Open dzlab opened 7 months ago
Updated. Can you try the latest commit and follow the instruction in README?
it runs into multiple compile issues
writing manifest file 'punica.egg-info/SOURCES.txt'
running build_ext
/usr/local/lib/python3.10/dist-packages/torch/utils/cpp_extension.py:424: UserWarning: There are no x86_64-linux-gnu-g++ version bounds defined for CUDA version 11.8
warnings.warn(f'There are no {compiler_name} version bounds defined for CUDA version {cuda_str_version}')
building 'punica.ops._kernels' extension
creating /content/punica/build/temp.linux-x86_64-cpython-310
creating /content/punica/build/temp.linux-x86_64-cpython-310/csrc
creating /content/punica/build/temp.linux-x86_64-cpython-310/csrc/bgmv
creating /content/punica/build/temp.linux-x86_64-cpython-310/csrc/flashinfer_adapter
creating /content/punica/build/temp.linux-x86_64-cpython-310/csrc/rms_norm
creating /content/punica/build/temp.linux-x86_64-cpython-310/csrc/sgmv
creating /content/punica/build/temp.linux-x86_64-cpython-310/csrc/sgmv_flashinfer
Emitting ninja build file /content/punica/build/temp.linux-x86_64-cpython-310/build.ninja...
Compiling objects...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
[1/6] /usr/local/cuda/bin/nvcc -I/content/punica/third_party/cutlass/include -I/usr/local/lib/python3.10/dist-packages/torch/include -I/usr/local/lib/python3.10/dist-packages/torch/include/torch/csrc/api/include -I/usr/local/lib/python3.10/dist-packages/torch/include/TH -I/usr/local/lib/python3.10/dist-packages/torch/include/THC -I/usr/local/cuda/include -I/usr/include/python3.10 -c -c /content/punica/csrc/bgmv/bgmv_all.cu -o /content/punica/build/temp.linux-x86_64-cpython-310/csrc/bgmv/bgmv_all.o --expt-relaxed-constexpr --compiler-options ''"'"'-fPIC'"'"'' -DTORCH_API_INCLUDE_EXTENSION_H '-DPYBIND11_COMPILER_TYPE="_gcc"' '-DPYBIND11_STDLIB="_libstdcpp"' '-DPYBIND11_BUILD_ABI="_cxxabi1011"' -DTORCH_EXTENSION_NAME=_kernels -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_75,code=compute_75 -gencode=arch=compute_75,code=sm_75 -std=c++17
FAILED: /content/punica/build/temp.linux-x86_64-cpython-310/csrc/bgmv/bgmv_all.o
/usr/local/cuda/bin/nvcc -I/content/punica/third_party/cutlass/include -I/usr/local/lib/python3.10/dist-packages/torch/include -I/usr/local/lib/python3.10/dist-packages/torch/include/torch/csrc/api/include -I/usr/local/lib/python3.10/dist-packages/torch/include/TH -I/usr/local/lib/python3.10/dist-packages/torch/include/THC -I/usr/local/cuda/include -I/usr/include/python3.10 -c -c /content/punica/csrc/bgmv/bgmv_all.cu -o /content/punica/build/temp.linux-x86_64-cpython-310/csrc/bgmv/bgmv_all.o --expt-relaxed-constexpr --compiler-options ''"'"'-fPIC'"'"'' -DTORCH_API_INCLUDE_EXTENSION_H '-DPYBIND11_COMPILER_TYPE="_gcc"' '-DPYBIND11_STDLIB="_libstdcpp"' '-DPYBIND11_BUILD_ABI="_cxxabi1011"' -DTORCH_EXTENSION_NAME=_kernels -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_75,code=compute_75 -gencode=arch=compute_75,code=sm_75 -std=c++17
/content/punica/csrc/bgmv/../flashinfer/vec_dtypes.cuh(935): error: identifier "make_bfloat162" is undefined
/content/punica/csrc/bgmv/../flashinfer/vec_dtypes.cuh(986): error: identifier "make_bfloat162" is undefined
/content/punica/csrc/bgmv/bgmv_impl.cuh(181): error: no operator "+=" matches these operands
operand types are: nv_bfloat16 += float
detected during:
instantiation of "void bgmv_expand_kernel<feat_in,feat_out,T>(T *, const T *, const T *, const int64_t *, int64_t, int64_t, float) [with feat_in=8, feat_out=768, T=nv_bfloat16]"
(199): here
instantiation of "void bgmv_kernel<feat_in,feat_out,T>(T *, const T *, const T *, const int64_t *, int64_t, int64_t, int64_t, float) [with feat_in=8, feat_out=768, T=nv_bfloat16]"
/content/punica/csrc/bgmv/bgmv_all.cu(5): here
/content/punica/csrc/bgmv/bgmv_impl.cuh(135): error: no operator "+=" matches these operands
operand types are: nv_bfloat16 += float
detected during:
instantiation of "void bgmv_shrink_kernel<feat_in,feat_out,T>(T *, const T *, const T *, const int64_t *, int64_t, int64_t, float) [with feat_in=768, feat_out=8, T=nv_bfloat16]"
(205): here
instantiation of "void bgmv_kernel<feat_in,feat_out,T>(T *, const T *, const T *, const int64_t *, int64_t, int64_t, int64_t, float) [with feat_in=768, feat_out=8, T=nv_bfloat16]"
/content/punica/csrc/bgmv/bgmv_all.cu(5): here
/content/punica/csrc/bgmv/bgmv_impl.cuh(181): error: no operator "+=" matches these operands
operand types are: nv_bfloat16 += float
detected during:
instantiation of "void bgmv_expand_kernel<feat_in,feat_out,T>(T *, const T *, const T *, const int64_t *, int64_t, int64_t, float) [with feat_in=8, feat_out=1024, T=nv_bfloat16]"
(199): here
instantiation of "void bgmv_kernel<feat_in,feat_out,T>(T *, const T *, const T *, const int64_t *, int64_t, int64_t, int64_t, float) [with feat_in=8, feat_out=1024, T=nv_bfloat16]"
/content/punica/csrc/bgmv/bgmv_all.cu(5): here
@dzlab seems your GPU architecture is sm_75, currently, our code base requires sm_80 or later. I'm adding sm75 support now.
Hey @yzh119 any update on sm_75 support for punica LORA kernels?
I'm not able to install this library on Colab. I tried this
But this is failing with the following error