Skip to content

c10::Scalar macOS build error on CUDA 10.1, cuDNN 7.6.5 for PyTorch version 1.8 #48903

Closed
Closed
c10::Scalar macOS build error on CUDA 10.1, cuDNN 7.6.5 for PyTorch version 1.8#48903
Labels
module: buildBuild system issuesmodule: cudaRelated to torch.cuda, and CUDA support in generalmodule: macosMac OS related issuestriagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate module
@jackdyson31

Description

@jackdyson31
jackdyson31
opened on Dec 6, 2020

Issue description

Building pyTorch on macOS High Sierra with CUDA 10.1 and cuDNN 7.6.5 (for CUDA 10.1) installed and running on an NVIDIA GTX 650M GPU fails at 77% with 24 errors connected to converting from long type to c10::Scalar type.

The problem seems to be connected to type conversions in lines 74-80 of file BatchLinearAlgebraLib.cu. The command where that happens is :

    Tensor self_array = at::arange(
      reinterpret_cast<long>(self_data),
      reinterpret_cast<long>(&self_data[(batch_size-1) * self_mat_stride]) + 1,
      static_cast<long>(self_mat_stride * sizeof(scalar_t)), self.options().dtype(at::kLong));
    Tensor self_inv_array = at::arange(
      reinterpret_cast<long>(self_inv_data),
      reinterpret_cast<long>(&self_inv_data[(batch_size-1) * self_inv_mat_stride]) + 1,
      static_cast<long>(self_inv_mat_stride * sizeof(scalar_t)), self.options().dtype(at::kLong));

I submit here in case others have seen this or in the hope that it has a solution.

To Reproduce the error

  1. Check that you are running Mac OS X High Sierra (10.13.6)

  2. Xcode version =10.1

  3. CUDA Driver Version: 418.163 from NVIDIA Web Driver Version =387.10.10.10.40.140

  4. .profile should have these defined:

export PATH=/Developer/NVIDIA/CUDA-10.0/bin${PATH:+:${PATH}}
export DYLD_LIBRARY_PATH=/usr/local/cuda/lib:$DYLD_LIBRARY_PATH

  1. cuDNN should be 7.6.5 for MacOS

  2. use Miniconda 4.9.2

  3. conda create --name test python=3.7

  4. conda activate test

  5. conda install numpy pyyaml mkl mkl-include setuptools cmake cffi typing

  6. Then:

git clone --recursive https://github.com/pytorch/pytorch
cd pytorch

  1. Then:

export CMAKE_PREFIX_PATH=${CONDA_PREFIX:-"$(dirname $(which conda))/../"}
MACOSX_DEPLOYMENT_TARGET=10.9 CC=clang CXX=clang++ python setup.py install

PS This can be fixed (see comment below) but before it completes it will stop again at 95% with a linker error which has been fixed by Tom Heaven in his report #42271.

Code example

The 24 errors produced are generic and look like this:

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(74): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=double]" 
(115): here... 

The full listing of the errors is:

[ 77%] Building NVCC (Device) object caffe2/CMakeFiles/torch_cuda.dir/__/aten/src/ATen/native/cuda/torch_cuda_generated_BinaryBitwiseOpsKernels.cu.o
/Users/jack/tmp/pytorch_cuda_build/pytorch/c10/util/BFloat16.h(57): warning: calling a __host__ function from a __host__ __device__ function is not allowed

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(74): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=double]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(75): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=double]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(76): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=double]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(78): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=double]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(79): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=double]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(80): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=double]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(74): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=float]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(75): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=float]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(76): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=float]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(78): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=float]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(79): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=float]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(80): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=float]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(74): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<double>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(75): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<double>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(76): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<double>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(78): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<double>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(79): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<double>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(80): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<double>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(74): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<float>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(75): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<float>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(76): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<float>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(78): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<float>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(79): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<float>]" 
(115): here

/Users/jack/tmp/pytorch_cuda_build/pytorch/aten/src/ATen/native/cuda/BatchLinearAlgebraLib.cu(80): error: more than one constructor applies to convert from "long" to "c10::Scalar":
            function "c10::Scalar::Scalar(uint8_t)"
            function "c10::Scalar::Scalar(int8_t)"
            function "c10::Scalar::Scalar(int16_t)"
            function "c10::Scalar::Scalar(int)"
            function "c10::Scalar::Scalar(int64_t)"
            function "c10::Scalar::Scalar(float)"
            function "c10::Scalar::Scalar(double)"
          detected during instantiation of "void at::native::apply_batched_inverse_lib<scalar_t>(at::Tensor &, at::Tensor &, at::Tensor &) [with scalar_t=c10::complex<float>]" 
(115): here

