The Search for a new HPC language
By Vtatkar-Oracle on Aug 29, 2006
Getting new languages adopted into the market is pretty rare; only a few have succeeded over time: Fortran, COBOL, C, C++ and Java are perhaps the best known of this lot. Adoption and popularity have often led to such fierce loyalty that getting new programming languages adopted has become a very high-barrier endeavor.
DARPA, through its HPCS program, is attempting to deliver a paradigm shift. The theory goes that creating a new general-purpose language for high performance computing will offer a different way to think about the problem.
The latest issue of HPCwire examines the three new emerging languages proposed as part of the DARPA effort: Chapel (Cray), X10 (IBM) and Fortress (Sun) .
I think the following links are a must-read for anyone who is interested in whats emerging in the HPC market and how leading vendors are going to offer a way out of the current limitations.
The first one discusses the motivation to Search for a new HPC language .
The second is an interview with Rusky Lusk, Acting Division Director of the Mathematics and Computer Science Division at Argonne National Laboratory, about how this part of the HPCS program will proceed as Phase II comes to a close.
I cant do enough justice in this blog to the topic; the above links say much more than I can and in more detail.
Personal note: I am sometimes asked at customer meetings about the direction that Sun is going to propose for development in the emerging CMT era (because Sun, more than anyone else, is seen as a technical leader in this market). My answers often wander around MPI, OpenMP, Parallel language extensions such as UPC, HPF, etc. I also mention the DARPA efforts in this regard. Clearly all of these are long-term efforts, not quick fixes in the short term to exploit the dual-core, quad-core and multi-core chips.
PPS. Sun has had other bloggers also talk about this aspect of the DARPA/HPCS program. One of the most profilic blogs is the HPCS languages moving forward and about the search for a new HPCL language . I'd urge y'all to go to this site from time to time; it has many interesting nuggets that I have learned from.