@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 5255d99dc5
Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git

Synced Nov 17, 05:30 UTC

[ROCm] Dockerfile fix for flash-attention build (#2885)

Hongxia Yang committed Feb 15, 2024 at 13:22 UTC 5255d99dc595f9ae7647842242d6542aa4145a4f
1 file changed +3 -3
Dockerfile.rocm
+3 -3
@@ -56,10 +56,10 @@ ENV LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/opt/rocm/lib/:/libtorch/lib:
56 ENV CPLUS_INCLUDE_PATH=$CPLUS_INCLUDE_PATH:/libtorch/include:/libtorch/include/torch/csrc/api/include/:/opt/rocm/include/:
57
58 # Install ROCm flash-attention
59 -RUN if [ "$BUILD_FA" == "1" ]; then \
59 +RUN if [ "$BUILD_FA" = "1" ]; then \
60 mkdir libs \
61 && cd libs \
62 - && git clone https://github.com/ROCmSoftwarePlatform/flash-attention.git \
62 + && git clone https://github.com/ROCm/flash-attention.git \
63 && cd flash-attention \
64 && git checkout ${FA_BRANCH} \
65 && git submodule update --init \
@@ -83,7 +83,7 @@ RUN if [ "$BASE_IMAGE" = "rocm/pytorch:rocm6.0_ubuntu20.04_py3.9_pytorch_2.1.1"
83 RUN cd /app \
84 && cd vllm \
85 && pip install -U -r requirements-rocm.txt \
86 - && if [ "$BUILD_FA" == "1" ]; then \
86 + && if [ "$BUILD_FA" = "1" ]; then \
87 bash patch_xformers.rocm.sh; fi \
88 && patch /opt/rocm/include/hip/amd_detail/amd_hip_bf16.h /app/vllm/rocm_patch/rocm_bf16.patch \
89 && python3 setup.py install \