cuda 8.0 in docker: nvcc not found

Hi,

I need to run a docker container with cuda to build some binary.
I took this image

8.0-runtime-ubuntu16.04

from the official dockerhub space https://hub.docker.com/r/nvidia/cuda/
The content of my dockerfile is basicly this simple:

FROM nvidia/cuda:8.0-runtime-ubuntu16.04

RUN apt-get update && apt-get install -y \
	 autoconf \
	 libtool

ENTRYPOINT ["sh", "-c", "run.sh && make"]

But I get an error: sh: 1: nvcc: not found

I checked this forum and checked the PATH and LD_LIBRARY_PATH are in place.
$ echo PATH /usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin echo $LD_LIBRARY_PATH
/usr/local/nvidia/lib:/usr/local/nvidia/lib64

What I noticed though that in the image there is no bin directory under /usr/local/cuda-8.0

$ ls -la /usr/local/cuda-8.0
drwxr-xr-x  4 root root 4096 Jun 20 23:44 .
drwxr-xr-x 13 root root 4096 Jun 27 14:22 ..
-rw-r--r--  1 root root  365 Jan 26 23:48 LICENSE
-rw-r--r--  1 root root  365 Jan 26 23:48 README
drwxr-xr-x  2 root root 4096 Jun 20 23:44 doc
lrwxrwxrwx  1 root root   24 Apr  6 22:24 lib64 -> targets/x86_64-linux/lib
drwxr-xr-x  3 root root 4096 Jun 20 23:44 targets
-rw-r--r--  1 root root   20 Jan 26 23:48 version.txt

And also there is no nvidia directory under /usr/local/

Could someone help me make this docker image operable?
Or is the docker image intended for something else?

Best regards,
Oleksii

Hello,

If you want nvcc, you should use 8.0-devel-ubuntu16.04.
The runtime images don’t provide the compiler toolchain.

Alternatively, you could also “apt-get install -y cuda-core-8-0” from your own Dockerfile.