June 19, 2006 9:00 PM PDT

Chip breaks speed record in deep freeze

IBM and Georgia Tech have coaxed a chip to run at 500GHz, a record for a silicon-based device, by dropping the temperature to minus 451 degrees Fahrenheit.

The experiment is part of a project to explore the ultimate speed limits of silicon-germanium (SiGe) chips. SiGe chips are similar to standard silicon chips, but they also contain germanium for better performance and lower power consumption.

Adding germanium, however, increases the price of producing wafers and chips that come out of the wafers, so SiGe chips are typically only found in a few select markets. IBM has sold hundreds of millions of SiGe chips since it began selling them in 1998, but the cell phone industry gobbles up billions of plain silicon chips annually. (Germanium is sprinkled into standard silicon chips: Intel adds minute amounts of the element to create strained silicon in its processors).

At room temperature, the IBM-Georgia Tech chip operates at 350GHz, or 350 billion cycles per second. That's far faster than standard PC processors today, which range from 3.8GHz to 1.8GHz. But SiGe chips can gain additional performance in colder temperatures.

To that end, IBM and Georgia Tech scientists turned down the temperature and cryogenically froze the chip at minus 451 F. It's about as cold as things get. An extremely cold temperature like that is found naturally only in outer space, but can be artificially achieved on Earth using ultracold materials such as liquid helium. Absolute zero comes at minus 459 F.

SiGe chips, the scientists theorized, could eventually hit 1 terahertz, or 1 trillion cycles a second.

Ultimately, high-performance SiGe chips could be used in defense systems, space exploration vehicles and remote sensing. Conceivably, you could also build a computer for Ted Williams and other cryogenically frozen celebrities.

See more CNET content tagged:
silicon chip, Georgia Tech, temperature, IBM Corp., scientist

26 comments

Join the conversation!
Add your comment
source
You can find the original press release at the GT Research News site.

<a class="jive-link-external" href="http://gtresearchnews.gatech.edu/newsrelease/half-terahertz.htm" target="_newWindow">http://gtresearchnews.gatech.edu/newsrelease/half-terahertz.htm</a>

They also show some interesting photos of the new SiGe chips.
Posted by pstamatiou (1 comment )
Reply Link Flag
Wait a minute 350 GHZ @ Room Temp!!!!?
I can understand the boost from -451F Cooloing but these chips can turn in 350GHZ @ room temp???!!! NO Cooling device?!!! I'd like to see the cooling device for the room temp aplication at that kind of speeds!
Posted by mssoot (169 comments )
Link Flag
Wow
That is amazing. 500ghz! Makes me want to cry. Does anyone else out there just hate to love and love to hate technology? I thought 4ghz was amazing when they overclocked a p4 on techtv (back when it was worth a crap) Anyways, i just have to say wow. Won't be long now until we're going "heh, you have a 500ghz processor, wow that is so 2010, step it up with the terahertz, here."
Posted by sintexxx (1 comment )
Reply Link Flag
e be like say
the dayz of the Mhz are now numbered
Posted by dapodollar (2 comments )
Reply Link Flag
e be like say
the dayz of the Mhz are now numbered.
Posted by dapodollar (2 comments )
Reply Link Flag
Whoa
Why is it taking so long for these chips to come to the personal desktop?
Posted by Michael00360 (58 comments )
Link Flag
Okay, but will it support 128bit encryption?
Now that Bill gates is flipping burgers at his weekend Bar-B-Q's.
Posted by Stalin Hornsby (60 comments )
Reply Link Flag
Support 128? try CRACK 1024!
Do you seriously think that the CIA, NSA or other intelligence agencys will not have a vat of liquid helium in a basement running an encryption cracking farm of these CPUs in a few years? 128 bit wireless can be cracked now in a day or two with 2 laptops. How long would it take 10-20 dedicated parallel 500Ghz CPUs to crack 128/256/1024 bits?
Posted by wyrm_ksc (4 comments )
Link Flag
3Ghz for me, 350Ghz for them!?!??
what what what?? if they can make a 350 Ghz chip then why on earth can we only mass produce chips running around 4ghz tops?!?!? i'm sure there are some people out there who would pay 10 or 20K for a 100Ghz proc. Heck you could charge a pretty penny for a 10Ghz proc...
Posted by jeffhesser (102 comments )
Reply Link Flag
Lack of information
Nothing was said about the transistor size. If you can only get a few hundred or even thousands on a chip, it's useless as a general purpose CPU.

Another issue is heat dissipation. How much power is used to reach these high speeds? I would imagine quite alot.

