Hello,
I think that this is a basic question but i did not find an answer:
did pgcc know *.cu file ? option to do it?
if not how to do to compile native CUDA code?
Hi bilel,
The you need to use C++ dirver, pgcpp (or pgCC), to compile CUDA X86 files. Add the flag “-Mcuda” if the file does not use the “.cu” file extension.
Hope this helps,
Mat
hello,
thank you for your Answer but my question is the flowing:
I have a good cuda file: test.cu , can compile it using nvcc and the output run on the GPU
Can I compile using the PGI suite in order to run on the GPU too?
Tks again,
bilel
Hi bilel,
Can I compile using the PGI suite in order to run on the GPU too?
Not yet. The first generation CUDA X86 product only targets multi-core x86.
Full details can be found here:Key Technologies | PGI
- Mat