EETQ wheel not building #27
Comments
|
@donjuanpond Installing EETQ takes a long time because the compilation of the cutlass library is slow. You can use |
|
Hi @SidaZh, What should I do about this? |
|
My system just updated CUDA to 12.6 and I can no longer compile EEQT. FAILED: /tmp/pip-req-build-jgyvqov6/build/temp.linux-x86_64-cpython-312/tmp/pip-req-build-jgyvqov6/csrc/cutlass_kernels/fpA_intB_gemm_wrapper.o
…
/usr/local/cuda/bin/nvcc --generate-dependencies-with-compile --dependency-output /tmp/pip-req-build-jgyvqov6/build/temp.linux-x86_64-cpython-312/tmp/pip-req-build-jgyvqov6/csrc/cutlass_kernels/fpA_intB_gemm_wrapper.o.d -DVERSION_INFO=1.0.0 '-I['"'"'/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include'"'"', '"'"'/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include/torch/csrc/api/include'"'"', '"'"'/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include/TH'"'"', '"'"'/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include/THC'"'"', '"'"'/usr/local/cuda/include'"'"']' -I/tmp/pip-req-build-jgyvqov6/csrc -I/tmp/pip-req-build-jgyvqov6/csrc/utils -I/tmp/pip-req-build-jgyvqov6/csrc/cutlass/include -I/tmp/pip-req-build-jgyvqov6/csrc/cutlass_kernels/include -I/tmp/pip-req-build-jgyvqov6/csrc/cutlass_extensions/include -I/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include -I/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include/torch/csrc/api/include -I/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include/TH -I/home/michael/miniconda3/envs/ai_newsletters/lib/python3.12/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/michael/miniconda3/envs/ai_newsletters/include/python3.12 -c -c /tmp/pip-req-build-jgyvqov6/csrc/cutlass_kernels/fpA_intB_gemm_wrapper.cu -o /tmp/pip-req-build-jgyvqov6/build/temp.linux-x86_64-cpython-312/tmp/pip-req-build-jgyvqov6/csrc/cutlass_kernels/fpA_intB_gemm_wrapper.o -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options ''"'"'-fPIC'"'"'' -O3 -std=c++17 -U__CUDA_NO_HALF_OPERATORS__ -U__CUDA_NO_HALF_CONVERSIONS__ -U__CUDA_NO_HALF2_OPERATORS__ -U__CUDA_NO_HALF2_CONVERSIONS__ -DTORCH_API_INCLUDE_EXTENSION_H '-DPYBIND11_COMPILER_TYPE="_gcc"' '-DPYBIND11_STDLIB="_libstdcpp"' '-DPYBIND11_BUILD_ABI="_cxxabi1011"' -DTORCH_EXTENSION_NAME=EETQ -D_GLIBCXX_USE_CXX11_ABI=0 -gencode=arch=compute_70,code=sm_70 -gencode=arch=compute_75,code=sm_75 -gencode=arch=compute_80,code=sm_80 -gencode=arch=compute_86,code=compute_86 -gencode=arch=compute_86,code=sm_86
…184 | #define NVTX_VERSION 3
|
/usr/local/cuda/include/nvToolsExt.h(248): error: invalid redeclaration of type name "nvtxStringHandle_t" (declared at line 292 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef struct nvtxStringHandle* nvtxStringHandle_t;
^
/usr/local/cuda/include/nvToolsExt.h(258): error: invalid redeclaration of type name "nvtxDomainHandle_t" (declared at line 278 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef struct nvtxDomainHandle* nvtxDomainHandle_t;
^
/usr/local/cuda/include/nvToolsExt.h(273): error: enum "nvtxColorType_t" has already been defined (previous definition at line 302 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef enum nvtxColorType_t
^
/usr/local/cuda/include/nvToolsExt.h(275): error: "NVTX_COLOR_UNKNOWN" has already been declared in the current scope
