Sign in with
Sign up | Sign in

Intel Patents Dynamic Core Swapping in Multi-core Systems

By - Source: USPTO | B 20 comments

Intel has just been granted a patent covering core swapping that could have big implications for processors that use different, but instruction set-compatible cores within one chip.

According to the document published on April 10, the patent covers at least two cores with different performance levels in a single processor, in which the first core can move some function to the second core. The second core would be powered up when it is needed, but it is described to consume substantially less power than the first core when performing a particular function. That idea sounds similar to what hybrid notebooks would take advantage of and it is especially similar to Nvidia is doing with Tegra 3 and its 4+1 architecture, where a fifth core is used to run certain features at a very low power state.

Intel filed the patent in December of 2008 and explicitly describes a power saving feature that is enabled by low power cores that are able to let a circuit power down much more power hungry-cores.

In Intel's description, "in [a] mobile platform, a high performance core is desired for operation when the platform is connected to a fixed infrastructure power network, such as when the mobile platform is docked to a desktop personal computer (PC) for data synchronization. However, when it is used in a battery mode, a low performance with low power consumption is preferred. A fixed microprocessor core with fixed performance characteristics may not be able to accommodate different usage conditions."

The inventors explain that a traditional homogeneous processor core architecture may not be able to achieve the best possible power savings, as cores are typically designed for high clock speeds and include, for example, many more transistors than a low-power core really needs to perform basic functions.

The patent granted to Intel fixes this problem with the idea to build different types of (x86) cores into the same processor.

Display 20 Comments.
This thread is closed for comments
  • -1 Hide
    echdskech , April 12, 2012 10:57 AM
    doesn't that tegra 4+1 do that already?
  • 0 Hide
    mihaimm , April 12, 2012 11:57 AM
    Bloody patents... it makes me sad whenever a new one is granted... especially for old technology that is commercially available and made by another company. There is something wrong with the patent office... maybe they're all working on the next TOE but somehow I doubt that...
  • 7 Hide
    tical2399 , April 12, 2012 12:25 PM
    mihaimmBloody patents... it makes me sad whenever a new one is granted... especially for old technology that is commercially available and made by another company. There is something wrong with the patent office... maybe they're all working on the next TOE but somehow I doubt that...


    There is nothing wrong with a a patent for a tech that somebody already has a similar version of. As long as you come up with a new method of doing the same thing. By your logic there should only be one of any kind of any product.
  • 6 Hide
    willard , April 12, 2012 12:34 PM
    echdskechdoesn't that tegra 4+1 do that already?

    You didn't read the article very carefully, did you?

    Quote:
    Intel filed the patent in December of 2008

    Intel filed for this patent at around the same time as the very first Tegra chip was released.
  • -5 Hide
    dan-n , April 12, 2012 12:37 PM
    I'm sorry but how is this different to ARM's big.LITTLE?
  • -2 Hide
    dan-n , April 12, 2012 12:37 PM
    I'm sorry but how is this different to ARM's big.LITTLE?
  • 5 Hide
    trumpeter1994 , April 12, 2012 12:48 PM
    dan-nI'm sorry but how is this different to ARM's big.LITTLE?

    Because it's on a cpu rather than a gpu and I know ARM makes cpu's but when was the last time you saw an ARM laptop?
  • -1 Hide
    trumpeter1994 , April 12, 2012 12:50 PM
    Wow this sounds really good though seeing as how intel has such a dominant presence in the notebook market, if combined with nvidia's optimus technology then I can see some massive battery life improvements
  • 2 Hide
    eddieroolz , April 12, 2012 2:19 PM
    Man, looks like NVidia's gotta think of a way to defend its product if Intel ever bites.
  • -1 Hide
    dan-n , April 12, 2012 5:26 PM
    trumpeter1994Wow this sounds really good though seeing as how intel has such a dominant presence in the notebook market, if combined with nvidia's optimus technology then I can see some massive battery life improvements


    Are you saying that ARM's concept is on GPUs?

    If so you're wrong... It's on CPUs, the Cortex-A7 and A15 theyre CPUs not GPUs. X86n is not GPUs thats CPUs too!

    And I'll think you'll find come Windows 8 it will be in a lot of laptops and when the v8 cores come out it will be in even more due to the 64-bit architecture. Furthermore, ARM is in almost 100% of tablets!

    ARMs CPUs are a hell of a lot more power efficient than Intel and have been for years!
  • 0 Hide
    dan-n , April 12, 2012 5:26 PM
    whoops wrong quote ^^^ I meant this one:

    trumpeter1994Because it's on a cpu rather than a gpu and I know ARM makes cpu's but when was the last time you saw an ARM laptop?

  • 2 Hide
    blazorthon , April 12, 2012 6:16 PM
    taetzschmiGood article! Thanks for sharing.


