Skip to content

Commit aec51d9

Browse files
mc-nvdmitry-tokarev-nv
mc-nv
and
dmitry-tokarev-nv
authored
TPRD-1710: Update default branches post-25.09 (#808)
* Update for 25.09 (#797) * Update versions * Fix version value * Update TensorRT-LLM submodule * Update Dockerfile.triton.trt_llm_backend - install pkg-config (#800) * Update Dockerfile.triton.trt_llm_backend - apt-get update * Update Dockerfile.triton.trt_llm_backend - libzmq3-dev * Update Dockerfile.triton.trt_llm_backend - libibverbs-dev * Update Dockerfile.triton.trt_llm_backend - rdma-core-devel * Update Dockerfile.triton.trt_llm_backend - mlx5 * Update Dockerfile.triton.trt_llm_backend - fixes * Update Dockerfile.triton.trt_llm_backend - syntax fix * bump TRT_VER to 10.13.0.35 * Update tensorrt_llm submodule to v1.0.0 ae8270b71 (#801) * switch back to TRT_VER=10.11.0.33 * Update Dockerfile by switching to public distribution of TensorRT-LLM (#803) * Update Dockerfile by switching to public distribution of TensorRT-LLM * Adding secrtets for internal build --------- Co-authored-by: Dmitry Tokarev <[email protected]>
1 parent 17bc7ab commit aec51d9

File tree

1 file changed

+20
-79
lines changed

1 file changed

+20
-79
lines changed

‎dockerfile/Dockerfile.triton.trt_llm_backend‎

Lines changed: 20 additions & 79 deletions
Original file line numberDiff line numberDiff line change
@@ -1,78 +1,31 @@
1-
ARG BASE_IMAGE=nvcr.io/nvidia/tritonserver:25.05-py3-min
2-
ARG PYTORCH_IMAGE=nvcr.io/nvidia/pytorch:25.05-py3
3-
ARG NVRTC_VER=12.9.41-1
1+
ARG BASE_IMAGE=nvcr.io/nvidia/tritonserver:25.06-py3-min
2+
ARG PYTORCH_IMAGE=nvcr.io/nvidia/pytorch:25.06-py3
3+
ARG NVRTC_VER=12.9.86-1
44
ARG TRT_VER=10.11.0.33
55
ARG NCCL_VER=2.27.5-1+cuda12.9
66
ARG RELEASE_URL_TRT_x86=https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.11.0/tars/TensorRT-${TRT_VER}.Linux.x86_64-gnu.cuda-12.9.tar.gz
77
ARG RELEASE_URL_TRT_ARM=https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.11.0/tars/TensorRT-${TRT_VER}.Linux.aarch64-gnu.cuda-12.9.tar.gz
88

99
# Versions of packages to copy from pytorch image
10-
ARG TORCH_VER=2.8.0a0+5228986c39.nv25.5
11-
ARG TORCHVISION_VER=0.22.0a0
10+
ARG TORCH_VER=2.8.0a0+5228986c39.nv25.6
11+
ARG TORCHVISION_VER=0.22.0a0+95f10a4e
1212
ARG SETUPTOOLS_VER=78.1.1
1313
ARG PYTORCH_TRITON_VER=3.3.0+git96316ce52.nvinternal
1414
ARG JINJA2_VER=3.1.6
15-
ARG NETWORKX_VER=3.4.2
15+
ARG NETWORKX_VER=3.5
1616
ARG SYMPY_VER=1.14.0
1717
ARG PACKAGING_VER=23.2
18-
ARG FLASH_ATTN_VER=2.7.3
18+
ARG FLASH_ATTN_VER=2.7.4.post1
19+
20+
ARG TENSORRTLLM_REPO=https://github.com/NVIDIA/TensorRT-LLM.git
21+
ARG TENSORRTLLM_REPO_TAG=release/1.0
22+
ARG TENSORRTLLM_VER=1.0.0
1923

2024
FROM ${PYTORCH_IMAGE} AS pytorch_image
2125
FROM ${BASE_IMAGE} AS install_dependencies
2226

23-
ARG TENSORRTLLM_REPO=https://github.com/NVIDIA/TensorRT-LLM.git
24-
ARG TENSORRTLLM_REPO_TAG=release/0.21
25-
2627
WORKDIR /workspace
2728

