DPCT error: dpct exited with error code -5

[Siddhanta_Shroff]

Hello,

I am trying to follow the steps given on https://software.intel.com/content/www/us/en/develop/documentation/get-started-with-intel-dpcpp-compatibility-tool/top.html 

Upon compilation, I get an error saying the path for CUDA header files is invalid. I am trying to use the compatibility tool on the devcloud server.

Now, to include the CUDA path I need to first install the CUDA toolkit but I cannot install it without sudo access. Could you please help me understand what is it that I can do to get it running on the server?

Thanks & regards,

Siddhanta

[RahulV_intel · Accepted solution]

Hi,

You need to install CUDA toolkit in your home directory (Home directory installation doesn't require sudo. You may find the instructions for the same on the Nvidia official webpage). Once you have installed the toolkit, use the below command to perform migration for a single source file:

dpct --cuda-include-path=/path/to/cuda/include vec_add.cu

This command will save the migrated files to the dpct_output directory by default.

Kindly refer to the DPCT user guide for more information:

https://software.intel.com/content/www/us/en/develop/documentation/intel-dpcpp-compatibility-tool-user-guide/top/usage-workflow-overview.html

For more than one source files, please note that you will have to generate a compile_commands.json file using your source application's Makefile (intercept-build make command) and perform the migration with the help of the generated json file. (Refer to the user guide for more info)

Please refer to the below link for supported CUDA versions and hardware/software requirements:

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

Thanks,

Rahul

[RahulV_intel]

Thanks for accepting the solution.

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