24 errors detected in the compilation of "/var/folders/yq/8p61d19111s1p7x1ny_czdr00000gn/T//tmpxft_0000ebbe_00000000-6_BatchLinearAlgebraLib.cpp1.ii".
CMake Error at torch_cuda_generated_BatchLinearAlgebraLib.cu.o.Release.cmake:281 (message):
  Error generating file
  /Users/jack/tmp/pytorch_cuda_build/pytorch/build/caffe2/CMakeFiles/torch_cuda.dir/__/aten/src/ATen/native/cuda/./torch_cuda_generated_BatchLinearAlgebraLib.cu.o


make[2]: *** [caffe2/CMakeFiles/torch_cuda.dir/__/aten/src/ATen/native/cuda/torch_cuda_generated_BatchLinearAlgebraLib.cu.o] Error 1
make[2]: *** Waiting for unfinished jobs....
/Users/jack/tmp/pytorch_cuda_build/pytorch/c10/util/BFloat16.h(57): warning: calling a __host__ function("isnan<float> ") from a __host__ __device__ function("c10::detail::round_to_nearest_even") is not allowed

/Users/jack/tmp/pytorch_cuda_build/pytorch/c10/util/BFloat16.h(57): warning: calling a __host__ function from a __host__ __device__ function is not allowed

/Users/jack/tmp/pytorch_cuda_build/pytorch/c10/util/BFloat16.h(57): warning: calling a __host__ function("isnan<float> ") from a __host__ __device__ function("c10::detail::round_to_nearest_even") is not allowed

/Users/jack/tmp/pytorch_cuda_build/pytorch/c10/util/BFloat16.h(57): warning: calling a __host__ function("isnan<float> ") from a __host__ __device__ function("c10::detail::round_to_nearest_even") is not allowed

make[1]: *** [caffe2/CMakeFiles/torch_cuda.dir/all] Error 2
make: *** [all] Error 2
Traceback (most recent call last):
  File "setup.py", line 773, in <module>
    build_deps()
  File "setup.py", line 320, in build_deps
    cmake=cmake)
  File "/Users/jack/tmp/pytorch_cuda_build/pytorch/tools/build_pytorch_libs.py", line 58, in build_caffe2
    cmake.build(my_env)
  File "/Users/jack/tmp/pytorch_cuda_build/pytorch/tools/setup_helpers/cmake.py", line 346, in build
    self.run(build_args, my_env)
  File "/Users/jack/tmp/pytorch_cuda_build/pytorch/tools/setup_helpers/cmake.py", line 141, in run
    check_call(command, cwd=self.build_dir, env=env)
  File "/opt/miniconda3/envs/torch/lib/python3.7/subprocess.py", line 363, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['cmake', '--build', '.', '--target', 'install', '--config', 'Release', '--', '-j', '8']' returned non-zero exit status 2.

System Info

Collecting environment information...
PyTorch version: N/A
Is debug build: No
CUDA used to build PyTorch: 7.6.5
ROCM used to build PyTorch: N/A

OS: macOS 10.13.6 (x86_64)
GCC version: Apple LLVM version 10.0.0 (clang-1000.11.45.5)
Clang version: Apple LLVM version 10.0.0 (clang-1000.11.45.5)
CMake version: version 3.18.2

Python version: 3.7 (64-bit runtime)
Is CUDA available:Yes
CUDA runtime version: Apple LLVM version 10.0.0 (clang-1000.11.45.5)
GPU models and configuration: GeForce 650M GTX NVIDIA with CUDA Driver 418.163
Nvidia driver version: 1.1.0
cuDNN version: cudnn-10.1-osx-x64-v7.6.5.32
Probably one of the following:
/usr/local/cuda/lib/libcudnn.7.dylib
/usr/local/cuda/lib/libcudnn_static.a
HIP runtime version: N/A
MIOpen runtime version: N/A

Versions of relevant libraries:
[pip3] numpy==1.19.2
[conda] mkl                       2020.2                      260  
[conda] mkl-include               2020.2                      260  
[conda] numpy                     1.19.2           py37h63973fd_0  
[conda] numpy-base                1.19.2           py37h68fea81_0

cc @malfet @seemethere @walterddr @ngimel

Metadata

Metadata

Assignees

No one assigned

    Labels

    module: buildBuild system issuesmodule: cudaRelated to torch.cuda, and CUDA support in generalmodule: macosMac OS related issuestriagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions