Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

CL_VERSION_2_0 for Nvida on Ubuntu 16.04? #237

Open
mw66 opened this issue Apr 7, 2018 · 4 comments
Open

CL_VERSION_2_0 for Nvida on Ubuntu 16.04? #237

mw66 opened this issue Apr 7, 2018 · 4 comments

Comments

@mw66
Copy link

mw66 commented Apr 7, 2018

Hi,

I saw in tests/svm.cpp

#if defined(CL_VERSION_2_0) || defined(VEXCL_BACKEND_CUDA)

I'm wondering where to get the CL_VERSION_2_0 dev files installed for Nvidia on Ubuntu 16.04?

Thanks.

============
$ grep -rIli CL_VERSION_2 /usr/local/cuda-*

shows nothing.

$ ls -lartd /usr/local/cuda-*
drwxr-xr-x 14 root root 4096 Mar 30 2017 /usr/local/cuda-8.0
drwxr-xr-x 11 root root 4096 Jun 11 2017 /usr/local/cuda-7.5
drwxr-xr-x 15 root root 4096 Jan 31 00:43 /usr/local/cuda-9.1

looks like only 1.x is there.

$ grep -rIi CL_VERSION_1 /usr/local/cuda-* | grep -w define
/usr/local/cuda-7.5/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_0 1
/usr/local/cuda-7.5/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_1 1
/usr/local/cuda-7.5/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_2 1
/usr/local/cuda-8.0/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_0 1
/usr/local/cuda-8.0/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_1 1
/usr/local/cuda-8.0/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_2 1
/usr/local/cuda-9.1/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_0 1
/usr/local/cuda-9.1/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_1 1
/usr/local/cuda-9.1/targets/x86_64-linux/include/CL/cl.h:#define CL_VERSION_1_2 1

@ddemidov
Copy link
Owner

ddemidov commented Apr 7, 2018

The test you mentioned is defined either for OpenCL 2.0 or the CUDA backend. I don't think NVIDIA supports OpenCL 2.0.

@mw66
Copy link
Author

mw66 commented Apr 7, 2018

I can find the news, but not the dev files:

https://streamhpc.com/blog/2017-02-22/nvidia-enables-opencl-2-0-beta-support/

@mw66
Copy link
Author

mw66 commented Apr 7, 2018

is it this one ?

NVIDIA/nvidia-docker#17
We just published the OpenCL images on DockerHub (finally!).

https://hub.docker.com/r/nvidia/opencl/

looks like it is defined:

root@91dbfd615000:~# grep CL_VERSION_ /usr/include/CL/* | grep -w define
/usr/include/CL/cl.h:#define CL_VERSION_1_0 1
/usr/include/CL/cl.h:#define CL_VERSION_1_1 1
/usr/include/CL/cl.h:#define CL_VERSION_1_2 1
/usr/include/CL/cl.h:#define CL_VERSION_2_0 1

@henryiii
Copy link
Contributor

henryiii commented Apr 16, 2018

On the latest NVidia drivers, 1.2 is the highest level officially supported. They've been adding bits of 2.0 support as developer previews for a year or so now as technical previews, but nothing official. (See the section: Experimental OpenCL 2.0, "Select features in OpenCL 2.0 are available in the driver for evaluation purposes only.")

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants