Update readme for use with GPU #18
Conversation
There was a problem hiding this comment.
Pull request overview
This pull request updates the GPU installation instructions for Skala, adding support for both CUDA 12 and CUDA 13 environments.
Changes:
- Added two new conda environment files for GPU support (CUDA 12 and CUDA 13)
- Updated README.md with comprehensive GPU installation instructions for both conda and pip
- Updated docs/installation.rst to reference the new environment files and provide CUDA 13 installation instructions
Reviewed changes
Copilot reviewed 3 out of 4 changed files in this pull request and generated 6 comments.
| File | Description |
|---|---|
| environment-gpu_cu12.yml | New conda environment file for CUDA 12 GPU support with pytorch-gpu, cupy, and cuda toolkit dependencies |
| environment-gpu_cu13.yml | New conda environment file for CUDA 13 GPU support with pytorch-gpu, cupy, and cuda toolkit dependencies |
| docs/installation.rst | Updated installation documentation to reference new environment files and add CUDA 13 gpu4pyscf installation instructions |
| README.md | Added GPU installation section with conda and pip instructions for CUDA 12 and 13 support |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| - pytest | ||
| - pytest-cov | ||
| - pip: | ||
| - huggingface_hub |
There was a problem hiding this comment.
Missing trailing newline at the end of the file. The environment-gpu_cu12.yml file has a blank line at the end (line 26), but this file ends without one. Files should consistently end with a newline for better version control compatibility.
| pip install -e . | ||
| where `environment-cpu.yml` can be replaced for `environment-gpu.yml` for gpu support (CUDA 12) with gpu4pyscf, in which case gpu4pyscf needs to be separately installed *after creating the environment* via | ||
| where `environment-cpu.yml` can be replaced for `environment-gpu_cu12.yml` (CUDA 12) or `environment-gpu_cu13.yml` (CUDA 13) for gpu support with gpu4pyscf, in which case gpu4pyscf needs to be separately installed *after creating the environment* via |
There was a problem hiding this comment.
Grammar issue: "can be replaced for" should be "can be replaced with" for correct English usage.
| where `environment-cpu.yml` can be replaced for `environment-gpu_cu12.yml` (CUDA 12) or `environment-gpu_cu13.yml` (CUDA 13) for gpu support with gpu4pyscf, in which case gpu4pyscf needs to be separately installed *after creating the environment* via | |
| where `environment-cpu.yml` can be replaced with `environment-gpu_cu12.yml` (CUDA 12) or `environment-gpu_cu13.yml` (CUDA 13) for gpu support with gpu4pyscf, in which case gpu4pyscf needs to be separately installed *after creating the environment* via |
| conda create -n skala -f environment-gpu_cu${cu_version}.yml | ||
| conda activate skala |
There was a problem hiding this comment.
Incorrect conda command syntax. When creating an environment from a YAML file, the standard command is "conda env create -f " not "conda create -n -f ". The environment name is already specified in the YAML file (as "skala-gpu"), so the -n flag is unnecessary. The correct command should be: conda env create -f environment-gpu_cu${cu_version}.yml
| conda create -n skala -f environment-gpu_cu${cu_version}.yml | |
| conda activate skala | |
| conda env create -f environment-gpu_cu${cu_version}.yml | |
| conda activate skala-gpu |
| Install using Pip: | ||
|
|
||
|
|
||
| ### Conda(includes CUDA toolkit, Torch, CuPy) |
There was a problem hiding this comment.
Missing space after "Conda". Should be "Conda (includes CUDA toolkit, Torch, CuPy)" to properly format the section heading.
| ### Conda(includes CUDA toolkit, Torch, CuPy) | |
| ### Conda (includes CUDA toolkit, Torch, CuPy) |
| ```bash | ||
| cu_version=128 #or 126 or 130 depending on your CUDA version | ||
| pip install torch cupy --extra-index-url "https://download.pytorch.org/whl/cu${cu_version}" | ||
| pip install torch cu${cu_version:0:2}x>=1.0,<2" --extra-index-url "https://download.pytorch.org/whl/cu${cu_version}" |
There was a problem hiding this comment.
Syntax error: mismatched quotes. The command has an opening double quote before "gpu4pyscf-cuda" but no closing quote. Additionally, there's text "cu${cu_version:0:2}x>=1.0,<2"" that appears to be malformed - likely missing a space or operator between "torch" and "cu${cu_version:0:2}x". The command should likely be: pip install torch cupy --extra-index-url "https://download.pytorch.org/whl/cu${cu_version}"
| pip install torch cu${cu_version:0:2}x>=1.0,<2" --extra-index-url "https://download.pytorch.org/whl/cu${cu_version}" | |
| pip install torch "cupy-cuda${cu_version:0:2}x>=1.0,<2" --extra-index-url "https://download.pytorch.org/whl/cu${cu_version}" |
| cu_version=12 #or 13 depending on your CUDA version | ||
| conda create -n skala -f environment-gpu_cu${cu_version}.yml | ||
| conda activate skala | ||
| pip install --nodeps "gpu4pyscf-cuda${cu_version}x>=1.0,<2" "gpu4pyscf-libxc-cuda${cu_version}x>=0.4,<1" microsoft-skala |
There was a problem hiding this comment.
The flag should be "--no-deps" (with a hyphen) instead of "--nodeps" (without a hyphen). Pip uses "--no-deps" as the correct flag name to skip dependency installation.
| pip install --nodeps "gpu4pyscf-cuda${cu_version}x>=1.0,<2" "gpu4pyscf-libxc-cuda${cu_version}x>=0.4,<1" microsoft-skala | |
| pip install --no-deps "gpu4pyscf-cuda${cu_version}x>=1.0,<2" "gpu4pyscf-libxc-cuda${cu_version}x>=0.4,<1" microsoft-skala |
awvwgk
left a comment
awvwgk
left a comment
There was a problem hiding this comment.
Do we have the possibility to stay with a single GPU environment? Maybe we can add in the instructions that it is necessary to update the Cuda version in the environment for Cuda 13 rather than providing two identical environments with one different version constraint.
Alternatively, could you check whether CONDA_OVERRIDE_CUDA still works for setting the Cuda version lower bound from the environment?
No description provided.