Build fails for me, ubuntu 16.04, with CUDA, Thrust include error

wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
source ~/conda3/bin/activate
git clone https://github.com/pytorch/pytorch
cd pytorch
source ~/conda3/bin/activate
conda create -n pytorch-dev python=3.6
source activate pytorch-dev
conda install --file requirements.txt
python setup.py install

What I expect to happen:

pytorch was built and installed successfully

What actually happens:

/usr/local/cuda/include/thrust/detail/reference.inl(127): error: no default constructor exists for class "thrust::detail
::execute_with_allocator<THCThrustAllocator, thrust::system::cuda::detail::execute_on_stream_base>"
          detected during:
            instantiation of "void thrust::reference<Element, Pointer, Derived>::assign_from(OtherPointer) [with Element
=signed long, Pointer=thrust::pointer<signed long, thrust::detail::execute_with_allocator<THCThrustAllocator, thrust::sy
stem::cuda::detail::execute_on_stream_base>, thrust::use_default, thrust::use_default>, Derived=thrust::use_default, Oth
erPointer=const signed long *]"
(65): here
            instantiation of "thrust::reference<Element, Pointer, Derived>::derived_type &thrust::reference<Element, Poi
nter, Derived>::operator=(const thrust::reference<Element, Pointer, Derived>::value_type &) [with Element=signed long, P
ointer=thrust::pointer<signed long, thrust::detail::execute_with_allocator<THCThrustAllocator, thrust::system::cuda::det
ail::execute_on_stream_base>, thrust::use_default, thrust::use_default>, Derived=thrust::use_default]"
/usr/local/cuda/include/thrust/system/cuda/detail/reduce_intervals.inl(150): here

Some diag:

ls -lh /usr/local
ls -lh /usr/local/include
gcc --version
cat /etc/lsb-release

Output:

(pytorch-dev) (master|✔) ~/git/pytorch$ ls -lh /usr/local
total 48K
drwxr-xr-x  2 root   root   4.0K Jun 22 09:48 bin
lrwxrwxrwx  1 root   root     19 Sep 25  2016 cuda -> /usr/local/cuda-8.0
drwxr-xr-x 15 root   root   4.0K Oct 13  2016 cuda-7.5
drwxrwxr-x 17 ubuntu ubuntu 4.0K Sep 25  2016 cuda-8.0
drwxr-xr-x  2 root   root   4.0K May 31 09:17 cuda_hidden
drwxr-xr-x  2 root   root   4.0K Sep  7  2016 etc
drwxr-xr-x  2 root   root   4.0K Sep  7  2016 games
drwxr-xr-x  2 root   root   4.0K Jun 22 09:47 include
drwxr-xr-x  5 root   root   4.0K Jun 21 00:50 lib
lrwxrwxrwx  1 root   root      9 Sep  7  2016 man -> share/man
drwxr-xr-x  3 root   root   4.0K Jun 27 08:27 opt
drwxr-xr-x  2 root   root   4.0K Sep  7  2016 sbin
drwxr-xr-x  8 root   root   4.0K Jun 27 10:16 share
drwxr-xr-x  2 root   root   4.0K Sep  7  2016 src
(pytorch-dev) (master|✔) ~/git/pytorch$ ls -lh /usr/local/include
total 0
(pytorch-dev) (master|✔) ~/git/pytorch$ gcc --version
gcc (Ubuntu 5.4.0-6ubuntu1~16.04.4) 5.4.0 20160609
Copyright (C) 2015 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

(pytorch-dev) (master|✔) ~/git/pytorch$ cat /etc/lsb-release
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=16.04
DISTRIB_CODENAME=xenial
DISTRIB_DESCRIPTION="Ubuntu 16.04.2 LTS"

Thoughts?

are you using CUDA 8.0.26 (the RC version of CUDA8)? If so this is a known problem, and you need to upgrade CUDA 8.0 to atleast 8.0.44 (final)

1 Like

I have the same error and cuda version is 9.0.176