Slashdot Mirror

Apparently DIPC (Distributed IPC) can run with MOSIX, although DIPC a few months ago did not optimize migrated processes. It could work, but works better when DIPC realizes that processes are able to run on other systems.

• Beowulf is one of a family of parallel programming API tools. Programs must use the API to accomplish parallel programming.
• SCI is fast hardware with support for distributed shared memory, messaging, and data transfers. Again, if you don't use the API then no gain.
• DIPC is distributed System V IPC. Programs which use the IPC API can be converted to DIPC easily, such as just by adding the DIPC flag to the IPC call.
• MOSIX is the most general-purpose. Processes are scattered across a cluster automatically without having to modify the programs. No API needed other than usual Unix-level process use. Allows parallel execution of any program, although full use requires a parallel program design.

• TreadMarks
• The Quarks DSM System (ports to other platforms are here and my port to Linux is here.)
• DIPC (or try here)
• SHRIMP, a high performance parallel system for Linux.
• PVM -- a message passing approach to parallel programming.

The statement that the open-source community is not a good innovator is, like most complex issues, both true and untrue.

On the one hand, only the most blind of observers would suggest that novel products are not emerging from that quarter. The sheer volume of announcements on Freshmeat is just flabbergasting, and scattered like jewels in among the 95% of fairly ordinary stuff are some really excellent software products and many priceless ideas.

But on the other hand, and maybe this is where SGI is coming from, innovation in the Linux kernel is comparatively minimal. I don't think anyone would go so far as to say that it is stifled, but the fact remains that the choice of which ideas are accepted into the official release and which are not is in the hands of a very few people (maybe three or four, or possibly just the one). That must have an effect on innovation, however much we respect the people in question.

As a little example of the above, the DIPC project implemented a gem of an idea (I have absolutely nothing to do with it, by the way): allowing processes that communicate through System V IPC mechanisms on a single host to do so even if they are on different machines, while maintaining 100% compile-time application compatibility because the only difference at the API is a single bit in the IPC headers which you'd flick off or on for single or multiple machine operation. That's innovation, usefulness and elegance rolled into one. But no, Linus didn't want to put it into the standard kernel, and to say that the developers were greatly dispirited is the understatement of the year.

It's worth reflecting that if a kernel facility isn't part of the standard distribution, or worse, if it's available only as a patch, then for all intents and purposes it doesn't exist. We musn't get ourselves into a situation where innovation in the Linux kernel suffers as a result of this possibility.

Doug Ward - dsward@cei.net

For that matter, the Linux Distributed IPC (DIPC) package was written by an someone from Iran, so open source parallel processing contributions are flowing in many directions already. And DIPC already has distributed shared memory, which MOSIX needs...maybe that will flow together.