World's First Polymorphic Computer
tdelama writes to mention Raytheon Company has developed the first polymorphic computer named the Morphable Networked Micro-Architecture (MONARCH) for the US Department of Defense. "'Typically, a chip is optimally designed either for front-end signal processing or back-end control and data processing,' explained Nick Uros, vice president for the Advanced Concepts and Technology group of Raytheon Space and Airborne Systems. 'The MONARCH micro-architecture is unique in its ability to reconfigure itself to optimize processing on the fly. MONARCH provides exceptional compute capacity and highly flexible data bandwidth capability with beyond state-of-the-art power efficiency, and it's fully programmable.'"
Is it fully buzzword compliant?
- None can love freedom heartily, but good men; the rest love not freedom, but license. -- John Milton
There is very, very little new under the sun.
Back in the early 1970s there was a mini computer called the "Meta 4" whose microprogramming could be changed on the fly. The purpose was to let you run software written for other vendors' instruction sets.
While the chip being discussed may do other spiffy stuff to optimize its performance in different roles, you really can't call it the first "polymorphic" computer.
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
"The MONARCH zaps itself with a wand of polymorph. The arch-lich hits! Oh no, it's using the touch of death! You die..."
How can we continue to believe in a just universe and freedom to eat crackers if we have no ale?
Get a 100k of these running in parallel, give em a self organizing and threading algorithm and run for cover. On the plus side Schwarzenegger's armageddon would be much more interesting than Gore's alternative.
An FPGA could provide the same functionality, but at a tremendous loss of efficiency. FPGAs have a very high overhead to support reconfiguration. But the reconfiguration is considerably more than this processor (likely) offers. (I couldn't read the article - dead link)
But with a claim of incredible power efficiency, it's decidedly not an FPGA. I imagine they borrow some of the concepts, but not entirely.
As a hybrid, FPOA (field-programmable object arrays) provide small programmable "objects" which are less granular than typical FPGA offerings. In the right application, an FPOA can achieve higher speeds and better power efficiency. In the "wrong" application, they're horrible.
It seems that this device would switch between the high computational efficiency of DSPs and things like graphics processors and the better branching / decision-making performance of general-purpose CPUs.
Their state-of-the-art-meter goes to eleven.
No, but they should. Not that I dislike Raytheon inherently, but they are certainly spinning this press release pretty hard. It's just programmable hardware. It's an attempt to catch the attention of the government because there are two Military-Industrial coalitions bidding right now for the military's next generation satellite system (which will be a contract worth tens of billions of dollars for about the next decade).
Since the press release is so light on detail, obviously the actual hardware isn't that impressive. Note things like these quotes:
Oh, really? And how many libraries of congress per fortnight is that?
Target audience, right there.
This is at least a little bit of information. However, those numbers are similar to current generation CPUs. I think the PS3 Cell can outperform this chip, so unless we have some power numbers it's unimpressive.
It's not a big surprise. It's just a press release and a slashvertisement.