28-
ARG CCACHE_REMOTE_STORAGE
29-
ARG CCACHE_URL
30-
ENV CCACHE_DEBUG=1
31-
32-
RUN if [ -n "${CCACHE_REMOTE_STORAGE}" ] ; then \
33-
curl -k -L ${CCACHE_URL} -o ccache.tar.gz ; \
34-
tar -xzf ccache.tar.gz -C /usr/local --strip-components=1 ; \
35-
rm ccache.tar.gz ; \
36-
ccache --set-config=remote_only=true ; \
37-
ccache --set-config=remote_storage=${CCACHE_REMOTE_STORAGE} ; \
38-
ccache --set-config=log_file=/tmp/ccache.log ; \
39-
ccache -p ; \
40-
fi
41-
42-
ARG TORCH_VER
43-
ARG TORCHVISION_VER
44-
ARG SETUPTOOLS_VER
45-
ARG PYTORCH_TRITON_VER
46-
ARG JINJA2_VER
47-
ARG NETWORKX_VER
48-
ARG SYMPY_VER
49-
ARG PACKAGING_VER
50-
ARG FLASH_ATTN_VER
51-
# Copy PyTorch package from PyTorch image
52-
COPY --from=pytorch_image /usr/local/lib/lib* /usr/local/lib/
53-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/torch /usr/local/lib/python3.12/dist-packages/torch
54-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/torch-${TORCH_VER}.dist-info /usr/local/lib/python3.12/dist-packages/torch-${TORCH_VER}.dist-info
55-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/torchgen /usr/local/lib/python3.12/dist-packages/torchgen
56-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/torchvision /usr/local/lib/python3.12/dist-packages/torchvision
57-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/torchvision-${TORCHVISION_VER}.dist-info /usr/local/lib/python3.12/dist-packages/torchvision-${TORCHVISION_VER}.dist-info
58-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/torchvision.libs /usr/local/lib/python3.12/dist-packages/torchvision.libs
59-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/setuptools /usr/local/lib/python3.12/dist-packages/setuptools
60-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/setuptools-${SETUPTOOLS_VER}.dist-info /usr/local/lib/python3.12/dist-packages/setuptools-${SETUPTOOLS_VER}.dist-info
61-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/functorch /usr/local/lib/python3.12/dist-packages/functorch
62-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/pytorch_triton-${PYTORCH_TRITON_VER}.dist-info /usr/local/lib/python3.12/dist-packages/pytorch_triton-${PYTORCH_TRITON_VER}.dist-info
63-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/triton /usr/local/lib/python3.12/dist-packages/triton
64-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/jinja2 /usr/local/lib/python3.12/dist-packages/jinja2
65-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/jinja2-${JINJA2_VER}.dist-info /usr/local/lib/python3.12/dist-packages/jinja2-${JINJA2_VER}.dist-info
66-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/networkx /usr/local/lib/python3.12/dist-packages/networkx
67-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/networkx-${NETWORKX_VER}.dist-info /usr/local/lib/python3.12/dist-packages/networkx-${NETWORKX_VER}.dist-info
68-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/sympy /usr/local/lib/python3.12/dist-packages/sympy
69-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/sympy-${SYMPY_VER}.dist-info /usr/local/lib/python3.12/dist-packages/sympy-${SYMPY_VER}.dist-info
70-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/packaging /usr/local/lib/python3.12/dist-packages/packaging
71-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/packaging-${PACKAGING_VER}.dist-info /usr/local/lib/python3.12/dist-packages/packaging-${PACKAGING_VER}.dist-info
72-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/flash_attn /usr/local/lib/python3.12/dist-packages/flash_attn
73-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/flash_attn-${FLASH_ATTN_VER}.dist-info /usr/local/lib/python3.12/dist-packages/flash_attn-${FLASH_ATTN_VER}.dist-info
74-
COPY --from=pytorch_image /usr/local/lib/python3.12/dist-packages/flash_attn_2_cuda.cpython-312-*-linux-gnu.so /usr/local/lib/python3.12/dist-packages/
75-
7629
# Might not need to copy cusparseLt in the future once it's included in DLFW cuda container
7730
COPY --from=pytorch_image /usr/local/cuda/lib64/libcusparseLt* /usr/local/cuda/lib64/
7831

