Slashdot Mirror


48-Core Chips Could Redefine Mobile Devices

CWmike writes "Intel researchers are working on a 48-core processor for smartphones and tablets, but it could be five to 10 years before it hits the market. Having a 48-core chip in a small mobile device would open up a whole new world of possibilities. 'If we're going to have this technology in five to 10 years, we could finally do things that take way too much processing power today,' said analyst Patrick Moorhead. 'This could really open up our concept of what is a computer... The phone would be smart enough to not just be a computer but it could be my computer.' Enric Herrero, a research scientist at Intel Labs in Barcelona, explained that with the prototype chip someone could, for instance, be encrypting an email while also working on other power-intensive apps at the same time — without hiccups. Same for HD video. Intel's Tanausu Ramirez said it could also boost battery life. 'The chip also can take the energy and split it up and distribute it between different applications,' he said. Justin Rattner, Intel's CTO, told Computerworld that a 48-core chip for small mobile devices could hit the market 'much sooner' than the researchers' 10-year prediction."

285 comments

  1. Desktop by Anonymous Coward · · Score: 5, Insightful

    Let's put a 48-core processor on a desktop or laptop before we talk about tablets or phones...

    1. Re:Desktop by tom17 · · Score: 4, Interesting

      Why not concentrate on tablets and phones first?

      1. They are the fastest growing segment and "everyone will have one"
      2. This will then be your primary computing device that follows you around. It's with you when you need it, because...
      3. It's easier to use a fully mobile device as a workstation device (just add a keyboard/monitor and fashionable pointing device) than it is to use a workstation device as a pocket computer.
      4. Power savings from this kind of architecture are more critical on pocket devices
      5. ...
      6. PROFIT! (Sorry)

      I think it's a good way forwards. But that's just opinion so...

    2. Re:Desktop by tomhath · · Score: 2

      I'm pretty sure that's kind of the plan here. A single device that's powerful enough to replace what we have today. You can put it on your desktop if you want, or slip it into your pocket.

    3. Re:Desktop by Lord+Lode · · Score: 2, Insightful

      Fully agree:

      5 years ago, the top of the line Intel consumer desktop CPU's had 4 cores.
      today, the top of the line Intel consumer desktop CPU's have 4 cores.

      So, get to it please! :)

    4. Re:Desktop by Z00L00K · · Score: 4, Informative

      Any desktop with a decent GPU has more than that already.

      But the difference between a desktop and a phone makes it harder to get good performance on the desktop with many cores - it's memory bandwidth that's the bottleneck. On a phone you can dedicate cores to certain well-defined tasks and optimize them for that.

      --
      If builders built buildings the way programmers wrote programs, then the first woodpecker would destroy civilization.
    5. Re:Desktop by CajunArson · · Score: 2

      No... 5 years ago the top of the line *mobile* CPUs had 2 cores.
      Today the top of the line mobile CPUs have 4 much more powerful cores.
      It turns out that the mobile CPUs are now used on most desktops too, but there are still more powerful desktop CPUs.

      If, however, you want the top of the line *desktop* CPU, the 3930K has 6 cores and wipes the floor with any quad core CPU in existence... assuming you are running software that can actually take advantage of all those cores. Since a whole lot of even "multithreaded" software cannot do so, you are basically not having to pay for unneeded silicon and power by getting a quad-core CPU.

      --
      AntiFA: An abbreviation for Anti First Amendment.
    6. Re:Desktop by ByOhTek · · Score: 5, Insightful

      Because desktops have one less criteria to meet than tablets and phones - they don't have nearly as small of a power envelope.

      The desktop, therefore could be seen as a logical step in the progression to getting it on the phone/tablet.

      --
      Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
    7. Re:Desktop by ByOhTek · · Score: 1

      Also, IRRC, weren't the quad core desktop CPUs from 4 years ago hyperthreaded, meaning 2 physical and 2 logical? That would make example, by comparison, a 12 core, as I believe Intel has sadly gone back to hyperthreading their CPUs.

      --
      Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
    8. Re:Desktop by Anonymous Coward · · Score: 1

      Let's put a 48-core processor on a desktop or laptop before we talk about tablets or phones...

      Because once you have a mobile chip, putting it in a desktop is easy.
      The reverse is generally impossible.

      And mobile is where the growth (and money) is now. Focus on a stagnant market like the desktop, and you'll very quickly get run over by competitors whose heads aren't in the sand.

      Focusing on the low end, high-volume market is how the pathetic little 16-bit x86 grew up to beat Motorola in the '80s and the RISC vendors in the '90s (and, for that matter, Itanium.) It's easier for the small to scale up than the big to scale down.

    9. Re:Desktop by Anonymous Coward · · Score: 3, Insightful

      And I don't want my 48-core machine to be fucking mobile. It means I would then be expected to take my work everywhere with me. Fuck that. Until we learn how to respect the sanctity of vacation time in the US and bring up the average vacation length for workers to something near European standard I would rather see this in a non-mobile version first.

    10. Re:Desktop by Anonymous Coward · · Score: 0

      http://en.wikipedia.org/wiki/Kentsfield_%28microprocessor%29

      6 years ago was the first true 4 core Intel CPU for desktops.

    11. Re:Desktop by Anonymous Coward · · Score: 2, Interesting

      I guess it works for me as I can easily separate my working life from my personal life (Yes, I live in North America but I derived my only-work-at-work mentality from the other side of the pond).

    12. Re:Desktop by Anonymous Coward · · Score: 0

      I'm AC parent. So that was 2 dual core chips packaged together (not hyperthreaded.)

      2008-11-17 was the core i7, which may have been 4 cores on 1 silicon die, I don't really know.

    13. Re:Desktop by jonadab · · Score: 5, Interesting

      > Why not concentrate on tablets and phones first?

      Because people expect significantly more from desktops than from phones.

      The article says this:
      > The phone would be smart enough to not just be a computer but it could be my computer.

      That would make any sense at all if, in addition to processing power, the phone also had multiple gigabytes of primary memory and could utilize multiple peripherals for input (keyboard, mouse, etc) and output (monitors, printers, speaker systems), store hundreds of gigabytes of data, connect to multiple networks (including high-speed wired ones), and run desktop applications.

      Traditionally, even the smartest phones aren't expected to do any of that. It isn't mostly the processing power that's holding them back.

      --
      Cut that out, or I will ship you to Norilsk in a box.
    14. Re:Desktop by Type44Q · · Score: 5, Insightful

      This will then be your primary computing device that follows you around.

      This will then be your primary computing device that:

      A) you leave on the roof of your car.
      B) gets dropped in the toilet.
      C) you spill your beverage on.
      D) gets chewed up by your dog.
      E) you get mugged for.
      F) you leave in your hotel room.
      G) you have confiscated by the authorities (should you find yourself at the wrong place/time)
      H) gets reverse-engineered/stress-tested by your toddler

      Shall I continue? Seriously; fuck all this smartphone nonsense; give me a borderline-disposable Nokia 2600-series and I'll happily call it a day! :p

    15. Re:Desktop by Jamu · · Score: 1

      You probably already have that many. Except they're in your graphics card.

      --
      Who ordered that?
    16. Re:Desktop by Cyko_01 · · Score: 1

      No point. Windows/microsoft doesn't have a clue how to harness and use that many cores anyways

    17. Re:Desktop by SuricouRaven · · Score: 1

      Programming for the GPU is also trickier - it's a specialised task, and so hiring a programmer experienced in the field is going to be a bit trickier and more expensive than hiring a programmer who can work with general-purpose processors.

    18. Re:Desktop by Anonymous Coward · · Score: 0

      My desktop has 64 cores you insensitive clod.

    19. Re:Desktop by Lord+Lode · · Score: 1

      The 3930K seems the one exception, but it's Sandy Bridge, and LGA 2011 chipset is borderline consumer! For Ivy Bridge, the newer architecture, it only goes up to 4 cores. And for the next architecture, Haswell, not even released yet, Wikipedia has the following feature to say: "Mainstream up to quad-core.".

      I'm curious why Intel went from 1 core to 4 cores in a short period of time, and then remains stuck at 4 cores for more than half a decade.

    20. Re:Desktop by MachineShedFred · · Score: 1

      Your top of the line must be different than mine.

      Everyone always says that Apple makes consumer electronics, yet they have a 12-core Mac Pro that they've been selling for 2+ years now...

      --
      Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
    21. Re:Desktop by Prof.Phreak · · Score: 1

      When they said ``computing device that follows you around,'' they might've meant as ``your own personal terminator, follows you around and protects you'' :-)

      --

      "If anything can go wrong, it will." - Murphy

    22. Re:Desktop by Kjella · · Score: 1

      today, the top of the line Intel consumer desktop CPU's have 4 cores.

      More like 6. And the LGA2011 platform is also nice if for some reason you need 64GB (8x8GB) of RAM. The real issue for Intel is that most of things that "enthusiasts" care about don't scale that well. Sure, there's always people that need 3D rendering or whatever that could scale to 8+ cores easily, but they're more the "workstation" market. You're much more likely to find enthusiasts with a 3770K overclocked to 4-5 GHz and a few 2000+ MHz sticks of RAM than anything the LGA2011 offers. It's been that way since Sandy Bridge and I don't think it's going to change any time soon. Intel has absolutely zero competition either so I'm guessing IVB-E will be a huge disappointment when it finally arrives sometime after Haswell's release.

