I am using pgi version 12.8
Enable for cuda and pgaccelerators.
But I am facing problems about the LD_LIBRARY_PATH.
#pgaccelinfo -v
libcuda.so not found
libamdcalcl.so not found
No accelerators found.
Check that you have installed the CUDA or CAL libraries properly
Check that your LD_LIBRARY_PATH environment variable points to the CUDA or CAL runtime installation directory
I could build the matrix multiplication example with accelator and build with the following options
pgfortran -fast -acc - Minf -g -pg matrx-acc.f90
Then after qsub to my cluster i got the gmoun.out
Then I tried to use the pgcollect and pgprof
pgcollect a.out
libcuda.so not found, exiting
Please check that the CUDA driver is installed and the shared object is in the install directory or on your LD_LIBRARY_PATH.
target process abnormal termination: Error: process received signal: Exception
no profile generated
The visual pgprof also failed.
Cuda is installed in my /usr/local but the cudalib.so does not exist there.
ls /usr/local/cuda/lib64/
libcublas.so libcufft.so libcurand.so libnpp.so
libcublas.so.4 libcufft.so.4 libcurand.so.4 libnpp.so.4
libcublas.so.4.2.9 libcufft.so.4.2.9 libcurand.so.4.2.9 libnpp.so.4.2.9
libcudart.so libcuinj.so libcusparse.so
libcudart.so.4 libcuinj.so.4 libcusparse.so.4
libcudart.so.4.2.9 libcuinj.so.4.2.9 libcusparse.so.4.2.9
Could you give any advice because this is critical issue for me to start parallelizing my project.
Thank you.
Koffi