I just download PGI accelerator Fortran/C/C++ workstation. As far as I can tell, I followed the install instructions. However, when I try to compile the first accelerator fortran code, I got:
pgfortran -ta=nvidia -Minfo main.F90
pgfortran-Warning-CUDA is not available in this installation; -ta=analysis assumed
What could be the reason for this problem? Thanks!
My system:
Ubuntu 10.04, x86_64
CUDA 4.0,
2 GPU: C2050/Quadro 5800.
That message occurs because the ‘cuda’ directory cannot
be found under $PGI/linux86-64/2011, or
the $PGI/linux86/2011
Have you defined $PGI ( I am assuming this is a Linux install),
and is the directory present? If not, reinstall, and make sure you
said ‘yes’ and ‘accept’ to the license terms and agreements.
I decided to re-use this thread since I’m now having a similar problem with the most recent installation of the PGI 12.1 compiler for Linux.
Now, I did not install this personally - our sys admin did.
When I try to link I get the following complaint:
pgfortran-Error-CUDA version 4.0 is not available in this installation
pgfortran-Warning-CUDA version 4.0 is not available in this installation; -ta=analysis assumed
Is there any way for me to determine if CUDA was installed with the PGI installation?
The CUDA components we ship would be located in the “$PGI/linux86-64/2012/cuda” directory. If it doesn’t exist, then you’ll need to ask your admin to re-install the compilers, and select ‘yes’ when prompted if the CUDA components should be installed.