AfterDawn: News

Intel shows 22nm SRAM chip

Intel has surprised some industry experts by showing off working 22-nanometer SRAM chips, proving that Moore's Law still has some fight left in it. Moore's Law, applied to chips, predicts a doubling of the number of transistors in chips roughly every 2 years.

Some industry experts admitted that they didn't believe that it was possible - considering that the new Intel chip contains three billion transistors in an area the size of a fingernail. Even with that, Intel said that there could be chips coming that push the physical limits even further.



"Moore's Law is not a law of physics, it's a law of human inventiveness," Intel president and chief executive Paul Otellini told BBC News. "There is nothing that says that humans can't continue to invent." The company is currently producing and selling chips based on 45nm technology, and will move to 32nm later in the year.

"We have at least four generations of what we are shipping today in development. That is as far out as we have ever been able to develop things," Otellini said. "Not everything is done but it gets harder for sure. It requires more invention and more money and it's our job to continue that pace."

Written by: James Delahunty @ 23 Sep 2009 18:02

• Previous article

• FCC's 'Open Internet' rules come under fire

• Next article

• Apple, Eminem publisher likely heading to trial

• 13 comments

• ZippyDSM

  So what dose this mean in terms of power/performance speed and power/heat?

  23.9.2009 18:27 #1

• Lothros

  I'm gonna guess a 3% real increase in performance with a 300% increase in price

  23.9.2009 18:28 #2

• ZippyDSM

  Originally posted by Lothros: I'm gonna guess a 3% real increase in performance with a 300% increase in price
  Ain't it normally 30% performance wise and 500% price wise?
  :P
  
  So we ain't anywhere near 4000MHZ ramm/FSB yet?

  23.9.2009 18:31 #3

• ThePastor

  So, computers of the future will contain "near absolute Zero" cooling devices to achieve these superconducting speeds?

  23.9.2009 18:42 #4

• ZippyDSM

  Originally posted by ThePastor: So, computers of the future will contain "near absolute Zero" cooling devices to achieve these superconducting speeds?I dunno we have hit 1333 so faster is not that hard to do....it just costs alot right now :P

  23.9.2009 18:55 #5

• mike.m

  They got these made sooner than I expected. Who knows, we might see these in laptops/desktops next year. Nice work Intel.

  23.9.2009 19:57 #6

• rmarier83

  With the numbers keep shrinking, what happens when they get very low, like 4nm or 2nm? Maybe they'll stretch it out with a decimal system "0.92nm chips"! Whatever happens, I just hope we soon get our dual-core atoms for netbooks and tablets.

  23.9.2009 22:28 #7

• ZippyDSM

  Originally posted by rmarier83: With the numbers keep shrinking, what happens when they get very low, like 4nm or 2nm? Maybe they'll stretch it out with a decimal system "0.92nm chips"! Whatever happens, I just hope we soon get our dual-core atoms for netbooks and tablets.I'd rather have a 6Ghz atom than a 2 core 1.2 .......but they can't seem to fix that pesky heating problem yet >>

  23.9.2009 22:30 #8

• xyqo

  Originally posted by rmarier83: With the numbers keep shrinking, what happens when they get very low, like 4nm or 2nm? Maybe they'll stretch it out with a decimal system "0.92nm chips"! Whatever happens, I just hope we soon get our dual-core atoms for netbooks and tablets.The limit is really in the 11nm scale which is basically nanoelectronics. Quantum tunnelling will affect anything thats attempted at such a scale.

  24.9.2009 10:06 #9

• shaffaaf

  well we have near 5GHz GDDR5 in GPUs, and near 3GHz GDDR3.
  
  adn what they do with turbo in the LGA 11156 i5/i7s is good. when a program only uses 1/2 core, the others are shut down and it increaces the multi for the 2 used cores which leaves them with a higher clock speed.

