Hello, I’m trying to compile a CUDA program for the very first time. I’m using one of the NVIDIA examples and their makefile. I’m getting errors I don’t understand. I think I need pointers on how to get started.
OBJECTIVE: I don’t care what example I compile or if it will run on may macbookpro, But I do want to be able to compile something as a way to get started.
mac book pro core duo 2
256MB Nvidia 8600 GT (should work accroding to NVIDIA list)
Software platform preparation:
installed mac developer tools
installed g95 for good measure from macports
Ran installer NVIDIA_CUDA_SDK_1.1_Beta_MACOSX , rebooted
What I tried:
downloaded Fortran cuda Blas
run a tsch
cd into this directory. run make without ANY edits to the makefile.
~/src/cuda/Fortran_Cuda_Blas] cems% make gcc -O3 -DCUBLAS_USE_THUNKING -I/usr/local/cuda/include Â -c Â fortran.c fortran.c:64:2: error: #error unsupported platform fortran.c:208:2: error: #error unsupported Fortran compiler make: *** [fortran.o] Error 1
Looking in the make file it is stumbling on the very first command:
gcc -O3 -DCUBLAS_USE_THUNKING -I/usr/local/cuda/include -c fortran.c fortran.c:64:2: error: #error unsupported platform fortran.c:208:2: error: #error unsupported Fortran compiler
since fortran.c is actually c-code not fortran, the cryptic unsupported fortran error hints that the problem lies in the included library.
Editing that command line to remove the include:
gcc -O3 -DCUBLAS_USE_THUNKING -c fortran.c
I get a wad of missing function errors but no unssuported platform error.
SO anyhow, I don’t care what example I compile or if it will run on may macbookpro, But I do want to be able to compile something as a way to get started.
Is my system misconfigured, do I need to edit the makefile? Should I try a different example, or what?