diff --git a/infra/docker/Dockerfile.ci-builder b/infra/docker/Dockerfile.ci-builder index 475aec1d2..690b67eee 100644 --- a/infra/docker/Dockerfile.ci-builder +++ b/infra/docker/Dockerfile.ci-builder @@ -55,13 +55,14 @@ RUN curl -fsSL https://downloads.rclone.org/v1.69.1/rclone-v1.69.1-linux-amd64.z RUN rustup component add clippy # Nsight Systems CLI — GPU profiling for CUDA graph node-level traces -# Not in the CUDA 13 devel repo — install from NVIDIA's nsight-systems repo -RUN curl -fsSL https://developer.download.nvidia.com/devtools/repos/ubuntu2404/amd64/nvidia-devtools-keyring_2024.1.gpg \ - -o /usr/share/keyrings/nvidia-devtools.gpg \ - && echo "deb [signed-by=/usr/share/keyrings/nvidia-devtools.gpg] https://developer.download.nvidia.com/devtools/repos/ubuntu2404/amd64/ /" \ - > /etc/apt/sources.list.d/nvidia-devtools.list \ - && apt-get update && apt-get install -y --no-install-recommends nsight-systems-cli \ - && rm -rf /var/lib/apt/lists/* +# The CUDA 13 devel image has NVIDIA repo configured but nsight-systems +# package name varies. Try the full package first, fall back to CLI-only. +RUN apt-get update \ + && (apt-get install -y --no-install-recommends nsight-systems-2025.2.1 \ + || apt-get install -y --no-install-recommends nsight-systems \ + || echo "WARNING: nsight-systems not available — nsys profiling disabled") \ + && rm -rf /var/lib/apt/lists/* \ + && (nsys --version 2>/dev/null || true) # CUDA env ENV CUDA_HOME=/usr/local/cuda