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

Ensure HPCX dependencies found in container #5922

Merged
merged 6 commits into from
Jun 11, 2023
Merged

Ensure HPCX dependencies found in container #5922

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
7057df1
Add HPCX dependencies to search path
dyastremsky Jun 8, 2023
ec23e61
Copy hpcx to CPU-only container
dyastremsky Jun 9, 2023
8222f11
Add ucc path to CPU-only image
dyastremsky Jun 9, 2023
c03110c
Fixed if statement
dyastremsky Jun 9, 2023
cfaebf2
Fix df variable
dyastremsky Jun 9, 2023
ca0962e
Combine hpcx LD_LIBRARY_PATH
dyastremsky Jun 9, 2023
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
13 changes: 13 additions & 0 deletions build.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1102,6 +1102,12 @@ def dockerfile_prepare_container_linux(argmap, backends, enable_gpu,
if 'onnxruntime' in backends:
df += '''
ENV LD_LIBRARY_PATH /opt/tritonserver/backends/onnxruntime:${LD_LIBRARY_PATH}
'''

# Necessary for libtorch.so to find correct HPCX libraries
if ('pytorch' in backends):
df += '''
ENV LD_LIBRARY_PATH /opt/hpcx/ucc/lib/:/opt/hpcx/ucx/lib/:${LD_LIBRARY_PATH}
'''

backend_dependencies = ""
Expand Down Expand Up @@ -1208,6 +1214,13 @@ def dockerfile_prepare_container_linux(argmap, backends, enable_gpu,
COPY --from=min_container /usr/local/cuda-12.1/targets/{cuda_arch}-linux/lib/libnvToolsExt.so.1 /usr/local/cuda/targets/{cuda_arch}-linux/lib/.
COPY --from=min_container /usr/local/cuda-12.1/targets/{cuda_arch}-linux/lib/libnvJitLink.so.12 /usr/local/cuda/targets/{cuda_arch}-linux/lib/.

RUN mkdir -p /opt/hpcx/ucc/lib/ /opt/hpcx/ucx/lib/
COPY --from=min_container /opt/hpcx/ucc/lib/libucc.so.1 /opt/hpcx/ucc/lib/libucc.so.1
COPY --from=min_container /opt/hpcx/ucx/lib/libucm.so.0 /opt/hpcx/ucx/lib/libucm.so.0
COPY --from=min_container /opt/hpcx/ucx/lib/libucp.so.0 /opt/hpcx/ucx/lib/libucp.so.0
COPY --from=min_container /opt/hpcx/ucx/lib/libucs.so.0 /opt/hpcx/ucx/lib/libucs.so.0
COPY --from=min_container /opt/hpcx/ucx/lib/libuct.so.0 /opt/hpcx/ucx/lib/libuct.so.0

COPY --from=min_container /usr/lib/{libs_arch}-linux-gnu/libcudnn.so.8 /usr/lib/{libs_arch}-linux-gnu/libcudnn.so.8

# patchelf is needed to add deps of libcublasLt.so.12 to libtorch_cuda.so
Expand Down