huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 04:14:52 +00:00
Code Issues Packages Projects Releases Wiki Activity

backend(trtllm): make sccache definition manually

Browse Source
This commit is contained in:
Morgan Funtowicz 2025-01-24 22:01:43 +01:00
parent a434c2ffc9
commit e7064c95da
3 changed files with 9 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
Dockerfile_trtllm
View File

@ -1,5 +1,4 @@
ARG cuda_arch_list="75-real;80-real;86-real;89-real;90-real"
ARG ompi_version="4.1.7"
ARG build_type=release
ARG sccache_gha_enabled=no
ARG actions_cache_url=""
@ -36,19 +35,16 @@ ENV TENSORRT_INSTALL_PREFIX=/usr/local/tensorrt
# Install OpenMPI
FROM cuda-builder AS mpi-builder
ARG ompi_version
ENV OMPI_TARBALL_FILENAME="openmpi-$ompi_version.tar.bz2"
ADD --checksum=sha256:54a33cb7ad81ff0976f15a6cc8003c3922f0f3d8ceed14e1813ef3603f22cd34 \
https://download.open-mpi.org/release/open-mpi/v4.1/$OMPI_TARBALL_FILENAME \
/opt/src/mpi/
WORKDIR /opt/src/mpi
RUN tar --strip-components=1 -xf $OMPI_TARBALL_FILENAME &&\
ADD --checksum=sha256:54a33cb7ad81ff0976f15a6cc8003c3922f0f3d8ceed14e1813ef3603f22cd34 \
https://download.open-mpi.org/release/open-mpi/v4.1/openmpi-4.1.7.tar.bz2 .
RUN tar --strip-components=1 -xf openmpi-4.1.7.tar.bz2 &&\
./configure --prefix=/usr/local/mpi --with-cuda=/usr/local/cuda --with-slurm && \
make -j all && \
make install && \
rm -rf "/opt/src/$OMPI_TARBALL_FILENAME"
rm -rf "/opt/src/openmpi-4.1.7.tar.bz2"
# Install TensorRT
FROM cuda-builder AS trt-builder
@ -97,6 +93,9 @@ COPY --from=trt-builder /usr/local/tensorrt /usr/local/tensorrt
COPY --from=mpi-builder /usr/local/mpi /usr/local/mpi
ENV RUSTC_WRAPPER=sccache
ENV CMAKE_C_COMPILER_LAUNCHER=sccache
ENV CMAKE_CXX_COMPILER_LAUNCHER=sccache
ENV CMAKE_CUDA_COMPILER_LAUNCHER=sccache
ENV CMAKE_INSTALL_PREFIX=$TGI_INSTALL_PREFIX
RUN mkdir $TGI_INSTALL_PREFIX && mkdir "$TGI_INSTALL_PREFIX/include" && mkdir "$TGI_INSTALL_PREFIX/lib" && \
cargo build --profile ${build_type} --package text-generation-backends-trtllm --bin text-generation-backends-trtllm && \

12
backends/trtllm/CMakeLists.txt
View File

@ -1,17 +1,5 @@
cmake_minimum_required(VERSION 3.20)
find_program(CCACHE_EXECUTABLE sccache)
if (NOT CCACHE_EXECUTABLE)
find_program(CCACHE_EXECUTABLE ccache)
endif ()
if (CCACHE_EXECUTABLE)
message(STATUS "Using compiler with caching: ${CCACHE_EXECUTABLE}")
set(CMAKE_C_COMPILER_LAUNCHER "${CCACHE_EXECUTABLE}")
set(CMAKE_CXX_COMPILER_LAUNCHER "${CCACHE_EXECUTABLE}")
set(CMAKE_CUDA_COMPILER_LAUNCHER "${CCACHE_EXECUTABLE}")
endif ()
if (CMAKE_VERSION VERSION_GREATER_EQUAL "3.24.0")
cmake_policy(SET CMP0135 NEW)
endif ()

4
backends/trtllm/build.rs
View File

@ -138,10 +138,6 @@ fn build_backend(is_debug: bool, opt_level: &str, out_dir: &PathBuf) -> (PathBuf
if let Some(wrapper) = option_env!("RUSTC_WRAPPER") {
println!("cargo:warning=Using caching tool: {wrapper}");
env::set_var("CMAKE_C_COMPILER_LAUNCHER", wrapper);
env::set_var("CMAKE_CXX_COMPILER_LAUNCHER", wrapper);
env::set_var("CMAKE_CUDA_COMPILER_LAUNCHER", wrapper);
}
// Allow to override which Python to use ...