I'm trying to run a very simple openACC program, but cannot get it to compile properly. I am running Ubuntu 20.04 on Windows 10.
First, I tried installing the pgcc compiler found at . After unpacking the .tar file, I ran install --help, because I am very unfamiliar with installation on Ubuntu. The help message contains the following:
This install program copies files (often just compiled) into destination locations you choose. If you want to download and install a ready-to-use package on a GNU/Linux system, you should instead be using a package manager like yum(1) or apt-get(1).
If I understand this correctly, I need to install the compiler using the package manager. However, I have no idea which command I should run in order to do this.
When I run the installation regardless, it gets installed in a sub-folder of the folder I am in. I have no idea where I should place these files, and how to access the pgcc compiler elsewhere.
Having had no success with the NVIDIA package, I tried using the GNU gcc compiler with the -fopenacc flag: . However, I get the following error:
lto-wrapper: fatal error: could not find accel/nvptx-none/mkoffload in /usr/lib/gcc/x86_64-linux-gnu/9/:/usr/lib/gcc/x86_64-linux-gnu/9/:/usr/lib/gcc/x86_64-linux-gnu/:/usr/lib/gcc/x86_64-linux-gnu/9/:/usr/lib/gcc/x86_64-linux-gnu/ (consider using ‘-B’)
If I understand this correctly, this has to do with offloading not being configured, as indicated here in the "Typical Compiler Usage" section: . Trying to figure out how to configure this properly leads me to a dead end.
At this point I am getting incredibly frustrated, as I've been trying to get an incredibly simple program to compile for several hours. I'm following a course on parallel computing, and openMP, MPI and Pthreads were all a matter of including a library or installing a package. I have no idea why it is so difficult to get a working openACC compiler...
3 Reset to default