      --
      Live today, because you never know what tomorrow brings
    23. Re:Desktop by stephanruby · · Score: 1

      ...explained that with the prototype chip someone could, for instance, be encrypting an email while also working on other power-intensive apps at the same time — without hiccups. Same for HD video.

      It sounds to me like the LG Optimus G is already there with its quad-core Snapdragon.

      I just played with one last week and I was quite amazed. The Optimus G really blew out of the water my Jellybeans Nexus 7 and my Jellybeans Galaxy Nexus with its super smooth UI and its HD video multi-tasking capabilities.

    24. Re:Desktop by angelbar · · Score: 1

      Nope.... We need "omnipresent" or "ghost" devices... accounts perhaps? a (trekkie here). With that kind of processing power we only need a system to send verbal and sign commands and receive data on augmented reality

      --
      -no sig today-
    25. Re:Desktop by rolfwind · · Score: 2

      It's already been here since last year:
      http://slashdot.org/story/06/09/26/1937237/intel-pledges-80-core-processor-in-5-years

      Oops, that was just Intel's promise in 2006. Nvm, carry on.

    26. Re:Desktop by wooferhound · · Score: 1

      640 cores should be enough for anybody

      --
      We are Dead Stars looking back Up at the Sky
    27. Re:Desktop by Anonymous Coward · · Score: 0

      Mac Pro can be purchased with dual 6-core CPUs same as any generic computer for the past couple years not a single CPU with 12 cores. Of course they only run at 2.4Ghz unlike my generic computer which uses a 6-core 3.0Ghz (overclocked to 4.16Ghz) CPU.

    28. Re:Desktop by jythie · · Score: 2, Insightful

      Tablets and phones will likely continue to represent scaled down devices.

      48 core chips would be silly to release for small devices first. Developing software for a large number of cores is non-trivial, and is often inefficient or unstable... it is one of those things like AI or fusion that has always be 'right around the corner' for general purpose computing, but has neve really arrived.... college classes still teach the techniques poorly and people who 'learn it on their own' also usually produce poor architectures.

      So if such a chip is going to be introduced, desktops, with more resources to waste and higher user tolerance for system failures, would be the logical place.

    29. Re:Desktop by Anonymous Coward · · Score: 0

      Your meme is bad and you should feel bad.

    30. Re:Desktop by omnichad · · Score: 2

      The same reason they're stuck at 3-4GHz. Heat and power usage. They have to shrink to an even smaller manufacturing process to avoid those issues.

    31. Re:Desktop by Anonymous Coward · · Score: 0

      Look, everyone won't have a tablet. It's a stupid, useless thing.
      Phones, ok. But I still can see a lot more use for multicore in something that you can actually do work on.

    32. Re:Desktop by wisty · · Score: 1

      SSH + HDMI + USB + EC2 + Wifi + a fibre.

      Too much lag for gaming / interactive 3D work though.

    33. Re:Desktop by AwesomeMcgee · · Score: 1

      Great point though; these are devices that have to be bought a second time far more often than a desktop, didn't even think about the potential value to intel in that fact.

    34. Re:Desktop by TheRaven64 · · Score: 1

      Not really. GPUs and CPUs are both general purpose processors, just optimised for different classes of algorithm. If you have an algorithm that makes sense on a GPU, then it's pretty easy to write OpenCL C to implement it. It's often much easier to implement efficiently than the CPU version. Getting the most possible performance out of a GPU is not so easy, but the same is true for a CPU.

      --
      I am TheRaven on Soylent News
    35. Re:Desktop by Anonymous Coward · · Score: 0

      GPU cores are like glorified SIMD units. A group of cores work as one unit and all execute the same instructions. This alone would make it useless for general computing.

    36. Re:Desktop by Anonymous Coward · · Score: 4, Interesting

      From Rob Pike:

      Twenty years ago, you expected a phone to be provided everywhere you went, and that phone worked the same everywhere. At a friend's house, or a restaurant, or a hotel, or a pay phone, you could pick up the receiver and make a call. You didn't carry a phone around with you; phones were part of the infrastructure. Computers, well, that was a different story. As laptops came in, people started carrying computers around with them everywhere. The reason was to have the state stored on the computer, not the computer itself. You carry around a computer so you can access its disk.

      In summary, it used to be that phones worked without you having to carry them around, but computers only worked if you did carry one around with you. The solution to this inconsistency was to break the way phones worked rather than fix the way computers work.

    37. Re:Desktop by isama · · Score: 1

      yup, it's really easy. after 5 o clock i simply hit logoff and go home. how hard can it be?

    38. Re:Desktop by ldobehardcore · · Score: 1

      Utility foglets. A cloud of nanobots that'll be basically everything, computer, phone, network dropdown, AR, medical treatment/diagnostics, safety gear (instant helmet, seatbelt, car seat, bungie cord etc) anything really. Like little self-propelled FPGA computers.

      --
      Hectice, baby, Mercator says hello to you
    39. Re:Desktop by hairyfeet · · Score: 1

      Because desktops are where you work out the power problems and is the test bed of the parts that end up in mobile? Its no coincidence that the parts start on desktops, the GPUs start as discretes and IGPs, because you don't have to worry about hitting the insanely low heat and power requirements on the mobiles as you refine the process while you still get paid.

      This is one advantage AMD and Intel has had over their competitors, because they don't have to get the power and cooling right off the bat, they can sell the chips and refine them before trying to squeeze them into a little piece of plastic.

      --
      ACs don't waste your time replying, your posts are never seen by me.
    40. Re:Desktop by ByOhTek · · Score: 1

      Likely we'll either all have moved to cloud storage, or a better form of 'store at home, access anywhere via the internet' will have been developed for normal users.

      --
      Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
    41. Re:Desktop by hairyfeet · · Score: 1

      This is why I think the ones that come out with a $50 disposable smartphone are gonna make a fucking killing, because all it takes is seeing what a 2 year old can do to an iPhone ONE time to make you realize that having insanely expensive small devices is a BAD idea for families. In that vein I've seen a lot of my customers giving up their iPhones for those sub $100 Android phones at Walmart, they don't get locked into a contract and if the phone gets flushed or stomped you don't have a damned coronary like you do when its a $700 smartphone.

      --
      ACs don't waste your time replying, your posts are never seen by me.
    42. Re:Desktop by ByOhTek · · Score: 1

      Yes, and the mac pros typically have TWO processors not 1. We are talking about cores per CPU here. Try to keep up.

      --
      Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
    43. Re:Desktop by SniffTheGlove · · Score: 1

      My old Moto Atrix has HDMI and USB plus Wifi and also came with a Bluetooth keyboard and mouse. Great it is as I just need to hook up to a HDMI monitor and away I go however I love my desktop much much more as the phone is just powerful enought (2 core Tergra) to do basic stuff like email and web. If the phone can be come a truly portable comparative of today's modern desktop PC with Nvidia CUDA core then bring it on, til then it's still my desktop

    44. Re:Desktop by Anonymous Coward · · Score: 0

      Fanboi much? There are plenty of options to get 12 cores on an x64 PC without resorting to overpriced Apples.

    45. Re:Desktop by surd1618 · · Score: 1

      Making something like this work on a desktop would be trivial for such a low-power consumption project. It will arrive on the desktop tangentially.

    46. Re:Desktop by MachineShedFred · · Score: 0

      And 6 is still greater than 4. Try going back to first grade math, asshat.

      --
      Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
    47. Re:Desktop by fatphil · · Score: 1

      TV out and HDMI out have been on phones for a while.
      USB host mode, under the guise of OTG, likewise - so that's your peripherals and wired network.
      They've had a gig of main memory for a while too (and still swap!). Moore's law would make one expect that many gigabytes will be coming soon.
      64GB flash memory storage has been on phones for ages, there's no reason to not expect that to grow rapidly too.

      There's nothing about modern phone technology that isn't as capable as a typical desktop 5 years ago. There's no reason to believe that mobile technology won't be as capable as a typical desktop now.

      --
      Also FatPhil on SoylentNews, id 863
    48. Re:Desktop by oPless · · Score: 1

      ``your own personal terminator, follows you around and protects you'' :-)

      Okay, I'll get right on it.

      What size would you like? 4ft,5ft, 5ft 5", 6ft, 6ft6", or 7ft?

    49. Re:Desktop by Anonymous Coward · · Score: 0

      I think it would be nice to be able to put an OS I like on a 48 core chip instead of the UTTER SHIT which is available on phones and tablets.
      But that's just opinion so...

    50. Re:Desktop by Type44Q · · Score: 1

      Is that the meaning behind your slashdot ID? :)

    51. Re:Desktop by AuMatar · · Score: 1

      Also lack of need. How many apps does the average person run at once? How much of the CPU do they peg? Most apps can't easily be parallelized, so they'll never see a speedup from more cores. For the average web browser/email/office app person, there's no gain even on 2->4 cores.

      --
      I still have more fans than freaks. WTF is wrong with you people?
    52. Re:Desktop by nightfire-unique · · Score: 2

      I don't mean to be a dick here... but..

      I've never had any of that stuff happen, and I've been a smartphone user since before they were called smartphones.