NVTX_COLOR_UNKNOWN = 0,
^
/usr/local/cuda/include/nvToolsExt.h(276): error: "NVTX_COLOR_ARGB" has already been declared in the current scope
NVTX_COLOR_ARGB = 1
^
/usr/local/cuda/include/nvToolsExt.h(277): error: invalid redeclaration of type name "nvtxColorType_t" (declared at line 306 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
} nvtxColorType_t;
^
/usr/local/cuda/include/nvToolsExt.h(282): error: enum "nvtxMessageType_t" has already been defined (previous definition at line 311 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef enum nvtxMessageType_t
^
/usr/local/cuda/include/nvToolsExt.h(284): error: "NVTX_MESSAGE_UNKNOWN" has already been declared in the current scope
NVTX_MESSAGE_UNKNOWN = 0,
^
/usr/local/cuda/include/nvToolsExt.h(285): error: "NVTX_MESSAGE_TYPE_ASCII" has already been declared in the current scope
NVTX_MESSAGE_TYPE_ASCII = 1,
^
/usr/local/cuda/include/nvToolsExt.h(286): error: "NVTX_MESSAGE_TYPE_UNICODE" has already been declared in the current scope
NVTX_MESSAGE_TYPE_UNICODE = 2,
^
/usr/local/cuda/include/nvToolsExt.h(288): error: "NVTX_MESSAGE_TYPE_REGISTERED" has already been declared in the current scope
NVTX_MESSAGE_TYPE_REGISTERED = 3
^
/usr/local/cuda/include/nvToolsExt.h(291): error: invalid redeclaration of type name "nvtxMessageType_t" (declared at line 320 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
} nvtxMessageType_t;
^
/usr/local/cuda/include/nvToolsExt.h(293): error: type "nvtxMessageValue_t" has already been defined (previous definition at line 328 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef union nvtxMessageValue_t
^
/usr/local/cuda/include/nvToolsExt.h(481): error: enum "nvtxPayloadType_t" has already been defined (previous definition at line 368 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef enum nvtxPayloadType_t
^
/usr/local/cuda/include/nvToolsExt.h(483): error: "NVTX_PAYLOAD_UNKNOWN" has already been declared in the current scope
NVTX_PAYLOAD_UNKNOWN = 0,
^
/usr/local/cuda/include/nvToolsExt.h(484): error: "NVTX_PAYLOAD_TYPE_UNSIGNED_INT64" has already been declared in the current scope
NVTX_PAYLOAD_TYPE_UNSIGNED_INT64 = 1,
^
/usr/local/cuda/include/nvToolsExt.h(485): error: "NVTX_PAYLOAD_TYPE_INT64" has already been declared in the current scope
NVTX_PAYLOAD_TYPE_INT64 = 2,
^
/usr/local/cuda/include/nvToolsExt.h(486): error: "NVTX_PAYLOAD_TYPE_DOUBLE" has already been declared in the current scope
NVTX_PAYLOAD_TYPE_DOUBLE = 3,
^
/usr/local/cuda/include/nvToolsExt.h(488): error: "NVTX_PAYLOAD_TYPE_UNSIGNED_INT32" has already been declared in the current scope
NVTX_PAYLOAD_TYPE_UNSIGNED_INT32 = 4,
^
/usr/local/cuda/include/nvToolsExt.h(489): error: "NVTX_PAYLOAD_TYPE_INT32" has already been declared in the current scope
NVTX_PAYLOAD_TYPE_INT32 = 5,
^
/usr/local/cuda/include/nvToolsExt.h(490): error: "NVTX_PAYLOAD_TYPE_FLOAT" has already been declared in the current scope
NVTX_PAYLOAD_TYPE_FLOAT = 6
^
/usr/local/cuda/include/nvToolsExt.h(491): error: invalid redeclaration of type name "nvtxPayloadType_t" (declared at line 378 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
} nvtxPayloadType_t;
^
/usr/local/cuda/include/nvToolsExt.h(570): error: type "nvtxEventAttributes_v2" has already been defined (previous definition at line 548 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef struct nvtxEventAttributes_v2
