I have problems running the prerelease version of MIT GCM C19 at a Linux PC cluster. The output fields of a forwrad run are unrealistic and differ from the ones obtained at IRIX64. I am wondering if anyone has ever run this version on a PC cluster? Thanks Elena