It looks like you're using an Ad Blocker.

Please white-list or disable AboveTopSecret.com in your ad-blocking tool.

Thank you.

 

Some features of ATS will be disabled while you continue to use an ad-blocker.

 

A Chip With The Power Of A Supercomputer

page: 2
9
<< 1   >>

log in

join
share:

posted on Feb, 14 2007 @ 03:32 AM
link   
The problem is not just finding a use for the 80 cores, this will happen eventually. The problem is getting all the cores to acctually work in sync, the hardware and software required for that is still a large issue.

What i find interesting is the use this could have for robotics. Being able to deal with more tasks may lead to further great leaps in control of balance and hand eye co-ordination. Think asimo but with less chips to help control it, this should mean it runs longer on less batteries.

The day a small robot can accurately calculate the path of a ball whilst it is flying and then catch it will be a major milestone. Some advances ahve already been made but the robots tend to be rather large and task dedicated.

Well done to intel though.



posted on Feb, 14 2007 @ 05:05 AM
link   

Originally posted by RedGolem
Could that much power be used for say optical cloaking?


Don't see a need for computers to optically camouflage.

[yvid]http://www.youtube.com/watch?v=JKPVQal851U[yvid]

Neat, huh?




www.youtube.com...

[edit on 14-2-2007 by LastOutfiniteVoiceEternal]



posted on Feb, 14 2007 @ 06:16 AM
link   

Originally posted by ImaginaryReality1984
The problem is not just finding a use for the 80 cores, this will happen eventually. The problem is getting all the cores to acctually work in sync, the hardware and software required for that is still a large issue.

What i find interesting is the use this could have for robotics. Being able to deal with more tasks may lead to further great leaps in control of balance and hand eye co-ordination. Think asimo but with less chips to help control it, this should mean it runs longer on less batteries.

The day a small robot can accurately calculate the path of a ball whilst it is flying and then catch it will be a major milestone. Some advances ahve already been made but the robots tend to be rather large and task dedicated.

Well done to intel though.


excellent point does any one know how many processors the human brain can handle at one time. thats a good point.



posted on Feb, 14 2007 @ 07:31 AM
link   

Originally posted by LastOutfiniteVoiceEternal

Originally posted by RedGolem
Could that much power be used for say optical cloaking?


Don't see a need for computers to optically camouflage.

[yvid]http://www.youtube.com/watch?v=JKPVQal851U[yvid]

Neat, huh?


width="425" height="350"> "http://www.youtube.com/v/JKPVQal851U" type="application/x-shockwave-flash" wmode="transparent" width="425" height="350">

www.youtube.com...

[edit on 14-2-2007 by LastOutfiniteVoiceEternal]


That looks impressive doesn't it? It's a coating painted on those objects similar to the stuff on a projection screen, they took a still shot of the background without the guy there, and there's a DLP projector in front under the camera. It's projecting the background scene onto anything that's sufficiently reflective.

There's another demo just like it where he wears a "cloak" on a downtown street, for that one they have a small camera behind him picking up the background scene for the projector.

The guys that did these "demos" published the details not long after.



posted on Feb, 17 2007 @ 11:03 PM
link   
11 years..too long. That's assuming silicon is still being used.

This is another, quicker route to tera-flopping on a desktop PC..64-bit double precision number-crunching using your graphics card's GPU:

www.tomshardware.co.uk...

[edit on 17-2-2007 by Ross Cross]



posted on Feb, 18 2007 @ 07:27 PM
link   
Red Cross,
that was a good link, thanks.
It will be really cool if people can turn the PCs inot super computers. It should bring a lot of interesting things in devolpment.



posted on Mar, 4 2007 @ 08:58 PM
link   
You're welcome, RedGolem.

Here's some news on AMD's latest accelerated computing platform.

AMD demonstrated a single system housing a dual-core AMD Operton processor (the server version of the Athlon64), together with two AMD R600 stream processors/co-processors.

This combination delivers over one trillion floating-point calculations per second..10 times faster than today's high-performance servers (although which servers were compared was not established).

AMD is also optimising the protocols on which such coprocessors communicate with the main processor, which in this case is running XP Pro. Such technology is aimed at a wide range of applications, including consumer.

This reminds me of Commodore's 1985 home computer, the Amiga. Essentially a trasnsputer, the Amiga housed a 16/32-bit 68000 CPU together with custom coprocessors heavily geared towards multimedia applications. As a result the Amiga could do tricks with video and audio that were out of this world..way ahead of it's time. The pre-emptive (ie real) multitasking on that machine was also incredibly smooth since it was built into the hardware from the ground up, unlike the inferior co-operative multi-tasking found on Windows-based machines at the time.

[edit on 4-3-2007 by Ross Cross]



posted on Mar, 5 2007 @ 03:56 AM
link   
Interesting to find some one else who rembers Comador

Around that time or so I was useing an apple, it is impressive how far they have come.



posted on Mar, 5 2007 @ 01:32 PM
link   

Originally posted by Tom Bedlam

There's no general purpose logic like you have in an x86 core; they're not really good for general computational work.

Vector processors don't run Doom, or Sim City or balance your checkbooks. They're used for crunching numbers. Lots and lots of numbers. A thing you could do with it, if it was a production chip, would be to run crypto and voice analysis on it:


I have seen on Crays site that these multicore supercomputers run in x86. The obvious bottleneck I can see is on the x86 software now, because the code is not written to take advantage of these multiple cores. Other wise I think you can play Doom...



posted on Mar, 5 2007 @ 01:54 PM
link   

Originally posted by tha stillz

I have seen on Crays site that these multicore supercomputers run in x86. The obvious bottleneck I can see is on the x86 software now, because the code is not written to take advantage of these multiple cores. Other wise I think you can play Doom...


Nuh-uh. Cray, I think, has MADE some parallel processing x86 machines, but the one in the article is not.

Each of the 80 cores on Polaris is a pair of FPU's, a mesh router, and about 10k of RAM. The mesh router and 3D RAM are things they're testing out. The rest of it is just there to do something, IMO.

Edit: SIMPLE FPU's at that, they're not full bore FPU's like you'd see on an x86 machine, I think it's just a pair of FMACs.

[edit on 5-3-2007 by Tom Bedlam]



new topics

top topics



 
9
<< 1   >>

log in

join