    You are a tricky spammer.
  • 0 Hide
    trumpeter1994 , April 12, 2012 10:55 PM
    dan-nwhoops wrong quote ^^^ I meant this one:

    No I was referring to the implumentation of Arm into tegra along with a gpu...... and I'm primarily talking about notebooks not tablets..... Honestly I don't care very much for tablets I'd rather have a notebook if i wanted a mobile computer.
  • 1 Hide
    dan-n , April 12, 2012 11:22 PM
    trumpeter1994No I was referring to the implumentation of Arm into tegra along with a gpu...... and I'm primarily talking about notebooks not tablets..... Honestly I don't care very much for tablets I'd rather have a notebook if i wanted a mobile computer.


    Dude its called a SoC and you obviously don't know what big.LITTLE is, because Tegra is not big.LITTLE thats NVidias 4+1 architecture completely different.

    If you want a low power "notebook" get a transformer prime with a battery life of 18hours, whilst under normal use with web browsing and HD video play back... None of this 10hours of battery life whilst we're not using it to do anything!

    The next arms race is not in performance, its in low power seeing as everything is shifting towards a cloud computing infrastructure! there was big hype when Intel anounced that it was using clock gating... ARM has been using this since it was founded in 1991!
  • 0 Hide
    f-14 , April 13, 2012 1:30 AM
    dan-nThe next arms race is not in performance, its in low power seeing as everything is shifting towards a cloud computing infrastructure!

    ask megaupload users how that cloud computing is working out for them. that's is just the tip of the iceberg on cloud computing problems. ask your ISP about it's bandwidth caps and download limits.
    figure in lag power outages, hackers. amazing you are not using microsofts terminal services to write this on your node.
  • 1 Hide
    blazorthon , April 13, 2012 3:27 AM
    dan-nDude its called a SoC and you obviously don't know what big.LITTLE is, because Tegra is not big.LITTLE thats NVidias 4+1 architecture completely different.If you want a low power "notebook" get a transformer prime with a battery life of 18hours, whilst under normal use with web browsing and HD video play back... None of this 10hours of battery life whilst we're not using it to do anything!The next arms race is not in performance, its in low power seeing as everything is shifting towards a cloud computing infrastructure! there was big hype when Intel anounced that it was using clock gating... ARM has been using this since it was founded in 1991!


    You had me agreeing with you until you mentioned cloud computing. Cloud computing is about as unreliable as any form of computing can get, not to mention the reduced speed and the costs associated with it.
  • 0 Hide
    hanskey , April 13, 2012 4:21 PM
    blazorthonYou had me agreeing with you until you mentioned cloud computing. Cloud computing is about as unreliable as any form of computing can get, not to mention the reduced speed and the costs associated with it.

    ditto.

    Obviously Intel is trying to get a foothold in the tablet market, with this, but I don't think they can compete with ARM any time soon.

    However, I do hope that Intel does well enough to force tablet performance to increase at a much faster pace with this move, because tablets are still nowhere near good enough to be cheap laptop replacements.
  • 0 Hide
    blazorthon , April 14, 2012 5:46 AM
    hanskeyditto.Obviously Intel is trying to get a foothold in the tablet market, with this, but I don't think they can compete with ARM any time soon.However, I do hope that Intel does well enough to force tablet performance to increase at a much faster pace with this move, because tablets are still nowhere near good enough to be cheap laptop replacements.


    Oh, Intel can compete with ARM.
  • 0 Hide
    kronos_cornelius , April 14, 2012 10:12 PM
    Intel has a piece of paper called a patent. Nvidia has a chip being manufactured at thousands per day. I would say Nvidia has more than enough evidence to show prior art. The Tegra 3 did not just come into existence in the last 4 years. These things take time. This just shows Nvidia puts more resources into engineering where Intel puts more resources on lawyers.
  • 0 Hide
    blazorthon , April 14, 2012 11:57 PM
    kronos_corneliusIntel has a piece of paper called a patent. Nvidia has a chip being manufactured at thousands per day. I would say Nvidia has more than enough evidence to show prior art. The Tegra 3 did not just come into existence in the last 4 years. These things take time. This just shows Nvidia puts more resources into engineering where Intel puts more resources on lawyers.


    Intel can have whatever patents that they want. Having a patent (even a stupid one) doesn't matter. Now, if Intel attacks Nvidia through this patent, then I will see problems. However, Intel might have patented this before Nvidia tried it out. Nvidia should have patented it or at least done something about it themselves, but they didn't.

    This does not show that Nvidia puts more money into work and Intel puts more money into lawyers at all. In fact, if Intel does not abuse this patent (they might, they might not), then it is a good thing that they patented it because it will stop someone who would abuse it from patenting it and suing everyone. Lately, Intel does not seem to be the kind to sue another company right now, so they probably won't do it. However, this patent's similarity to Tegra 3 is alarming, so it is possible.

    Besides all of that, from the looks of it, this seems to only apply to X86 CPUs, so it's no big deal if Nvidia keeps Tegra as an ARM based CPU.