      If you don't want one, you don't want one... fine. There's no need to rationalize it.

      --
      A government is a body of people notably ungoverned - AC
    53. Re:Desktop by godrik · · Score: 1

      It is called Xeon Phi (previously Intel MIC) and it is coming soon. I have been playing with prototypes for a year. On the server it will definitively happen. On the desktop, it depends on the price, but I do not see why not...

    54. Re:Desktop by Mike+Buddha · · Score: 1

      If the ultimate goal is to make portable power sipping devices, then do just that. Power consumption on Intel processors has been far behind ARM processors because the engineering tradeoffs made for x86 processors always assumed a desktop environment. You're going to make tradeoffs, and they should be made with the ultimate goal of having a mobile processor not a desktop.

      No doubt this technology will find its way to desktops, the same way that all-in-one desktops are using laptop components.

      --
      by Mike Buddha -- Someday the mountain might get him, but the law never will.
    55. Re:Desktop by omnichad · · Score: 1

      I don't know - web browsing probably could benefit even more from multicore than they do already. 1 entire core for flash, because it just performs so badly. 1 thread for rendering, 1 thread for executing javascript. Chrome puts each tab in its own process.

    56. Re:Desktop by Mike+Buddha · · Score: 5, Funny

      And I don't want my 48-core machine to be fucking mobile. It means I would then be expected to take my work everywhere with me. Fuck that. Until we learn how to respect the sanctity of vacation time in the US and bring up the average vacation length for workers to something near European standard I would rather see this in a non-mobile version first.

      I don't think you have anything to worry about. They're not making a portable deep fryer, this is a computer.

      --
      by Mike Buddha -- Someday the mountain might get him, but the law never will.
    57. Re:Desktop by Type44Q · · Score: 0

      Well, congrats; aren't you lucky (and/or a completely anal-retentive Aspie!). :p

    58. Re:Desktop by john.r.strohm · · Score: 1

      No, web browsing will not benefit significantly from multicore. The vast majority of the slowdown in web browsing is downloading and running gigabytes and gigabytes of Flash garbage animations and TV ads. AND YOU CAN'T STOP OR PAUSE THE ADS!

      This is not hypothetical. I bought a new machine earlier this year. It browsed like a bat out of hell with its tailfeathers on fire ... until ONE SITE I thought I needed to hit that day demanded I enable Flash. The moment I turned Flash on, all web browsing on that machine slowed to a crawl.

    59. Re:Desktop by Anonymous Coward · · Score: 0

      Hello, this is supposed to be a mobile device. I want a terminator that I can carry in my pants. Come with me if you want to live.

    60. Re:Desktop by AuMatar · · Score: 1

      Exactly what do you think is going on that requires full cores for each of those? The vast majority of load time is downloading files (images, videos, etc). The processing times are minimal. Worse, these activities are not isolated (rendering would require result updates from flash, jscript, etc), so breaking them into separate threads would just mean that one of them is stuck waiting for the others to finish. You'd have a single digit speed increase, and probably a low digit at that. It's just not a parallelizable problem.

      The gain Chrome got from breaking into multiple threads wasn't render speed. It was allowing you to change tabs if there was a misbehaved app (in flash, js, etc) on one tab. Having multiple threads on the same core would give them the same benefit, it was breaking it out into threads that was the win, not running them on parallel cores.

      --
      I still have more fans than freaks. WTF is wrong with you people?
    61. Re:Desktop by viperidaenz · · Score: 1

      Perhaps they think its easier to make a low power many-core CPU because you can completely switch off all but one core when there is little load. One idle low power core is probably going to consume less power than an idle high performance core.

    62. Re:Desktop by colesw · · Score: 1

      Xeon is not usually considered a consumer product but a business product.

    63. Re:Desktop by argStyopa · · Score: 3, Funny

      This will then be your primary computing device that:

      A) you leave on the roof of your car.
      B) gets dropped in the toilet.
      C) you spill your beverage on.
      D) gets chewed up by your dog.
      E) you get mugged for.
      F) you leave in your hotel room.
      G) you have confiscated by the authorities (should you find yourself at the wrong place/time)
      H) gets reverse-engineered/stress-tested by your toddler

      You say that as if such destruction wouldn't cause you to (insurance-covered or not) buy ANOTHER phone thus stimulating market activity, as if that's not good?
      Look guys, we need to work together to rescue the economy. /Keynes

      --
      -Styopa
    64. Re:Desktop by omnichad · · Score: 1

      OK. Flash can have its own core and the rest are on another. Rendering and compositing are not the same thing. They are two separate steps on modern browsers. Apparently Chrome uses a completely separate thread for compositing already - so that the page scrolls smoothly no matter what's going on.

      Rendering a DOM tree is a rather complex task - a breeze for a modern CPU, but can still drag on with poorly designed sites using nested tables. Would be nice if all of your pictures were already decompressed by the time the DOM is fully rendered.

      Trust me, on my connection I don't lose a lot of time on downloads. DNS latency is a factor, but even just 3 simultaneous HTTP connections are enough to finish that job very fast. And most sites use subdomains for image and CSS assets to get around HTTP connection limits.

      And as the site loads, you are sometimes completely re-rendering and re-compositing a page several times, especially if some content is loaded via AJAX after the main page loads.

    65. Re:Desktop by Type44Q · · Score: 1

      I like your economic recovery plan: it sounds like it implements both Supply-Side and Keynesian economics... how can that not be good?? ;)

    66. Re:Desktop by Sulphur · · Score: 1

      This will then be your primary computing device that:

      A) you leave on the roof of your car.
      B) gets dropped in the toilet.
      C) you spill your beverage on.
      D) gets chewed up by your dog.
      E) you get mugged for.
      F) you leave in your hotel room.
      G) you have confiscated by the authorities (should you find yourself at the wrong place/time)
      H) gets reverse-engineered/stress-tested by your toddler

      You say that as if such destruction wouldn't cause you to (insurance-covered or not) buy ANOTHER phone thus stimulating market activity, as if that's not good?
      Look guys, we need to work together to rescue the economy. /Keynes

      The smartphone is a stimulus?

    67. Re:Desktop by lister+king+of+smeg · · Score: 1

      Hello, this is supposed to be a mobile device.

      if it can walk on it own it is mobile no mater the size so why not go for the full sized summer glau model?

      I want a terminator that I can carry in my pants.

      wouldn't you rather get in her pants anyway. :-)

      --
      ---Saying gnome 3 is better than windows 8 not so much a compliment as it is damning with light praise.
    68. Re:Desktop by PRMan · · Score: 2

      Actually, the core multitasking engine as rewritten by Mark Russinovich and company in Windows now supports multiple threads very well in Windows 7. Run a program that is not written for multiple threads (even a batch file) and watch as all your CPUs level out to nearly equal amounts (within ~10% usually). This happens with nearly any process these days.

      So, 6 cores would work just fine for people. I think the real reason for the limit is that Intel doesn't want to cannibalize the overpriced virtualization server market.

      --
      Peter predicted that you would "deliberately forget" creation 2000 years ago...
    69. Re:Desktop by narcc · · Score: 1

      None of these things has ever happened to me, my wife, or (to my knowledge) anyone we know.

      We're either lucky, or we're just not as irresponsible as ... you?

    70. Re:Desktop by MachineShedFred · · Score: 1
      --
      Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
    71. Re:Desktop by Type44Q · · Score: 1
      Me, neither*; guess I'm just able to envision a wide range of possibilities.

      *There was one time when I went for a drive in the country with my girlfriend and I passed an area by the side of the road where people were four-wheeling in their pickups and ATV's. I couldn't resist; there was no way they were going to have all that fun to themselves, so I pulled off the road, aired-down the tires in my ancient '86 5000 Turbo Quattro (200 Quattro in Euro-speak) and proceeded to have the time of my life (well, outside the sack at any rate). However, when one of the puddles turned out to be a foot or so deeper than I expected, water came pouring in through the door seals... and when the car launched itself out of the puddle with a vengeance (damn,what a rush!), my old Nokia bounced out of its cubbyhole in the console and landed in the muddy water at our feet... however, it was nothing a little drying out on the dash couldn't fix. :)

    72. Re:Desktop by narcc · · Score: 1

      Developing software for a large number of cores is non-trivial, and is often inefficient or unstable...

      If only there was a mobile OS that's absurdly stable, efficient, and thrives in a multi-processor and/or mutli-core enviroment...

      While we're making wishes, why not make it POSIX compliant, and a hard real-time OS suitable for mission critical applications where failure can result in injury and death?

      We could call it QNX. Yeah, that would be cool.

    73. Re:Desktop by mlts · · Score: 1

      I can see a large number of cores being useful... but with a hypervisor and a VM layer. This way, a phone can have multiple "worlds" (to use the ARM term) where work stuff is completely separate from home stuff, and there is the ability to have each client be on a separate VM. Of course, there would need to be deduplication for apps (so the MS word viewer only takes up one instance, not for each VM) and encryption for the app data, but this isn't something that is impossible.

