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.

Sign up for GitHub

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

Jump to bottom

base: CUDA 12.6.2 and Ubuntu 24.04 #1122

Closed
wants to merge 2 commits into from
Closed

base: CUDA 12.6.2 and Ubuntu 24.04 #1122

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
19b1ca2
base: CUDA 12.6.2 and Ubuntu 24.04
olupton Oct 28, 2024
1119f84
Install to global /opt/venv
olupton Oct 28, 2024
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
19 changes: 12 additions & 7 deletions .github/container/Dockerfile.base
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
# syntax=docker/dockerfile:1-labs
ARG BASE_IMAGE=nvidia/cuda:12.6.1-devel-ubuntu22.04
ARG BASE_IMAGE=nvidia/cuda:12.6.2-devel-ubuntu24.04
ARG GIT_USER_NAME="JAX Toolbox"
ARG [email protected]
ARG CLANG_VERSION=18
Expand All @@ -8,17 +8,17 @@ ARG CLANG_VERSION=18
## Obtain GCP's NCCL TCPx plugin
###############################################################################

FROM us-docker.pkg.dev/gce-ai-infra/gpudirect-tcpx/nccl-plugin-gpudirecttcpx:v3.1.10 as tcpx-installer-amd64
FROM us-docker.pkg.dev/gce-ai-infra/gpudirect-tcpx/nccl-plugin-gpudirecttcpx:v3.1.10 AS tcpx-installer-amd64

# make a stub arm64 container because GCP does not provide an arm64 version of the plugin
FROM ubuntu as tcpx-installer-arm64
FROM ubuntu AS tcpx-installer-arm64
RUN <<"OUTEREOF" bash -ex
mkdir -p /scripts /var/lib/tcpx/lib64
echo '#!/bin/bash' > /scripts/container_entry.sh
chmod +x /scripts/container_entry.sh
OUTEREOF

FROM tcpx-installer-${TARGETARCH} as tcpx-installer
FROM tcpx-installer-${TARGETARCH} AS tcpx-installer
RUN /scripts/container_entry.sh install

###############################################################################
Expand Down Expand Up @@ -51,14 +51,16 @@ apt_packages=(
liblzma-dev
python-is-python3
python3-pip
python3-venv
rsync
vim
wget
jq
# llvm.sh
lsb-release software-properties-common
lsb-release
software-properties-common
# GCP autoconfig
pciutils hwloc bind9-host
# pciutils hwloc bind9-host
)
if [[ $(dpkg --print-architecture) == arm64 ]]; then
# h5py: The newest release of of h5py (3.11.0) does not include ARM wheels and causes pip to build h5py.
Expand Down Expand Up @@ -125,7 +127,10 @@ git apply </opt/pip/pip-vcs-equivalency.patch
git add -u
git commit -m 'Adds JAX_TOOLBOX_VCS_EQUIVALENCY as a trigger to treat all github VCS installs for a package as equivalent. The spec of the last encountered version will be used'
EOF
RUN pip install --upgrade --no-cache-dir -e /opt/pip pip-tools && rm -rf ~/.cache/*
# Create a system-wide venv for the pip-installed world inside the containers
RUN python -m venv --prompt jax /opt/venv && /opt/venv/bin/pip install --no-cache-dir -e /opt/pip pip-tools
# Make sure `python` refers to the venv version
ENV PATH=/opt/venv/bin:${PATH}

###############################################################################
## Install TCPx
Expand Down
Loading