Speaker: Jaspal Subhlok
High Performance Fortran - Past, Present and Future
Date: February 16, 1995
Abstract: High Performance Fortran (HPF) is the only hope for a standard, portable, and widely accepted language for today's parallel machines, at least in the near future. The HPF Forum, which had its first meeting in Jan 92 has accomplished a fair amount in a short time. The HPF language document was released in June 1993 and several parallel machines and third party vendors are selling or developing HPF products. However, I believe that the HPF Effort is at crossroads and facing several challenges. It is difficult to compile HPF efficiently and the available compilers have not proven to be particularly successful. At the same time, many users complain that HPF simply does not have enough functionality for their applications, but adding more functionality to HPF would make it even more difficult to build good compilers. Further, it is not clear if the target market, basically scientific computing, is large enough to justify the efforts and costs. Also, an increasing number of users believe that C/C++ is the way to go, but that is a whole different story, and the problems there seem even more daunting.
I will present the basics of HPF Fortran and discuss the above issues.