      What would be nice is to see cores that do different tasks, and a scheduler smart enough to use them. For example, one type core is mainly a GPU, another a FPU, another a slow (but energy saving) integer cruncher, another a relatively fast CPU, another a DSP, and another a FPGA. This would provide a device with immense flexibility. For optimizing calculations (such as AES which require array shifts), the FPGA can be configured and used. When the touchscreen is being used, the tasks that monitor that can be shifted from a slow core to a faster one until a few seconds elapse after the finger is removed. If the hypervisor is crunching a lot of context shifts, it can move itself to a faster core to handle the load, then back down to a slower core.

      Of course, the car example is simple -- check the task at hand and find the right engine to suit. A low-wattage generator gets a one cylinder, two stroke engine. A 20,000 pound freight shipment gets a turbo diesel. A motorcycle gets a small gasoline engine.

      This also would be useful if coupled with tiered storage like a mini SAN controller. SLC flash gets used for data that is more often accessed than stuff on slower MLC cells.

    74. Re:Desktop by Anonymous Coward · · Score: 0

      Not to mention that most software out there can't leverage more than 1 or 2 cores to begin with.

    75. Re:Desktop by Type44Q · · Score: 1

      Something tells me neither you nor your wife have ever had an experience like that, either... :p

    76. Re:Desktop by Anonymous Coward · · Score: 0

      Software doesn't even use the four-cores I have now. More cores are useless unless the software can take advantage of it.

    77. Re:Desktop by Anonymous Coward · · Score: 0

      Your top of the line must be different than mine.

      Everyone always says that Apple makes consumer electronics, yet they have a 12-core Mac Pro that they've been selling for 2+ years now...

      uh and they are workstations not consumer pcs.
      also they are xeons, how many people buy xeons? people think apple makes consumer electronics because apple markets themselves that way.

    78. Re:Desktop by Darinbob · · Score: 2

      People think too one-dimensionally I think. They figure out that 2 or 4 cores means something is faster, therefore 48 cores would be even better (and then they get so excited they say something incredibly stupid like "redefine mobile devices"). But think in more than one dimeions. Maybe make the base CPU faster instead, add more memory, optimize things that are bottlenecks, etc.

      48 cores is a really bad idea given the clumsy shared memory model current implementations use.

      And what slows down smart phones anyway, is it that they weren't threaded enough and thus more cores will fix it, or is it because some resources are bottlenecked? Plus figure out what it really is that you want to be fast on a mobile device.

      Maybe it's a good idea to have slower devices that save more battery power than having lots of idle cpu cores sucking up juice.

    79. Re:Desktop by Darinbob · · Score: 1

      Mobile devices as far as I can tell are not computing devices. Or at least I've never seen anyone use one for computing or for any other purpose that requires having a fast computer. Instead they're all used for basically two things; media consumption and communication. These things are not "computing".

    80. Re:Desktop by Darinbob · · Score: 1

      If you put a core to a well defined task then it can be a separate CPU altogether and doesn't need to be a core! Ie, you don't have one of the cores do signal processing, you offload that to a DSP instead where it can do it's work with it's own resources. You don't use a core to do graphics, you use a GPU with it's own resources.

      Unless definitions have changed, "core" means a part of the general purpose CPU that shares the memory bus with other identical cores.

    81. Re:Desktop by Anonymous Coward · · Score: 0

      If you could get a couple terabytes of memory on the device then professionals could :
      7) use a partition for work that is fully encrypted. Heck you could up the encryption to 256 bit levels with that kind of power.
      8) fully support encrpyted email and multiple emails reporting to one device.
      9) this would push wireless internet providers into the forefront... where they want to be anyway
      10) --- final nail in the coffin for landlines
      11) --- possibly first nail in the coffin for cable companies that have not merge with wireless providers
      12) would force companies such as Dell and Gateway to fully enter the smartphone market - driving down the cost of smart phones
      13) remember thin client -- yeah it's back
      14) Welcome the newest version of Android -- now complete with memory management and process control -- kinda like a real OS ( ooops...)
      15) the advent of the mobile virus -- much like the rhino virus, contact, and data fluids can now infect your phone
      16) security is paramount -- really its even more important

    82. Re:Desktop by Darinbob · · Score: 1

      Right, because you want generic programmers dealing with signal processing, not programmers who are specialized in signal processing.

    83. Re:Desktop by aczisny · · Score: 2

      Power consumption on Intel processors has been far behind ARM processors because the engineering tradeoffs made for x86 processors always assumed a desktop environment

      While that was true, you might check out how the Razr i (which has an Intel chip) compares to the Razr m (snapdragon). They're nearly identical phones by the same manufacturer, and the Intel chip performs better than the snapdragon and has better battery life. Here's Engadget's review. Maybe someday the "x86 is horribly power inefficient because of a huge decoder" meme will finally die.

      --
      Now, landing thrusters.. landing thrusters, hmm. Now if I were a landing thruster, which one of these would I be?
    84. Re:Desktop by narcc · · Score: 1

      Your assumption is correct. My ancient '86 Saab 900 has no turbo.

    85. Re:Desktop by jeffmeden · · Score: 1

      I guess it works for me as I can easily separate my working life from my personal life (Yes, I live in North America but I derived my only-work-at-work mentality from the other side of the pond).

      Or in the smartphone era, "sync email after hours?" OFF. How hard was that box to find?

    86. Re:Desktop by jeffmeden · · Score: 2

      That would make any sense at all if, in addition to processing power, the phone also had multiple gigabytes of primary memory and could utilize multiple peripherals for input (keyboard, mouse, etc) and output (monitors, printers, speaker systems), store hundreds of gigabytes of data, connect to multiple networks (including high-speed wired ones).

      Let's see: 2GB primary memory (a little light but hey, it's almost 6 months old) 96GB nonvolatile storage, dual core 1.5ghz, 100+Mbit A/B/G/N wifi plus 100Mbit LTE (wires are for old people)... full monitor output via HDMI, plus bluetooth or usb peripherals. I'm sure in another year or two they will be able to add whatever it is that is still missing...

    87. Re:Desktop by jeffmeden · · Score: 1

      This will then be your primary computing device that follows you around.

      This will then be your primary computing device that:

      A) you leave on the roof of your car.

      B) gets dropped in the toilet.

      C) you spill your beverage on.

      D) gets chewed up by your dog.

      E) you get mugged for.

      F) you leave in your hotel room.

      G) you have confiscated by the authorities (should you find yourself at the wrong place/time)

      H) gets reverse-engineered/stress-tested by your toddler

      Shall I continue? Seriously; fuck all this smartphone nonsense; give me a borderline-disposable Nokia 2600-series and I'll happily call it a day! :p

      Seriously? After about 5 years of carrying a smartphone on a daily basis the only test mine have faced have been H, and they all managed to pass (i have a drawer of obsolete but otherwise perfect smartphones...)

    88. Re:Desktop by Anonymous Coward · · Score: 0

      More cores doesn't have to mean more performance.
      The reason for pushing multicore is exactly the smaller power envelope.
      Theoretically, if you split up a process on two cores with half the clock-speed,
      you use half the power.

    89. Re:Desktop by leromarinvit · · Score: 3, Funny

      I don't think you have anything to worry about. They're not making a portable deep fryer, this is a computer.

      Just put today's 48 core chip in a laptop, and there's your portable fryer.

      --
      Proud member of the Ferengi Socialist Party.
    90. Re:Desktop by Nyder · · Score: 1

      Let's put a 48-core processor on a desktop or laptop before we talk about tablets or phones...

      Lets first discovery batteries that can actually last more then a few hours...

      --
      Be seeing you...
    91. Re:Desktop by samoanbiscuit · · Score: 1

      The only thing the Razr i beats the Razr M on is the Sunspider benchmark, which has been optimized for x86 processors since it's inception. By every other metric it is inferior. Not embarrassingly so, but enough to let Qualcomm and nVidia executives sleep soundly at night. "x86 is horribly power inefficient because of a huge decoder" is not just a meme, it's also true. Intel's plan of attack is to shrink their process nodes faster than the competition can keep up, using the power and heat shaved off by the shrinks to make up for a less efficient architecture. Intel needs to ramp their process size down as quickly as possible, ARM needs to ramp up their design's power and IPC as quickly as possible. It's an interesting time in the tech market.

    92. Re:Desktop by ByOhTek · · Score: 1

      I never tried to argue otherwise, try going back to first grade reading.

      --
      Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
    93. Re:Desktop by Anonymous Coward · · Score: 1

      I develop software, and I'm a self taught coder. I had to learn on my own because college wasn't available to 8 year olds in the 80's... Anyhow, it takes significantly more time to develop programs that can scale their workloads up to an arbitrary number of cores. Furthermore, many problems are linear in nature and can't be parallelized easily if at all. Even if you do get a massive number of parallel processors working, you can end up being a lot slower than the single core implementation with equivalent (summed) GHz due to the time wasted waiting for one core to sync with another core via spinlocks and other such mutexes. It's a shame that Erlang isn't more popular, I think it was heading in the right direction. A language / compiler that's smart enough to take code, detect isolated bits and split programs into multiple cores is what's needed. Compilers that automatically utilize SIMD instructions are getting better, but automatic thread separation? -- Unfortunately that will never be as good as if a person manually designs their code to be parallelized.

      If it weren't for me taking an interest in multi-core software design I wouldn't have learned it at all, often people have no other recourse but to 'learned it on their own', because as you say college courses don't adequately cover the subject. This means that almost anyone who produces an architecture that isn't 'poor' is self taught... I totally agree with the other parts of your statements though.

