Parallelisation with compilation options only ? (FORTRAN)

Hi,

I am currently, working on a complex code. In this way, it is out of question to modify it, even if with flags.

That the reason why I would like to know if it would be possible to access to the Nvidia’s graphics card resources without any modification of the code itself. In other words only with compilation options.

Thank,

Have a nice day.

Hi AGy,

You may want to investigate the use of the PGI Accelerator Model (http://www.pgroup.com/resources/accel.htm) which uses directives to outline the portions of the code you wish to off-load to the GPU.

Note that programing for a GPU is very different than programming for a CPU. For complex code, you will most likely need or want to make modifications to your source for both portability and performance.

  • Mat

Hi Mat,

Thank you for the confirmation.