DPCT use in cuda header files (.cuh).

[MountainLogic]

What is the correct way to handle #include xxx.cuh file from a cuda kernel with the DPCT tool?

I have a single kernel.cu file that I a porting to sycl using DPCT that has a cuda include file, kernel.cuh.  I seem to be getting a correct port, except the kernal.dp.cpp file keeps the "#include kernel.cuh"  

I also seem to be having a problem with device constants such as 

__device__ __constant__ int d_foo;

__device__ __constant__ float d_bar;

these see to be ignored by DCPT

[NoorjahanSk_Intel]

Hi,

Thanks for reaching out to us.

>>What is the correct way to handle #include xxx.cuh file from a cuda kernel with the DPCT tool?

 Header files which are located in same directory of source files are migratable when specified by --in-root.

 We can also use --extra-arg flag which is used to include header files explicitly.

 For more details you can refer below link:

 https://software.intel.com/content/www/us/en/develop/documentation/intel-dpcpp-compatibility-tool-user-guide/top/migrate-a-project/migrate-a-project-on-linux.html

>> problem with device constants

Please try above methods during migration of header files(.cuh), if the issue still persists please provide us a sample reproducer and steps you have followed to migrate.

Thanks & Regards

Noorjahan.

[NoorjahanSk_Intel]

Hi,

 Has the information provided helped? If not, then could you please provide the above-mentioned details.

Thanks & Regards

Noorjahan.

[NoorjahanSk_Intel]

Hi,

I have not heard back from you. This thread will no longer be monitored by Intel. If you need further assistance, please post a new question

Thanks & Regards

Noorjahan.