    94. Re:Desktop by Anonymous Coward · · Score: 0

      Moron.

    95. Re:Desktop by aczisny · · Score: 1

      The only thing the Razr i beats the Razr M on is the Sunspider benchmark, which has been optimized for x86 processors since it's inception. By every other metric it is inferior

      And battery life, where it gets about an extra hour of active battery, and a significantly longer standby. You know, that thing Intel's obviously going to suck at because it only produces high power chips, right? You don't think it's at all interesting that a 32nm x86 chip draws less power than a 28nm ARM chip with similar performance? I'm sure that Intel's process is better (Has anyone else shipped product with high-K metal gates yet?) , but I'd say that's a pretty big indication that the tables are turning on ARM.

      The decoder things drives me nuts. People are arguing that something that, generously, might be .1% of the die is the huge power hog. Really? It's got relatively high power consumption and you can't turn it off sure, but this is hardly the thing driving your overall power envelope.

      --
      Now, landing thrusters.. landing thrusters, hmm. Now if I were a landing thruster, which one of these would I be?
    96. Re:Desktop by SniffTheGlove · · Score: 1

      Nice one :-)

    97. Re:Desktop by samoanbiscuit · · Score: 1

      It's interesting to note that the Razr M reviewed had LTE, while the Razr i does not. I wonder how that affected battery life between otherwise identical phones... Engadget is good for news and rumours, but they're not the best reviewers out there. I've read much more detailed reviews and write-ups about Medfield, and it DOES suck down the power compared to equivalent ARM SoCs. It's a very good showing, but still not there. Anandtech and Tom's Hardware both had a look at the Medfield prototypes and that Orange phone, and they seem to agree that Medfield is more power hungry. Also, Medfield's GPU is underpowered, and the next generation had better be a big improvement if they want to have any hopes of getting a design win for devices with retina-like displays.

      It's not just the decoder that makes x86 inefficient, it's all the little design decisions that led them into the server room and into workstations. It took Intel several tries before they put out a decent laptop CPU.

    98. Re:Desktop by Anonymous Coward · · Score: 0

      Desktops are a declining market.

    99. Re:Desktop by armanox · · Score: 1

      Yes. The Nehalem based Core i5 (some) and i7 processors were quad cores on 1 die - the i7 also including Hyper-Threading (some of the i5 line were DC chips with HT instead of QC).

      --
      I'm starting to think GNU is the problem with "GNU/Linux" these days.
    100. Re:Desktop by MachineShedFred · · Score: 1

      Ok, I know you're a little slow, so I'll try to catch you up with the rest of us.

      Mac Pros use CPUs that either have 4, or 6 cores per CPU. This is how they can have configurations of either 8 or 12 cores - by combining either two 4-core CPUs, or two 6-core CPUs.

      Thus my comment that 6 is more than 4. I was talking about cores per CPU too, but you were too thick to catch on. What, did you think that they used 3 CPUs or something?

      You do know that Intel makes a 6-core i7, right? Does that not count either?

      --
      Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
    101. Re:Desktop by lsatenstein · · Score: 1

      Let's put a 48-core processor on a desktop or laptop before we talk about tablets or phones...

      ====
      Adding cores rarely translates into increased throughput.

      Think of this idea from queuing theory. Which is more efficient, one server running at 2x speed, or two servers running at x speed? Equate cpu and server as one.

      If the two servers are contending for a common serialized resource (disk, ssd, etc) , then they lock each other out until the serialized task is completed. Queuing theory suggests that they trip over each other and this results in less throughput than with the one server running twice as fast. When things are slow, many servers sits idle. This is true for a 48 server situation.
      There is also a large overhead that is required to manage the active servers from the 48 server.

      For large numbers of cores, the scheduling algorthms have to change. My view is that there will be a queue of available cpus, and a task dispatcher that selects the first cpu from the queue. When that CPU based task is finished, the cpu goes back into the queue. There will be rules to indicate affinity to cpu or group, requiring a task to run only on a specific cpu, or independence and freedom of choice. We hardly handle quad cores well, I wait to think of the design and Q/A tasks leadingup to a generalized task scheduler for large core systems.

      If all servers are identical, one can return it to the back of the line. If it is a faster server, it would go closer to the front.
      Naturally, one hopes that resource contention concerns can be handled by the dispatcher-scheduler.

      My thoughts about the above are mine. I don't have task manager design experience, but I only have my years of experience to go on.

      Thanks for the read.

      --
      Leslie Satenstein Montreal Quebec Canada
    102. Re:Desktop by angelbar · · Score: 1

      Thats excelent... really But I prefeer my technical breakthroughs on multiples of 5 years. (And the gray goo its scary)

      --
      -no sig today-
  2. excessive by fizzer06 · · Score: 2, Insightful

    will it include a car battery?

    1. Re:excessive by Anonymous Coward · · Score: 0

      More cores (with better parallelism) should mean _less_ power consumption, if cores can be powered up and down as needed.

    2. Re:excessive by tom17 · · Score: 1

      I think this way forward likely HELPS power usage if anything.

      Let's take 2 power-equivalent processors. Both use 1W at the lowest non-idle power setting. Processor A is a 2 core model, B is a 48 core.

      Each core on the 2 core model A uses approx 0.5W (ignoring overheads etc)
      Each core on the 48 core model B uses approx 0.02W (ignoring overheads etc)
      In this instance, it's likely that each of the model B cores will be one 24th as computationally powerful as a model A core.

      If you have some trivial tasks (i.e. most stuff that is running when you are not doing heavy lifting) running, on model A they might use 1 core on it's lowest power setting of 0.5W.
      That same trivial task on the 48 core model B might still only use 1 core which would then obviously greatly reduce the power usage. This would still work out as a power saving if it needed more than 1 model B core (anywhere up to around 15-20 cores depending on SMP overhead/efficiency)

      Of course, this is only a generalisation, ignores overheads and will only help with processes that need significantly less than 1 core on a conventional 1/2/4 core architecture; which is most of the non-lifting workload of any CPU.

      I think it's a good thing...

    3. Re:excessive by LordLimecat · · Score: 1

      (ignoring overheads etc)

      I think I found the problem with your assumptions. Ignoring overheads can get you into a lot of trouble in parallel computing.

    4. Re:excessive by ifiwereasculptor · · Score: 1

      That's not how those things scale, though. In your example, for you device to be borderline useable, you'd have to use about 18 cores, minimum, at most times. Assuming perfectly threaded software, inifnitely scalable. Sleep state is a different beast, but then you can simply put an extra low power core that takes over in such cases, like the Tegra 3. Consider there are tons of ARM clusters around. When power consumption is equal, a system with dozens of cores is incredibly inneficient against an Intel Core for any given workload. You don't even have to look at ARM, look at AMD's Bulldozer, here to prove that scalability isn't a magic bullet. Multiple cores are for big computational demands when frugality isn't that big a deal.

    5. Re:excessive by jythie · · Score: 1

      And flying cars should mean an end to traffic congestion...

      This type of parallelism with cores switching on and off as been done before, and it worked well in the domain of supercomputers and other specialized devices, but the programming for them tended to be rather unusual and translated poorly into general purpose use.

    6. Re:excessive by Anonymous Coward · · Score: 0

      Modern CPUs already power gate their cores and have the ability to not only sleep the cores, but completely shut-off power and remove power leakage from that silicon.

      The only difference is the number of cores.

    7. Re:excessive by Anonymous Coward · · Score: 0

      Since individual processes should have extremely little resource sharing, overhead should not be an issue.

    8. Re:excessive by jythie · · Score: 1

      That is a significant difference. The problem tends to scale exponentially.. there are good reasons why were have not seen a huge rise in the number of processors over the last 2 decades, with most staying in the 1-8 range. They have made progress, but it isn't quite as simple as the low core count solutions make it sound.

  3. 48 cores on a mobile device by Anonymous Coward · · Score: 1

    And a battery life of four fifths of a second!

  4. Before we know it we will be at 640 cores by tom17 · · Score: 4, Funny

    And we can stop then because, well, you know, 640 cores should be enough for anyone.

    1. Re:Before we know it we will be at 640 cores by robthebloke · · Score: 3, Funny

      Well, you could selotape together 640 Sinclair ZX81's, which would give you 640 Z80 cores, AND 640Kb RAM!!!! Combine that with 640 x 64 x 48 pixel displays, and 640 cassette tape interconnects, and you'd have far more computing power than anyone could ever need!

    2. Re:Before we know it we will be at 640 cores by Bill_the_Engineer · · Score: 5, Funny

      I was looking at my razor and thinking I should have more than 6 blades!

      --
      These comments are my own and do not necessarily reflect the views or opinions of my employer or colleagues...
    3. Re:Before we know it we will be at 640 cores by Anonymous Coward · · Score: 2, Funny

      Next up, Sinclair C5 NASCAR...

    4. Re:Before we know it we will be at 640 cores by Dogbertius · · Score: 1

      Well NOW there is!

      http://www.youtube.com/watch?v=4F7TMlrDXtw

      Spishak present the Mach20, to guarantee you the closest shave of your life!

    5. Re:Before we know it we will be at 640 cores by Anonymous Coward · · Score: 0
    6. Re:Before we know it we will be at 640 cores by Anonymous Coward · · Score: 0

      640 blades should be enough for anyone.

