Slashdot Mirror

It's certainly not the case that "almost all" problems decompose into data parallelism. Likewise, while there are some tasks that GPUs can do very well, there are others where they don't do well at all. What Linus is arguing is that the cases where they don't do well at all dominate. I concur. Programming this stuff typically locks you deeply into a single GPU's architecture, and, oh, you have zero cache, zero pipelining. There are other problems, too. A good overview of the technology as a whole is here: http://cstar.iiit.ac.in/~kkish...

GPGPU's have had some impressive successes, but CPUs are still more versatile, and, like Linus says, I don't see Intel giving up single core performance so people can program a bunch of tiny little ant-processors that can't communicate with each other in less than 500 cycles.

Actually I just found a paper about it: http://dli.iiit.ac.in/ijcai/IJCAI-81-VOL-2/PDF/072.pdf

The author of said paper might know Chris Fenton in some way, since the verilog code mentions a company named

// Company: D.E. Shaw Research
// Engineer: Christopher Fenton
//
// Create Date: 19:54:35 01/29/2008

Actually, I found a book at google

"Strategic Computing, By Alex Roland, Philip Shiman" that mentions DADO and NonVon being canceled because they didn't offer anything fundamentally new. Compared to the CM they had little staying power.

I had no idea it would be findable on the web - the Digital Library of India did a real nice job of scanning a paper on it into a nice, searchable OCRed PDF:

http://dli.iiit.ac.in/ijcai/IJCAI-81-VOL-2/PDF/071.pdf

ZMOB consisted of 256 home-grown Z80 boards and 256 home-grown communication boards. The comm hardware controlled a 48-bit wide bus that was essentially a shift register running in a loop around all 256 processor/comm pairs.

If we had started the project a few years later, we would have probably used 68000 processors, or maybe 8085/8087.

ZMOB was Maryland's first big hardware project, and we learned a lot about how not to do projects of this nature, like the worst ways to maneuver money through state channels, and that when you do large scale machines, careful signal engineering matters.

ZMOB per se was a failure - we never really got all 256 machines running at once. The comm hardware was eventually broken up into smaller 16 and 32 machine loops, and the Z80s were replaced by 68000s.

Another company seems to have developed speech recognition engines for embedded devices in languages other than english. Speech recognition has a potentially huge user base(in tens or hundreds of millions atleast) if they can crack the problem for native indian and chinese languages.

There is a lot of scope for the sister technology(speech synthesis) too ,if it can be implemented with reasonable success in native languages.Ideally ,this technology could act like a google translate for voice.It could break the language barrier at one stroke.unfortunately ,speech synthesis seems to be much more nascent.

Another link here

http://dli.iiit.ac.in/

Same at my institute as well. Open Source, Free Software, are not just buzzwords. People involved with these are treated as technical gurus in our campus, since they are the people who get things working with the most efficacy. Needless to mention, all our server use open source software, and all the sysadmins are hardcore linux fans.

I agree with u. Isn't our email client enough and even better for such works. Then why to spend HUGE amount on a silly mouse. By the u can if u have the money like bill gates ------ Visit my blog for more or Visit this