It's one thing to achieve these results in a lab and quite another to create a viable product.

It does sound promising though...
Posted by freemarket--2008 (5058 comments )
Link Flag
Pssh thats nothing
I dont care how fast they make chips... all i care about is that they taste like Nacho Cheese.... oh wait wrong chips...


Or could they make that taste like nacho cheese?
Posted by Mackath (1 comment )
Link Flag
A boon for Blue Brain Project
This is great for the Health industry as far as simulating the human brain or real life. I am surprised they missed that. Things like the Blue Brain project
Posted by Blito (436 comments )
Reply Link Flag
When ,...
taht little baby hits the market, processor speed officially becomes
irrelevant.

No software maker on the planet is going to write software that
that chip cannot handle anytime soon,...
Posted by corelogik (680 comments )
Reply Link Flag
Consumers?
Why can't these chips be out to consumers in the near future, say 2years from now? I'd be willing to pay $2500 for a processor that fast!
Posted by samus1225 (1 comment )
Reply Link Flag
$2500 + more...
The cooling, energy use, and hardware that hooks into something like that would cost ridiculous amounts money to keep it going and such. Not to mention that when they say "Room temperature" I believe they means running it cooled to room temperature (I could be wrong), which is still significantly cooler than the average processor runs at even with cooling. (Mine runs on average 32 degrees celsius, which is something like the 90's in farenheit, 42 (over 104 farneheit) under stress)
Posted by Bobbias (55 comments )
Link Flag
But what's on the chip?
Are these full fledged processors? Chances are they aren't. These are more likely simple custom logic chips with a fraction of the number of transistors you would find on a CPU... just enough to perform a simple logic function to prove it can be done at that speed. The wavelength of the clock speed alone would play considerable havoc across the current footprint of a CPU. You'd wouldn't be able to keep the entire chip in phase with the clock.
Posted by Seaspray0 (9714 comments )
Reply Link Flag
But what's on the chip?
Are these full fledged processors? Chances are they aren't. These are more likely simple custom logic chips with a fraction of the number of transistors you would find on a CPU... just enough to perform a simple logic function to prove it can be done at that speed. The wavelength of the clock speed alone would play considerable havoc across the current footprint of a CPU. You'd wouldn't be able to keep the entire chip in phase with the clock.
Posted by Seaspray0 (9714 comments )
Reply Link Flag
Radiation?
I heard that the main barrier to having computer chips above 4 GHz is that the radiation they give off is in the region of microwaves.

So what kind of radiation do these chips give off??
Posted by ayteebee (32 comments )
Reply Link Flag
Radiation - answer
Ok, I did some research on Wikipedia. It seems that 500 GHz is at the upper end of the Far Infra Red spectrum. Microwaves go up to about 300 GHz. So instead of having a chip that microwaves your mind, you will be computing with an infrared lamp! Sounds like an interesting cooling problem.

Could be an interesting problem for fighter pilots; flying over an urban area with millions of infra red sources all over the city! Just hope they don't launch any heat-seeking missiles...
Posted by ayteebee (32 comments )
Link Flag
This article is WRONG!!!
The speed record does not measure how fast we can make CMOS computer processors. The test measures the max frequency at which a single transistor can produce current gain. There was an article published in New Scientist in 2002 that was similar to this one: <a class="jive-link-external" href="http://www.newscientist.com/article/mg17323320.600.html" target="_newWindow">http://www.newscientist.com/article/mg17323320.600.html</a>

Question: If IBM created a computer chip that ran at 110GHz in 2002, don't you think it would be on the market today?
Posted by HeyYoTyson (3 comments )
Reply Link Flag
you ever think...
that IBM was experimenting with the SiGe to see if it would acctuialy be feasible... and now they are trying to push the limits of the SiGe mixture to see how far it can go. This is far away from being retail just yet.
Posted by CaptDave86 (30 comments )
Link Flag
Question
Did YOUR "Blue Brain" shut off mid-sentence there?
Posted by handdrawn (8 comments )
Reply Link Flag
?????????
-451*F=-268.33333*C! How they done that I don't know! But It's the STRANGEST ARTICLE ever. But is It really -268.33 Celsius??
Posted by Michael-TNT (1 comment )
Reply Link Flag
kool ...
still need to dump the clock if they really have a need for speed :)
Posted by Lolo Gecko (131 comments )
Reply Link Flag
 

Join the conversation

Add your comment

The posting of advertisements, profanity, or personal attacks is prohibited. Click here to review our Terms of Use.

What's Hot

Discussions

Shared

RSS Feeds

Add headlines from CNET News to your homepage or feedreader.