Problem with RDC

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi, I am trying to run the following program.

https://github.com/yhf8377/cuda-fp-growth

I am working on CUDA 8.0 and Visual Studio 2013. When setting -rdc = false, I got the following error.

error : kernel launch from __device__ or __global__ functions requires separate compilation mode

However, after setting -rdc = true, I got the following errors.

error LNK2001: unresolved external symbol ___fatbinwrap_66_tmpxft_0000a66c_00000000_18_cuda_device_runtime_compute_61_cpp1_ii_8b1a5d37

error LNK2019: unresolved external symbol _main referenced in function ___tmainCRTStartup

How can I solve this problem? Thank you!

2

It’s not clear how you set -rdc=true

If you simply added that compile switch, that’s not the correct way to do it in visual studio. You should use the project configuration to change to a relocatable device code project. This should address the first (LNK2001) error.

The second error is arising due to the fact that this project has an unusual mechanism for providing a main() routine, and it is apparently not working. You’ll need to investigate that.

3

Thanks for the response. I set -rdc=true in Project → Properties → Configuration Properties → CUDA C/C++ → Common → Generate Relocatable Device Code, and this does not seem to address the LNK2001 error. Are there any steps that I am missing? Thanks!

4

I think my problem is the same as the one described in [url]c++ - CUDA Dynamic Parallelism error : LNK2001 - Stack Overflow. Hopefully this can help others with the same issue.