Compilation error

Accelerated Computing HPC Compilers Legacy PGI Compilers
1

Hello,

I installed PGI 14.10 on x86-64 CentOS, CUDA V5.5 and two nvidia geForce gtx 295. The pgCC compiler works fine, but when I include the file “cuda_profiler_api.h”:

#include <cuda_profiler_api.h>

in my C++ program, to use the functions:

cudaProfilerStart(); and cudaProfilerStop();

I get the following error from the pgCC compiler :

"acc_c1.cpp", line 15: catastrophic error: cannot open source file
          "cuda_profiler_api.h"
  #include <cuda_profiler_api.h>

I am compiling the program using the command line:

pgCC -fast -acc -ta=tesla:cc13 acc_c1.cpp -o executable

and I have also checked that $PATH and $LD_LIBRARY_PATH include /usr/local/cuda-5.5/bin and /usr/local/cuda-5.5/lib64. I found the file cuda_profiler_api.h at /usr/local/cuda-5.5/include and add this directory to the $PATH but still does not work.

2

You have a University development license, which allows you to use the
current releases of the compilers. pgCC is no longer a PGI product, and
you should be using pgc++ . OpenACC is an evolving technology, and the compilers and CUDA versions available now are better at achieving goals.

Install the 16.9 release compilers, and a current CUDA version.

dave