^
/usr/local/cuda/include/nvToolsExt.h(663): error: type "nvtxEventAttributes_t" has already been defined (previous definition at line 550 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef struct nvtxEventAttributes_v2 nvtxEventAttributes_t;
^
/usr/local/cuda/include/nvToolsExt.h(1091): error: enum "nvtxResourceGenericType_t" has already been defined (previous definition at line 978 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef enum nvtxResourceGenericType_t
^
/usr/local/cuda/include/nvToolsExt.h(1093): error: "NVTX_RESOURCE_TYPE_UNKNOWN" has already been declared in the current scope
NVTX_RESOURCE_TYPE_UNKNOWN = 0,
^
/usr/local/cuda/include/nvToolsExt.h(1094): error: "NVTX_RESOURCE_TYPE_GENERIC_POINTER" has already been declared in the current scope
NVTX_RESOURCE_TYPE_GENERIC_POINTER = ((((uint32_t)(1))<<16)|((uint32_t)(1))),
^
/usr/local/cuda/include/nvToolsExt.h(1095): error: "NVTX_RESOURCE_TYPE_GENERIC_HANDLE" has already been declared in the current scope
NVTX_RESOURCE_TYPE_GENERIC_HANDLE = ((((uint32_t)(1))<<16)|((uint32_t)(2))),
^
/usr/local/cuda/include/nvToolsExt.h(1096): error: "NVTX_RESOURCE_TYPE_GENERIC_THREAD_NATIVE" has already been declared in the current scope
NVTX_RESOURCE_TYPE_GENERIC_THREAD_NATIVE = ((((uint32_t)(1))<<16)|((uint32_t)(3))),
^
/usr/local/cuda/include/nvToolsExt.h(1097): error: "NVTX_RESOURCE_TYPE_GENERIC_THREAD_POSIX" has already been declared in the current scope
NVTX_RESOURCE_TYPE_GENERIC_THREAD_POSIX = ((((uint32_t)(1))<<16)|((uint32_t)(4)))
^
/usr/local/cuda/include/nvToolsExt.h(1098): error: invalid redeclaration of type name "nvtxResourceGenericType_t" (declared at line 985 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
} nvtxResourceGenericType_t;
^
/usr/local/cuda/include/nvToolsExt.h(1177): error: type "nvtxResourceAttributes_v0" has already been defined (previous definition at line 1121 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef struct nvtxResourceAttributes_v0
^
/usr/local/cuda/include/nvToolsExt.h(1236): error: type "nvtxResourceAttributes_t" has already been defined (previous definition at line 1123 of /usr/local/cuda/include/nvtx3/nvToolsExt.h)
typedef struct nvtxResourceAttributes_v0 nvtxResourceAttributes_t;
^
/usr/local/cuda/include/nvToolsExt.h(1502): error: cannot overload functions distinguished by return type alone
nvtxDomainHandle_t nvtxDomainCreateA(const char* name);
^
/usr/local/cuda/include/nvToolsExt.h(1503): error: cannot overload functions distinguished by return type alone
nvtxDomainHandle_t nvtxDomainCreateW(const wchar_t* name);
^
35 errors detected in the compilation of "/tmp/pip-req-build-jgyvqov6/csrc/cutlass_kernels/fpA_intB_gemm_wrapper.cu".
dpkg --list | grep cuda-
ii cuda-12-6 12.6.0-1 amd64 CUDA 12.6 meta-package
ii cuda-cccl-12-6 12.6.37-1 amd64 CUDA CCCL
ii cuda-command-line-tools-12-6 12.6.0-1 amd64 CUDA command-line tools
ii cuda-compiler-12-6 12.6.0-1 amd64 CUDA compiler
ii cuda-crt-12-6 12.6.20-1 amd64 CUDA crt
ii cuda-cudart-12-6 12.6.37-1 amd64 CUDA Runtime native Libraries
ii cuda-cudart-dev-12-6 12.6.37-1 amd64 CUDA Runtime native dev links, headers
ii cuda-cuobjdump-12-6 12.6.20-1 amd64 CUDA cuobjdump
ii cuda-cupti-12-6 12.6.37-1 amd64 CUDA profiling tools runtime libs.
