Skip to content
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

pytorch for gh200 #123

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions config.yaml
Original file line number Diff line number Diff line change
Expand Up @@ -178,6 +178,12 @@ uenvs:
deploy:
eiger: [zen2]
todi: [gh200]
pytorch:
"24.7":
recipes:
gh200: '24.7/gh200'
deploy:
todi: [gh200]
quantumespresso:
"v7.1":
recipes:
Expand Down
5 changes: 5 additions & 0 deletions recipes/pytorch/24.7/gh200/compilers.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
bootstrap:
spec: gcc@12
gcc:
specs:
- [email protected]
7 changes: 7 additions & 0 deletions recipes/pytorch/24.7/gh200/config.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,7 @@
name: pytorch
store: /user-environment
description: PyTorch for GH200
spack:
repo: https://github.com/spack/spack.git
commit: releases/v0.22
modules: true
33 changes: 33 additions & 0 deletions recipes/pytorch/24.7/gh200/environments.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,33 @@
pytorch:
compiler:
- toolchain: gcc
spec: [email protected]
mpi:
spec: [email protected]
gpu: cuda
unify: true
specs:
- aws-ofi-nccl@master
- cmake
- cuda@12
- cray-mpich
- nccl
- ninja
- [email protected]
- [email protected]
- py-numpy
- py-tqdm
- py-lightning
- py-typing-extensions
- [email protected] +distributed +cuda cuda_arch=90 +cudnn +nccl +mpi
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

thanks Ben!
we should probably include at least the basic torch packages torchvision torchaudio like in https://pytorch.org/get-started/locally/ as they also need to compiled with the correct cuda version (the wheels might not match)

variants:
- +mpi
- +cuda
- cuda_arch=90
views:
pytorch:
link: roots
uenv:
add_compilers: true
prefix_paths:
LD_LIBRARY_PATH: [lib, lib64]
23 changes: 23 additions & 0 deletions recipes/pytorch/24.7/gh200/modules.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,23 @@
modules:
# Paths to check when creating modules for all module sets
prefix_inspections:
bin:
- PATH
lib:
- LD_LIBRARY_PATH
lib64:
- LD_LIBRARY_PATH

default:
arch_folder: false
# Where to install modules
roots:
tcl: /user-environment/modules
tcl:
all:
autoload: none
hash_length: 0
exclude_implicits: true
exclude: ['%[email protected]', 'gcc %[email protected]']
projections:
all: '{name}/{version}'