:
	* https://github.com/pytorch/pytorch - PyTorch project page on GitHub.		* https://github.com/pytorch/pytorch - PyTorch project page on GitHub.
	* https://docs.computecanada.ca/wiki/PyTorch - Compute Canada article is not directly applicable on ARC but contains a lot of good information.		* https://docs.computecanada.ca/wiki/PyTorch - Compute Canada article is not directly applicable on ARC but contains a lot of good information.
	* https://pytorch.org/tutorials/recipes/recipes/saving_and_loading_a_general_checkpoint.html - Checkpointing tutorial

	= Links =		= Links =

← Older revision		Revision as of 22:36, 7 March 2024
Line 157:		Line 157:
	== Checkpointing ==		== Checkpointing ==
	Refer to the checkpointing tutorial at https://pytorch.org/tutorials/recipes/recipes/saving_and_loading_a_general_checkpoint.html.		Refer to the checkpointing tutorial at https://pytorch.org/tutorials/recipes/recipes/saving_and_loading_a_general_checkpoint.html.

	~~== See also ==~~

	* https://github.com/pytorch/pytorch - PyTorch project page on GitHub.
	* https://docs.computecanada.ca/wiki/PyTorch - Compute Canada article is not directly applicable on ARC but contains a lot of good information.

	= Links =		= Links =