Skip to content

Dockerfile - upgrade nccl version and install ucx to fix bug in cuda 12.4 docker file #646

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

Merged
yukirora merged 4 commits into from
Sep 20, 2024
Merged

Dockerfile - upgrade nccl version and install ucx to fix bug in cuda 12.4 docker file #646

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
74293df
upgrade nccl due to OOM bug in nccl v2.20 graph mode
Sep 17, 2024
f2698df
install ucx 1.16 for mutli thread support for mpi in ib-traffic
Sep 17, 2024
d8360e5
add ${NUM_MAKE_JOBS} for nccl and ucx build
yukirora Sep 20, 2024
607b893
Merge branch 'release/0.11' into yutji/upgrade-nccl
yukirora Sep 20, 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: 18 additions & 1 deletion dockerfile/cuda12.4.dockerfile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,7 @@ FROM nvcr.io/nvidia/pytorch:24.03-py3
# - CUDA: 12.4.0
# - cuDNN: 9.0.0.306
# - cuBLAS: 12.4.2.65
# - NCCL: v2.20
# - NCCL: v2.23.4-1
# - TransformerEngine 1.4
# Mellanox:
# - OFED: 23.07-0.5.1.2
Expand Down Expand Up @@ -115,6 +115,23 @@ RUN cd /tmp && \
mv amd-blis /opt/AMD && \
rm -rf aocl-blis-linux-aocc-4.0.tar.gz

# Install NCCL 2.23.4
RUN cd /tmp && \
git clone -b v2.23.4-1 https://github.com/NVIDIA/nccl.git && \
cd nccl && \
make -j ${NUM_MAKE_JOBS} src.build && \
make install && \
rm -rf /tmp/nccl

# Install UCX v1.16.0 with multi-threading support
RUN cd /tmp && \
wget https://github.com/openucx/ucx/releases/download/v1.16.0/ucx-1.16.0.tar.gz && \
tar xzf ucx-1.16.0.tar.gz && \
cd ucx-1.16.0 && \
./contrib/configure-release-mt --prefix=/usr/local && \
make -j ${NUM_MAKE_JOBS} && \
make install

ENV PATH="${PATH}" \
LD_LIBRARY_PATH="/usr/local/lib:/usr/local/mpi/lib:${LD_LIBRARY_PATH}" \
SB_HOME=/opt/superbench \
Expand Down