  24.9.2009 14:20 #10

• handsom

  Honestly, I don't know what expert could still be surprised by this. People have been decrying Moore's law for years. And it continues to prove itself. Just because YOU can't fathom how they'd do it, doesn't mean others can't. It's a shortsighted assumption.
  
  That said, it's nice to see progress... Even if Intel is practically making out with the press "It's our job to continue that pace." Could you kiss any more @ss?
  
  Obviously, we'll see their competitor AMD make similar jumps in a relatively close timeframe, because that's just how competition has always worked.
  
  I do think that it's important for microprocessor manufacturers to continue making this kind of progress; because it allows us to move forward. We seem to have hit a kind of wall with raw core speed, just around the 3-4Ghz mark. Sure, we could very slowly get better, but going in the direction of multi-threaded program design, and multi-core operations is a much more sensible technology path for us. Just because the individual cores aren't necessarily getting faster (In some cases they're slower), doesn't mean this isn't a step forward.
  
  More cores means the ability to do more at once, just because a single core doesn't handle a single operation as fast as it might have before doesn't mean that being able to get more cores, more transistors into the same space won't allow for significant performance boosts.
  
  The interesting thing right now is that a lot of users and consumers are crying out "That's great, but it doesn't do me any good, I don't need this technology." A lot of folks feel like progressively faster computing is moving too slow, and it's simply not true. In fact, it's nearly the opposite. Manufacturers are creating processors that are capable of more than software engineers know what to do with just yet. It's an unusual case, but I think it's important to remember that the hardware is still moving forward at a consistent rate, we just don't have software that fully utilizes the latest progressions. We will in the coming years, and these multi-core CPU's will suddenly become much more useful for everyone.

  24.9.2009 18:09 #11

• ZippyDSM

  Originally posted by handsom: Honestly, I don't know what expert could still be surprised by this. People have been decrying Moore's law for years. And it continues to prove itself. Just because YOU can't fathom how they'd do it, doesn't mean others can't. It's a shortsighted assumption.
  
  That said, it's nice to see progress... Even if Intel is practically making out with the press "It's our job to continue that pace." Could you kiss any more @ss?
  
  Obviously, we'll see their competitor AMD make similar jumps in a relatively close timeframe, because that's just how competition has always worked.
  
  I do think that it's important for microprocessor manufacturers to continue making this kind of progress; because it allows us to move forward. We seem to have hit a kind of wall with raw core speed, just around the 3-4Ghz mark. Sure, we could very slowly get better, but going in the direction of multi-threaded program design, and multi-core operations is a much more sensible technology path for us. Just because the individual cores aren't necessarily getting faster (In some cases they're slower), doesn't mean this isn't a step forward.
  
  More cores means the ability to do more at once, just because a single core doesn't handle a single operation as fast as it might have before doesn't mean that being able to get more cores, more transistors into the same space won't allow for significant performance boosts.
  
  The interesting thing right now is that a lot of users and consumers are crying out "That's great, but it doesn't do me any good, I don't need this technology." A lot of folks feel like progressively faster computing is moving too slow, and it's simply not true. In fact, it's nearly the opposite. Manufacturers are creating processors that are capable of more than software engineers know what to do with just yet. It's an unusual case, but I think it's important to remember that the hardware is still moving forward at a consistent rate, we just don't have software that fully utilizes the latest progressions. We will in the coming years, and these multi-core CPU's will suddenly become much more useful for everyone.Well real world results basically come down to slower speeds sower lower performance, at least for the next 5 years, beyond that everything should be ready to handle real multi core processing and CPU speed wont matter..

  24.9.2009 18:15 #12

• handsom

  ZippDSM, I couldn't agree with you more as far as real world results are concerned. For me, it's just a matter of remembering that the hardware folks are there, they're ready, and they're getting the equipment out in a [mostly] cost and time effective manner. A lot of consumers don't fully realize that what's holding this hardware back isn't the hardware itself, it's how we [fail to] utilize it.

  24.9.2009 18:40 #13