Error running vector-add on NVIDIA GPU

[John_Johnson]

Hello everyone!

This thread is related to https://software.intel.com/en-us/forums/oneapi-data-parallel-c-compiler/topic/850410?language=en-us&https=1 and I wanted to write there, but post was locked.

My problem is that when running vector-add example on NVIDIA Tesla P100 example I get error:

Command I run:

$ SYCL_DEVICE_TYPE=GPU SYCL_DEVICE_WHITE_LIST="" SYCL_DEVICE_ALLOWLIST="" ./vector-add

Device: Tesla P100-PCIE-12GB
terminate called after throwing an instance of 'cl::sycl::runtime_error'
  what():  OpenCL API failed. OpenCL API returns: -34 (CL_INVALID_CONTEXT) -34 (CL_INVALID_CONTEXT)

Does someone have any about that?

[RahulV_intel]

Hi John,

Could you try running this sample on CPU and let us know if the issue persists?

What version of oneAPI base-toolkit are you using?

--Rahul

[John_Johnson]

On CPU inteloneapi works just fine. Command I use to run on CPU:

$ SYCL_DEVICE_TYPE=CPU ./vector-add

Device: Intel(R) Xeon(R) CPU E5-2630 v4 @ 2.20GHz
success

* My IntelOneApi version is beta04

[GouthamK_Intel]

Hi John,

Intel oneAPI currently doesn't support NVIDIA GPU's. 

Regards

Goutham

[John_Johnson]

Hmmm... I checked here and I see that it should support NVIDIA GPU's:

https://www.phoronix.com/scan.php?page=news_item&px=Codeplay-Does-SYCL-On-NVIDIA

[GouthamK_Intel]

Hi John,

Currently, Intel oneAPI basekit beta04 doesn't support Nvidia GPU's. 

The link which you referred is Codeplay's distribution of oneAPI So you may need to install it separately and work on it.

Regards

Goutham

[GouthamK_Intel]

Hi,

We hope the resolution provided helped you.

In case of any further queries please feel free to raise a new thread.

Thanks

Goutham