dpct exited with code -5

[Shukla__Gagandeep]

Hi,

I was converting knncuda.cu from kNN-CUDA repo using latest dpct (beta08) but I'm constantly getting the error:

dpct exited with code: -5 (Error: Path for CUDA header files is invalid or not available. Specify with --cuda-include-path)

even when I specify cuda path in the command

Command:

dpct --cuda-include-path=/usr/local/cuda/include --out-root=dpct knncuda.cu

dpct version: 2021.1-beta08

System: Ubuntu 18.04.4 LTS

I have installed cuda (release11.0, V11.0.194) and cuda include files are present in /usr/local/cuda/include folder. I am able to run intercept-build to generate commpile_commands.json file andable to build application using nvcc.

I even reinstalled cuda and oneapi base toolkit again but I am still getting the same error.

What can cause dpct to fail with this error? Any suggestions?

Regards,
Gagan

[RahulV_intel · Accepted solution]

Hi Gagan,

CUDA version 11.0 is currently not supported as per the latest DPCT system requirements. The supported CUDA versions are 8.0, 9.x, 10.1, 10.2.

Kindly refer to the DPCT system requirements link for more details:

https://software.intel.com/content/www/us/en/develop/articles/intel-dpc-compatibility-tool-system-requirements.html

I have not come across any issues while migrating knncuda.cu using CUDA 10.2 headers.

Let me know if it helps.

Regards,

Rahul

[Shukla__Gagandeep]

Hi Rahul,

Sorry, my bad. I did not check the requirements and installed wrong version.
I installed CUDA 10.2 now and dpct is working fine.

Regards,
Gagan

[RahulV_intel]

Hi Gagan,

Good to know that it works.

Let me know if I can close the thread from my end.

Thanks,

Rahul

[RahulV_intel]

Intel will no longer monitor this thread. However, this thread will remain open for any community discussion.