      I think I can safely stand by this prediction.

    7. Re:Before we know it we will be at 640 cores by Anonymous Coward · · Score: 0

      Apparently the very first episode of SNL had a skit making fun of 3 blade razor called triple trac

    8. Re:Before we know it we will be at 640 cores by Anonymous Coward · · Score: 0

      Watch out for those cases where all the other blades are waiting for that middle one, or when they are all trying to cut the same facial hair, all at once.

  5. This is just.... by StrayEddy · · Score: 0

    overkill !

    1. Re:This is just.... by Anonymous Coward · · Score: 0

      Overkill is underrated!

    2. Re:This is just.... by Anonymous Coward · · Score: 1

      being underrated is overrated.

    3. Re:This is just.... by Anonymous Coward · · Score: 1

      Yes. IMHO it's much better than Ace of Spades.

    4. Re:This is just.... by HeadBanger606 · · Score: 1

      Haha, had a genuine laugh at this. :D Was just going to make a similar reference.

      --
      --- Amateur musician: http://josh.morine.net/headbanger/
    5. Re:This is just.... by ZombieThoughts · · Score: 1

      With a name like yours, you would have had to.

  6. What can't it do! by Anonymous Coward · · Score: 0

    It can also julien fries and deep fry them while you play games and hd movies at the same time on your 2 inch screen!

    1. Re:What can't it do! by Jeng · · Score: 1

      What it can't do is pass a charging station.

      --
      Don't know something? Look it up. Still don't know? Then ask.
  7. It's just Larrabee by Balial · · Score: 4, Insightful

    Keep floggin' that dead Larrabee horse, Intel.

  8. Not exactly a unique idea... by hattig · · Score: 1

    Given that even today's mobile phone SoCs have dedicated hardware "cores" for encryption, video encoding and decoding, etc, this is only Intel trying to generalise the functionality back into the CPU - which is pretty much all Intel know - rather than the more suitable, lower power, dedicated function blocks that are but pinheads even on today's SoCs, never mind the 10nm SoCs in years to come.

    Today's quad-core mobile phone SoC on 32nm could be a 16-core cluster-on-a-chip on 16nm, and 64-cores on 10nm.

    Never mind the compute-assist on the GPUs in these SoCs - Exynos 5 is over 70 GFLOPS for example, and supports OpenCL 1.1. Never mind the 4K video decode/encode support...

  9. Maybe...but not soon by Overzeetop · · Score: 4, Interesting

    Sure, you could put the power of today's typical desktop in a phone with such a beast, but by then we should have desktop boxes with an order of magnitude more power than one - and we'll find a way to "need" that extra power on the desktop. It's not just about chip capability and battery life - I'm mean you can always plug in a phone and run it full power 24/7. But you start running up against the limits of thermal dissipation. It's no surprise that maximum TDP has not changed a whole lot on the desktop per processor. We're still limited by the ability to aircool a chip that's really a 130+/- watt heater. That won't change. Remember also that until you standardize a dock with a real video connector (or crazy fast wireless video - not this compressed crap we use for movies), you're still limited to that little tiny window on the portable device.

    The biggest potential savings is if they can shut down 47 cores and run one thread at low power when I'm not "using" it so the battery lasts as long as possible.

    --
    Is it just my observation, or are there way too many stupid people in the world?
    1. Re:Maybe...but not soon by david.given · · Score: 2

      A while back I was working on site with a customer... a major featurephone manufacturer, who will remain nameless so as to protect the incompetent. We discovered that the phone would crash if our software ran the CPU flat-out, even if it was at a lower priority thread than anything else.

      It turns out that the sorry excuse for an operating system this thing ran was doing system-critical tasks in the idle thread, so if the CPU didn't idle some message queues would overflow and the OS would reboot. When I asked if there was a way around this, the response was that it was deliberate because the hardware would overheat if the CPU ran continuously for more than about thirty seconds at a time.

      (This OS, BTW, was such an epic failure that when it was cancelled the entire staff went with it... and so did the building.)

    2. Re:Maybe...but not soon by Anonymous Coward · · Score: 0

      > The biggest potential savings is if they can shut down 47 cores and run one thread at low power when I'm not "using" it so the battery lasts as long as possible.

      But how will your phone display it's AeroAquaMetroChrome then, huh?

    3. Re:Maybe...but not soon by Jesus_666 · · Score: 1

      The solution is obvious: Just make each core as powerful as a current singlecore desktop CPU and have all of them work at the same time, yet stay within a passively-cooled mobile processor heat envelope. That's how I understand TFA: More cores = every parallelizable task becomes trivial for a handheld mobile device because apparently waste heat doesn't scale with number of active cores. They even point out that you can run "some big apps" (presumably apps which would be taxing on a current mobile processor) and "nothing will steal performance from each other" because a small number of cores will be potent enough to run one of these "big apps". Seems to me that each core would be of similar power to a current mobile processor core.

      This chip can enable nice things such as more efficient power usage - but I agree that "magically turn mobile phones into full-fledged desktop PCs with no heat problems whatsoever" seems rather unlikely.

      Well, or future smartphones will come with a watercooling rig with a huge radiator attached to them. That might work.

      --
      USE HOT GRITS WITH STATUE OF NATALIE PORTMAN (NAKED AND PETRIFIED)
    4. Re:Maybe...but not soon by Overzeetop · · Score: 1

      I don't mind shutting most cores down, or running them at kHz speed for battery, and ramping up for hard core use. Even ignoring thermal dissipation, you still have the wireless communication they noted. Right now, I work on a 4960x1600 desktop. I certainly hope in 5-10 years I'm not still stuck at 90dpi on my desktop. But even at 2560x1600 - my central monitor - I need 5.8Gbps to drive the monitor at 60Hz with no overhead, uncompressed. Simply moving to 180dpi on my setup would mean 45Gbps of information be passed over a wireless link. That seems...unlikely.

      --
      Is it just my observation, or are there way too many stupid people in the world?
  10. 3 options, pick any 2. by Anonymous Coward · · Score: 2, Insightful

    Hmm. 48 cores. More processing power. Better battery life. And how large of a backpack will I need to transport those long-life batteries? Just trying to plan ahead.

  11. Imagine a Beowulf cluster... by ducomputergeek · · Score: 5, Funny

    had to be said.

    --
    "The problem with socialism is eventually you run out of other people's money" - Thatcher.
    1. Re:Imagine a Beowulf cluster... by slim · · Score: 1

      +1 Obligatory

    2. Re:Imagine a Beowulf cluster... by Anonymous Coward · · Score: 0

      No.

    3. Re:Imagine a Beowulf cluster... by rasmusbr · · Score: 2

      You're joking, but half-seriously, imagine if a manufacturer would make a cheap 'tablet' without a display and with physical network ports and just enough ventilation that you could stack multiple tablets as high as you please without overheating. Imagine a Beowulf cluster of those!

      I imagine the power adapters and the power strips connected in series would look really silly.

    4. Re:Imagine a Beowulf cluster... by oodaloop · · Score: 1

      What about a beowulf cluster of raspberry pi's? Mmmmmmmmmmmm pi.

      --
      Tic-Tac-Toe, Global Thermonuclear War, and relationships all have the same winning move.
  12. Dirk Meyer's last words by CajunArson · · Score: 5, Funny

    Just before the AMD board executed him, Dirk Meyer screamed out: "MOAR COARZ!" And today we have Bulldozer.

    The moral of the story is, MOAR != more all of the time. Especially in a freakin' cellphone, where, despite what some Slashdotters think, the primary use case is *not* performing massively parallel scientific simulations.

    --
    AntiFA: An abbreviation for Anti First Amendment.
    1. Re:Dirk Meyer's last words by Anonymous Coward · · Score: 0

      The primary use of a phone is to make phone calls. No one wants a fancy touch screen and internet access, it's mobile phone used only to make calls. The point I'm making is you may not foresee a use for all those cores in a phone, but make it available and I can almost guarantee you someone will.

  13. Architecture? by Anonymous Coward · · Score: 0

    Probably all x86 cores too.
    Oh the poor suffering batteries (and us poor users who will need a wheelbarrow for them.)

  14. If only by Anonymous Coward · · Score: 0

    diminishing returns weren't such a big problem with multi-core processors. The speed of most applications will still be limited by how fast a single core can do work, to say nothing of relatively slow I/O.

  15. And programmers will take 40 years. by Anonymous Coward · · Score: 0

    This will hit the market, and all phones will seem much slower for many years, as too many developers think in terms of single-threaded code.

    This thing will purr along, saturated at 2.08% CPU usage most of the time.

  16. 10 years? by Omnifarious · · Score: 3, Informative

    Hah! The Parallela Kickstarter project was from a group that already had a 64 core CPU that consumed only a watt or 4 of power when running full-tilt. If this takes Intel 10 years to design they'll be left in the dust.

    BTW, they wanted the money to fund developing the mask for the Fab technique needed to mass produce them for $100 apiece instead of the few hundred apiece they now cost to manufacture. The chips already exist.