ii cuda-cupti-dev-12-6 12.6.37-1 amd64 CUDA profiling tools interface.
ii cuda-cuxxfilt-12-6 12.6.20-1 amd64 CUDA cuxxfilt
ii cuda-demo-suite-12-6 12.6.37-1 amd64 Demo suite for CUDA
ii cuda-documentation-12-6 12.6.37-1 amd64 CUDA documentation
ii cuda-driver-dev-12-6 12.6.37-1 amd64 CUDA Driver native dev stub library
ii cuda-gdb-12-6 12.6.37-1 amd64 CUDA-GDB
ii cuda-keyring 1.1-1 all GPG keyring for the CUDA repository
ii cuda-libraries-12-6 12.6.0-1 amd64 CUDA Libraries 12.6 meta-package
ii cuda-libraries-dev-12-6 12.6.0-1 amd64 CUDA Libraries 12.6 development meta-package
ii cuda-nsight-12-6 12.6.20-1 amd64 CUDA nsight
ii cuda-nsight-compute-12-6 12.6.0-1 amd64 NVIDIA Nsight Compute
ii cuda-nsight-systems-12-6 12.6.0-1 amd64 NVIDIA Nsight Systems
ii cuda-nvcc-12-6 12.6.20-1 amd64 CUDA nvcc
ii cuda-nvdisasm-12-6 12.6.20-1 amd64 CUDA disassembler
ii cuda-nvml-dev-12-6 12.6.37-1 amd64 NVML native dev links, headers
ii cuda-nvprof-12-6 12.6.37-1 amd64 CUDA Profiler tools
ii cuda-nvprune-12-6 12.6.20-1 amd64 CUDA nvprune
ii cuda-nvrtc-12-6 12.6.20-1 amd64 NVRTC native runtime libraries
ii cuda-nvrtc-dev-12-6 12.6.20-1 amd64 NVRTC native dev links, headers
ii cuda-nvtx-12-6 12.6.37-1 amd64 NVIDIA Tools Extension
ii cuda-nvvm-12-6 12.6.20-1 amd64 CUDA nvvm
ii cuda-nvvp-12-6 12.6.37-1 amd64 CUDA Profiler tools
ii cuda-opencl-12-6 12.6.37-1 amd64 CUDA OpenCL native Libraries
ii cuda-opencl-dev-12-6 12.6.37-1 amd64 CUDA OpenCL native dev links, headers
ii cuda-profiler-api-12-6 12.6.37-1 amd64 CUDA Profiler API
ii cuda-runtime-12-6 12.6.0-1 amd64 CUDA Runtime 12.6 meta-package
ii cuda-sanitizer-12-6 12.6.34-1 amd64 CUDA Sanitizer
ii cuda-toolkit 12.6.0-1 amd64 CUDA Toolkit meta-package
ii cuda-toolkit-12-6 12.6.0-1 amd64 CUDA Toolkit 12.6 meta-package
ii cuda-toolkit-12-6-config-common 12.6.37-1 all Common config package for CUDA Toolkit 12.6.
ii cuda-toolkit-12-config-common 12.6.37-1 all Common config package for CUDA Toolkit 12.
ii cuda-toolkit-config-common 12.6.37-1 all Common config package for CUDA Toolkit.
ii cuda-tools-12-6 12.6.0-1 amd64 CUDA Tools meta-package
ii cuda-visual-tools-12-6 12.6.0-1 amd64 CUDA visual tools
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I'm trying to install EETQ in a Google Colab notebook using the following command:
pip install git+https://github.com/NetEase-FuXi/EETQ.git. When installing, all steps seem to work, except that the installation gets stuck atBuilding wheels for collected packages: EETQ- it's been at this stage for over 10 minutes now. What should I do to fix this?P.S., the same problem happens when I try to install with the method as described in the readme.
The text was updated successfully, but these errors were encountered: