Re:Linus Torvalds on Garbage Collection (2002)
by ramchip on Saturday April 23, 2011 @04:41AM
(#2476514)
My point was not aimed at the Tera specifically, but rather at this architecture as a solution to the memory problem in general. As you say, supercomputers have been using multi-threaded code for a long time now, so it's not a technical problem for this case, but it can be for the kind of applications 'regular users' may need.
Thanks for the references. I'm in embedded systems but I hadn't heard of the Parallax Propeller before, it's an interesting architecture.
Re:Linus Torvalds on Garbage Collection (2002)
by kragen on Saturday April 23, 2011 @05:27AM
(#2476565)
I suspect that the killer feature of Tera-like architectures for embedded systems could be their determinism. Embedded microcontrollers like the LPC2100 series are starting to get caches, branch prediction, and the like, and that makes me really nervous, because it's going to make worst-case timing very hard to reproduce in testing. (In a way, this has been the case since the introduction of interrupts in the 1960s, but I think it makes the problem much worse.)
If multithreaded or massively multicore processors were a viable alternative, you could get deterministic timings without sacrificing throughput. You could even do away with interrupts. The GreenArrays chip doesn't have interrupts at all; instead, its cores go into a low-power shutdown state whenever they're waiting on I/O.
But that's all pretty speculative.
This started happening in earnest about 20 years ago in the supercomputer market, which is where the Tera was sold. About 10 years ago, it started happening in desktop CPUs (check out Herb Sutter's article about "the end of the free lunch") and now it's starting to happen in embedded microcontrollers, with the Parallax Propeller and the GreenArrays chips.
As it happens, the Tera didn't lose to faster single-threaded supercomputer CPUs. By the time the MTA came out, even the most stalwart defenders of the fast-single-threaded-performance approach, like the Cray SV1 and the NEC SX-5, had succumbed to the necessity of CPU parallelism. But the approach that was taking over the supercomputer market at the time was actually far more parallel, and far more difficult to program efficiently --- NUMA machines and then Beowulfs.
So that's why I don't think it was single-threaded programs that made the Tera fail in the supercomputer market.
The question of how to meaningfully parallelize XML and HTML parsing, compilation, text processing in general, and web browsers are very interesting indeed. It's not obvious how to do it, but it might turn out to be tractable. A group at Berkeley was doing some research on it in 2007 and 2008: http://www.eecs.berkeley.edu/~lmeyerov/projects/pbrowser/