    1. Re:10 years? by nanodroid · · Score: 4, Informative

      Context:

      http://www.kickstarter.com/projects/adapteva/parallella-a-supercomputer-for-everyone

    2. Re:10 years? by Anonymous Coward · · Score: 0

      It's not a 64-core CPU, it's a floating-point processor. The basic Parallela model comes with a 2-core ARM CPU.

    3. Re:10 years? by Anonymous Coward · · Score: 0

      It is speculated that the CPU and GPU in an iPhone has a TDP of 0.8 watts (that's at full load). If the Parallela device uses 1 watt when it idling doesn't bode well for mobile applications.

    4. Re:10 years? by jomegat · · Score: 1

      It also comes with either a 16- or 64-core floating-point Epiphany chip.

      --

      In theory, practice and theory are the same. In practice, they're not.

    5. Re:10 years? by Anonymous Coward · · Score: 0

      One would assume that Intel doesn't have an architecture that's as brain-dead as the Paralella.

    6. Re:10 years? by cavok · · Score: 1

      It's not a 64-core CPU, it's a floating-point processor. The basic Parallela model comes with a 2-core ARM CPU.

      Ok, they are not full x86 cores - this may also explain the different power consumption - but Intel is targeting the same use(r)s of Adapteva. By the time Intel is ready I hope Linux has taken the world and GPUs are eventually implemented blob-free on top of beds of the Adapteva chips.

    7. Re:10 years? by Omnifarious · · Score: 1

      No, that's an inaccurate characterization. It is a full CPU that is optimized for floating point operations. That's not the same thing at all as a dedicated floating point co-processor like the old 80387. You could, for example, use it to perform SHA-1 calculations, though that would be a bit of a waste.

    8. Re:10 years? by Omnifarious · · Score: 1

      What's brain-dead about their architecture? The NUMA model with each core having fast local memory seems like a good way to go to me. It's not very feasible to have memory that can respond equally to all cores that's not a bottleneck.

    9. Re:10 years? by Omnifarious · · Score: 1

      Thanks! I was typing that out on my phone and couldn't do my normal research and hunting down links.

    10. Re:10 years? by viperidaenz · · Score: 1

      Not only is it not 64 general purpose cores, it doesn't even support double precision floating point in its 64 floating point cores. Its main cores are two ARM ones.

    11. Re:10 years? by viperidaenz · · Score: 1

      because 64k of RAM is enough for any core? It's a severely limited architecture.

    12. Re:10 years? by Anonymous Coward · · Score: 0

      The 64-cores is still part of the coprocessor, not the central processing unit. Saying the Parallela has a 64-core CPU is factually incorrect.

    13. Re:10 years? by Omnifarious · · Score: 1

      Well, 1Mib of RAM, not 64Kib. I admit that's a bit small by modern standards. Though it isn't really that tiny. There's a lot you can do with that architecture. It would be nice if there was more memory per core.

      Is it possible to get a whole ton more memory on a CPU these days? What would you do?

    14. Re:10 years? by Omnifarious · · Score: 1

      Is it or is it not true that you could run a program that did any arbitrary thing on each core? The example I saw had each core running general purpose C code. That seems pretty CPU-like to me.

    15. Re:10 years? by Anonymous Coward · · Score: 0

      The language being used does not determine if it's a CPU (Central Processing Unit), neither does it determine if it's a General-Purpose CPU.

      Programs would have to be specifically designed and modified to work with Adapteva's Epiphany chips. Could mobile phones or computers use this type of processors? Sure, but it cannot replace a CPU on its own, even the Parallela device requires a separate ARM CPU to operate.

      You don't seem to understand what the Parallela project is. I recommend starting here to learn more about this proposed platform and what its pitfalls are.

    16. Re:10 years? by viperidaenz · · Score: 1

      The Epiphany processor has 32KiB of RAM per core, split in to 8 4KiB pages. Each processor only has 0x7FFF of local address space. A theoretical limit of 128KiB and an implementation of 32KiB. All other memory access is either to main memory or to the memory of another node.

    17. Re:10 years? by Anonymous Coward · · Score: 0

      Hah! The Parallela Kickstarter project was from a group that already had a 64 core CPU that consumed only a watt or 4 of power when running full-tilt. If this takes Intel 10 years to design they'll be left in the dust.

      You are being stupid. Don't be stupid.

      Intel is way ahead of Parallela in terms of meaningful impact on supercomputers from multicore chips. Look up Xeon Phi. It's a 64-core Intel x86 chip which can do more than 1 teraflop (which is much more than this toy Parallela chip). The first supercomputer using Xeon Phi is going to come online January 2013, and it's expected to rank as the #3 supercomputer in the world on the TOP500 list.

      Don't get me wrong with the "toy" remark, it's a cool little chip, and (if what's said about it is actually true) an impressive achievement for a 3 man team. But merely having a lot of cores doesn't mean it automatically competes head to head with everything else under the sun. It isn't going to be a revolution in either the big supercomputer or the ultra low power space, nor is a 3-man team actually capable of keeping pace with Intel's resources (and I'm pretty sure those 3 dudes would tell you the same). Parallela's chip is going to be most useful in academic research, and for geeks to play with.

    18. Re:10 years? by Anonymous Coward · · Score: 0

      Yeah, I was thinking that. BUT the Parallela 64-core chip takes 5W - that's 1A at 5V, which is what the iPhone charger provides. So it'll suck on battery life.

      Looking forward to my Epiphany-III board :)

    19. Re:10 years? by Omnifarious · · Score: 1

      I will have to verify that, as that's at odds with my understanding. But if you're right, that is pretty limited. But I still think the idea is pretty interesting. Maybe Intel can do better. But I still maintain that they'd better take less than 10 years to do it.

  17. And in 20 to 30 years... by Anonymous Coward · · Score: 0

    We could have 128 core chips. It will open up a whole new world of possibilities.

    1. Re:And in 20 to 30 years... by Anonymous Coward · · Score: 1

      Yeah, you'll be able to run 128 instances of Angry Birds! No wonder you're excited!

  18. My Computer? by superdave80 · · Score: 1

    The phone would be smart enough to not just be a computer but it could be my computer.

    Anyone have a clue what he means by "my computer"? All the computers I currently own and use are my computer already.

  19. 48 Cores? by QuantumHack · · Score: 5, Informative

    Not useful without a serious change in computer architecture.

    Amdahl's law. It's a bitch, baby.

    --
    www.backwoodsengineer.com
    1. Re:48 Cores? by c0lo · · Score: 4, Informative

      I see your Amdahl's Law and raise you a Gustafson's law.

      --
      Questions raise, answers kill. Raise questions to stay alive.
    2. Re:48 Cores? by Anonymous Coward · · Score: 0

      I don't think most vendors give a crap about Amdahl's law. All they want is a larger penis than Apple.

    3. Re:48 Cores? by Anonymous Coward · · Score: 0

      bullshit. Gustafson's assumes an arbitrarily large dataset. Graphics are the only arbitrarily large dataset anything deals with, and playing Duke Nukem Forever isn't going to happen on a mobile phone.

    4. Re:48 Cores? by Anonymous Coward · · Score: 0

      And if you don't understand either, Amdahl's law is like doing a Google search for a topic and knowing at minimum it is going to take you so much time to find what you are looking for no matter how many tabs you have open, Gustafson's law is more like you start looking for something particular in Google and even though you just end up just looking at porn, pretty quickly you are quite satisfied with the results because you got a lot more than what you were looking for.

    5. Re:48 Cores? by c0lo · · Score: 1

      bullshit. Gustafson's assumes an arbitrarily large dataset. Graphics are the only arbitrarily large dataset anything deals with, and playing Duke Nukem Forever isn't going to happen on a mobile phone.

      Yeah, well... you sure about graphics? Ummm... maybe you could actually run a decent "Siri" on your phone, with no round-trip to the cloud?

      My point: there was a person who once said "640k ought to be enough for everybody". Nowadays, it's mainly quoted as a reminder of stupid assumptions one can make.

      --
      Questions raise, answers kill. Raise questions to stay alive.
    6. Re:48 Cores? by godrik · · Score: 1

      People manage to find GPGPU useful, they will find how to use 48 cores as well. I have been using prototype of Intel MIC (now Xeon Phi) for a year or so and it is quite promising for stuff GPUs do not do easily (unstructured pointer chasing parallel algorithm. Typically graphs).

    7. Re:48 Cores? by viperidaenz · · Score: 1

      "Graphics are the only arbitrarily large dataset anything deals with" - AC

    8. Re:48 cores? by m.dillon · · Score: 1

      Not only possible but very easy to do, except for the issue that nobody writes complex programs in assembly any more, and very few programmers are used to working in a high level language where the basic unit of data is 8 or 16 bits wide. So the hardware is fairly easy to construct, but getting sufficiently advanced software running on top of it is not.

      In anycase, all current many-core implementations offer a significant reduction in complexity. There is nothing inherently difficult about having a less complex 32 or 64 bit cpu... no real need to go revert all the way back to 8 or 16 bits. 80% (or more) of the real-estate on current generation cpus exists simply to enhance performance. If performance isn't needed... that is, if performance can be fanned-out across many cpu cores instead of concentrated in one, then it's easy.

      Get rid of the branch prediction, reduce the size of the L1 cache, remove the L2 and L3 caches, revert certain direct instruction implementations back to microcode, reduce the number of execution units, simplify cache coherency semantics, simplify the floating point unit, reduce the size of the TLB and simplify the MMU... that saves a lot of real-estate. Performance will be 1/4 what it was, but it could still be a win if it means you can put 10x more cores in the same die area.

