codefuse-ai / FasterTransformer4CodeFuse

High-performance LLM inference based on our optimized version of FastTransfomer
Other
123 stars 9 forks source link

cmake -DCMAKE_BUILD_TYPE=Release -DSM="80;75" -DBUILD_PYT=ON -DSPARSITY_SUPPORT=OFF -DMEASURE_BUILD_TIME=ON \ -DBUILD_CUTLASS_MIXED_GEMM=ON -DBUILD_MULTI_GPU=ON -DBUILD_TRT=OFF \ -DENABLE_FP8=OFF -DBUILD_PYBIND=ON -DTORCH_PYTHON_LIBRARIES=${TORCH_PYTHON _LIBRARIES} .. _LIBRARIES} .. #4

Closed quida01 closed 11 months ago

quida01 commented 11 months ago

Branch/Tag/Commit

main

Docker Image Version

conda

GPU name

A6000

CUDA Driver

530.30.02

Reproduced Steps

cmake -DCMAKE_BUILD_TYPE=Release -DSM="80;75"                                                                                                              -DBUILD_PYT=ON -DSPARSITY_SUPPORT=OFF -DMEASURE_BUILD_TIME=ON \
      -DBUILD_CUTLASS_MIXED_GEMM=ON -DBUILD_MULTI_GPU=ON -DBUILD_TRT=OFF \
      -DENABLE_FP8=OFF -DBUILD_PYBIND=ON -DTORCH_PYTHON_LIBRARIES=${TORCH_PYTHON                                                                                                             _LIBRARIES} ..
CMake Warning (dev) at CMakeLists.txt:17 (find_package):
  Policy CMP0146 is not set: The FindCUDA module is removed.  Run "cmake
  --help-policy CMP0146" for policy details.  Use the cmake_policy command to
  set the policy and suppress this warning.

This warning is for project developers.  Use -Wno-dev to suppress it.

CUDA_VERSION 12.2 is greater or equal than 11.0, enable -DENABLE_BF16 flag
-- Could NOT find CUDNN (missing: CUDNN_LIBRARY_PATH CUDNN_INCLUDE_PATH)
-- Add DBUILD_CUTLASS_MOE, requires CUTLASS. Increases compilation time
-- Add DBUILD_CUTLASS_MIXED_GEMM, requires CUTLASS. Increases compilation time
-- Add DBUILD_MULTI_GPU, requires MPI and NCCL
CMake Error at /root/miniconda3/envs/CodeFuse/lib/python3.8/site-packages/cmake/                                                                                                             data/share/cmake-3.27/Modules/FindPackageHandleStandardArgs.cmake:230 (message):
  Could NOT find NCCL (missing: NCCL_INCLUDE_DIRS NCCL_LIBRARIES)
Call Stack (most recent call first):
  /root/miniconda3/envs/CodeFuse/lib/python3.8/site-packages/cmake/data/share/cm                                                                                                             ake-3.27/Modules/FindPackageHandleStandardArgs.cmake:600 (_FPHSA_FAILURE_MESSAGE                                                                                                             )
  cmake/Modules/FindNCCL.cmake:126 (find_package_handle_standard_args)
  CMakeLists.txt:84 (find_package)
zhang-ge-hao commented 11 months ago

Thanks for your response!

It looks like you didn't reproduce it using the recommended docker image. Please align your environment with nvcr.io/nvidia/pytorch:22.09-py3 and try another time.

If you have any problems when reproducing this repository using a suitable docker image, please feel free to re-open this issue.