@@ -134,7 +87,7 @@ RUN pip3 install /usr/local/tensorrt/python/tensorrt-*-cp$( python3 -c "import s
13487
ENV LD_LIBRARY_PATH=/usr/local/tensorrt/lib:${LD_LIBRARY_PATH}
13588
ENV TRT_ROOT=/usr/local/tensorrt
13689

137-
FROM install_dependencies AS tensorrt_llm_build
90+
FROM install_dependencies AS tensorrt_llm_code
13891

13992
WORKDIR /workspace
14093

@@ -143,16 +96,6 @@ ARG TENSORRTLLM_REPO_TAG
14396
RUN git-lfs install \
14497
&& git clone --single-branch --recurse-submodules --depth=1 -b ${TENSORRTLLM_REPO_TAG} ${TENSORRTLLM_REPO} tensorrt_llm
14598

146-
RUN pip3 install --no-cache-dir polygraphy==0.49.9 mpi4py==3.1.5 cmake==4.0.3 ninja
147-
ENV CMAKE_POLICY_VERSION_MINIMUM=3.5
148-
149-
RUN cd tensorrt_llm && \
150-
if [ -n "${CCACHE_REMOTE_STORAGE}" ] ; then \
151-
python3 scripts/build_wheel.py --trt_root="${TRT_ROOT}" --clean --use_ccache ; \
152-
else \
153-
python3 scripts/build_wheel.py --trt_root="${TRT_ROOT}" --clean ; \
154-
fi
155-
15699
# Final stage to build the TRT-LLM container
157100
FROM ${BASE_IMAGE} AS final_stage
158101

@@ -239,17 +182,15 @@ ENV TRT_VERSION=$TRT_VER
239182
ENV LD_LIBRARY_PATH=/usr/local/tensorrt/lib:${LD_LIBRARY_PATH}
240183
ENV TRT_ROOT=/usr/local/tensorrt
241184

242-
WORKDIR /tmp
243-
244185
# Install TRT-LLM wheel after all the dependencies are installed
245-
COPY --from=tensorrt_llm_build /workspace/tensorrt_llm/build/tensorrt_llm*whl .
246-
RUN pip3 install --no-cache-dir tensorrt_llm*.whl \
247-
&& rm -f tensorrt_llm*.whl
186+
ARG TENSORRTLLM_VER
187+
RUN --mount=type=secret,id=pypi_extra_values,env=PYPI_EXTRA_VALUES \
188+
pip install --no-cache-dir ${PYPI_EXTRA_VALUES} tensorrt_llm==${TENSORRTLLM_VER}
248189

249190
# Copying the Tensorrt LLM scripts and applications
250191
WORKDIR /app
251-
COPY --from=tensorrt_llm_build /workspace/tensorrt_llm/triton_backend/scripts scripts
252-
COPY --from=tensorrt_llm_build /workspace/tensorrt_llm/triton_backend/all_models all_models
253-
COPY --from=tensorrt_llm_build /workspace/tensorrt_llm/triton_backend/inflight_batcher_llm/client client
254-
COPY --from=tensorrt_llm_build /workspace/tensorrt_llm/triton_backend/tools tools
255-
COPY --from=tensorrt_llm_build /workspace/tensorrt_llm/examples examples
192+
COPY --from=tensorrt_llm_code /workspace/tensorrt_llm/triton_backend/scripts scripts
193+
COPY --from=tensorrt_llm_code /workspace/tensorrt_llm/triton_backend/all_models all_models
194+
COPY --from=tensorrt_llm_code /workspace/tensorrt_llm/triton_backend/inflight_batcher_llm/client client
195+
COPY --from=tensorrt_llm_code /workspace/tensorrt_llm/triton_backend/tools tools
196+
COPY --from=tensorrt_llm_code /workspace/tensorrt_llm/examples examples

0 commit comments

Comments
 (0)