      Of course, there's the other niggling little problem in that SMP / multi-threaded programming is still barely in its infancy. Most applications these days still assume one core and are only able to spool off batch tasks to other cores. Games are probably the most advanced users of multi-core technologies, with video and photo processing coming in second. But not much else.

      Cell phones at least have dedicated processes that can make use of this sort of technology. You have your cellular voice subsystem, cellular data, wifi, UI, basic I/O pipeline, application space. None of these require a lot of cpu individually so the cell phone / mobile model fits the concept very well.

      -Matt

    9. Re:48 Cores? by Anonymous Coward · · Score: 0

      Amdahl's law is not based on architecture. It is a law. And a very trite one at that. All it says is, parallel programs can only be as fast as they can be. It doesn't say how fast that is. I've paralellized at least two algorithms which common sense says can't be parallelized*. So, no, it's not a bitch, Daddy.

      * Favourite job interview response: "Why would you want to do that?" I didn't take that job.

  20. Been there, done that? by bradley13 · · Score: 4, Informative

    Haven't we already been here? When multi-core processors first became widely available, I recall a study that showed that anything over 8 cores was counterproductive. First, very few people have enough background stuff running to need more processing power than that. Second, coordinating multi-tasking on multiple cores requires a lot of complex work by the operating system, unless you just dedicate one to each process (not to each thread - that opens up problems with cache and data consistency). The benchmarks on desktop computers showed that adding more than 8 cores to a general purpose system actually slowed the system down due to added OS overhead.

    About the only way this many cores can be useful is for graphics processing (or, in TFA, video processing): many simple cores work in parallel for the same process, on different parts of the same data. This, of course, is what graphics chips already do for a living.

    --
    Enjoy life! This is not a dress rehearsal.
    1. Re:Been there, done that? by fozzy1015 · · Score: 1

      . unless you just dedicate one to each process (not to each thread - that opens up problems with cache and data consistency).

      How so? Any issues with cache consistency have to do with each core having their own L1/L2 caches but sharing the same memory. This is what hardware based cache coherence protocols like MESI were invented for and have nothing to with running multiple processes vs. multiple threads. Are you're referring to the fact that threads in the same process share the same address space? There has to be care taken to serialize access to critical sections(such as using a lock based on a mutex), and while blocking threads at critical sections can be detrimental to performance by reducing parallelism, the OS scheduler can just as effectively schedule multiple threads in one process to run on multiple cores as it can schedule a single thread in multiple processes. Multiple processes require the same sort of serialization for accessing shared memory between them. The difference between the two is how the MMU is used to configure address spaces, not scheduling.

    2. Re:Been there, done that? by PhrostyMcByte · · Score: 1

      Haven't we already been here? When multi-core processors first became widely available, I recall a study that showed that anything over 8 cores was counterproductive.

      This assumes he means equally-powerful cores.

      I can easily see having 8 high-speed cores for games and intensive processing, 4 super-low-speed ones for background tasks during idle moments with the screen turned off, and some medium-speed cores for normal use with the screen on.

      As far as efficiently using all 48 cores for one task goes -- the programming capable of doing that would be horrendously difficult or embarrassingly easy without much room in between. Amdahl's law shows that even with mythical 99% scalable code, a program would only get a pretty pathetic ~35x speedup when using 48 cores.

      The question burning in my mind is: would Android be able to have a smooth UI with such a CPU? Even with the quad-core CPU and their new tripple-buffering on my Nexus 7, Android is not nearly as smooth as an iPhone or gen-1 Windows Phone 7 devices (which use the same single-core hardware as the gen-1 Android launch devices!).

    3. Re:Been there, done that? by hattig · · Score: 1

      And this is pretty much what ARM is doing with big.LITTLE - powerful ARM Cortex A15 cores coupled with very low power ARM Cortex A7 cores - both have exactly the same ISA. Depending on software need, things either run on the A7s (most of the time probably) and then the A15s kick in when something actually stresses the A7s.

      The policy for doing so does have to take into account task power as well as instantaneous power. A slow-running job on the A7s could complete far faster on the A15s, and thus actually use less power overall. Something running constantly but not doing much should run on the A7s.

    4. Re:Been there, done that? by Anonymous Coward · · Score: 1

      There are two hurdles to parallel computing.
      Sequential algorithms, or how even in programs or algorithms that have easily divided work loads, there will still be a hard sequence of events that cannot be sped up except by a faster single CPU.

      Unneccessarily sequential code, where even though the calucluations in a loop do not depend on eachother at all, it is still written and compiled as a single thread. Purely sequential code is much easier to debug and test, inter-thread communications and thread management require more effort to test and can display more bugs that are hardware dependant. This leads to a problem that developers if pressed for time will have to either favor stability or potential speed. Even though there will be a patch in a week, the first buyers influence future sales by quite a bit.

    5. Re:Been there, done that? by Anonymous Coward · · Score: 0

      Graphics chips have a specific architecture which only functions well for embarrassingly parallel problems with highly ordered data access and very little per-instance local storage. There is a large class of problem which (a) can be parallelized but (b) cannot (as far as we know) be implemented efficiently on GPUs. Raytracing, for instance, because of the local stack. The way to think about this is that we currently have two extremes - small number of ALUs with large internal memories (CPU) and huge number of ALUs with tiny internal memories (GPU). There is a whole spectrum of processor node designs in between. I suspect future computers will have a mix of processors at various points on this spectrum.

      Having a large number of ALUs under individual OS control probably is counter-productive, yes - you would want to have a pool of them which applications can use; perhaps timeslice the pool, perhaps not, but you wouldn't be timeslicing each individual ALU because the overheads would kill you. Also, memory protection would be performed per pool, not per core, because it's too expensive putting MMUs onto every ALU. This is how GPUs already work, with over 1,000 scalar ALUs in some cases, so I wonder what that study has to say about that? :)

      Anyway, current mobile graphics chips have like 2 or 4 cores. This 48 ALU chip could well be useful as a mobile GPU.

    6. Re:Been there, done that? by strikethree · · Score: 1

      When multi-core processors first became widely available, I recall a study that showed that anything over 8 cores was counterproductive.

      Every operating system that I use has more than 8 separate processes active at any given time. Task switching involves serious overhead. Why does my 8 (16?) core computer constantly perform context switches to those stupid silly processes when I am trying to run something heavy? Seriously. If I am doing something that is eating up 100% of a single core's time, why does the process keep getting preempted to load any other processes at all?

      No modern consumer level operating system has sane policies for using multiple cores. THAT is why more than 8 cores is pointless.

      --
      "Someone needs to talk to the tree of liberty about its ghoulish drinking problem." by ohnocitizen
  21. Already here, kinda by AmiMoJo · · Score: 4, Interesting

    Modern phones already have many, many cores in them. They are just not general purpose cores available for use by the OS or applications.

    For example the radios in phones are software defined, meaning they have an RF front end that just feeds the signal to a dedicated signal processing core to do decoding. They have at least one for the mobile network radio, one for WiFi, one for Bluetooth, one for NFC, one for GPS. The audio codec will have a signal processor that can do mixing, digital filtering and parametric equalization. The cameras will each have processors to handle some image processing before the data is handed to the main CPU.

    You could combine all these cores into a single CPU, and then you would have the advantage of being able to use them for other things when they are not busy handling the aforementioned hardware. The problem is that these more general purpose cores tend to use a lot more power than dedicated ones designed specifically for one application, but presumably Intel things it will get the power budget down to something acceptable in 10 years time (or batteries will get a lot better).

    --
    const int one = 65536; (Silvermoon, Texture.cs)
    SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
    1. Re:Already here, kinda by pointyhat · · Score: 1

      Most of those cores are DSPs. x86 variants are crap for DSP related operations. TBH x86 is crappy for pretty much everything.

      The way we should be going is reconfigurable logic. For example when an mp3 is played, the device is reconfigured to contain a hardware codec. This can work on an async clock so it will only tick on data availability. When it's not being played, it turns of the macrocells that it was built on. There should be analogue and digital macrocells which can take on the RF and computational duties respectively.

      The problem is that this is hard.

    2. Re:Already here, kinda by Anonymous Coward · · Score: 0

      I work with FPGAs. What you describe sounds like it would have to be a $3000 phone, even ten years from now.

    3. Re:Already here, kinda by viperidaenz · · Score: 1

      My PC has many many cores too. There are 4 in the CPU, 1440 in the GPU, The network card has one, the hard drives each have one, the dvd writer has one, the power supply has one, there are probably half a dozen on the motherboard doing various things as well... the keyboard has one, the mouse has one, the wireless keyboard/mouse dongle has one, the memory card reader has one, the monitor probably has a few.

      They're all optimised to their own set of tasks. Some of them are probably only 8 bit. There is no reason, whatsoever, to put all of those processors inside a single chip. For a start, you'd have to throw away your LGA2011 socket, because 2011 pins wouldn't be enough. Welcome the LGA16384! You'd also have a pretty shit time routing the high speed signals from the various peripherals. From hard drive head signals to RF from the wireless, wcdma, gps and nfc. You've also got physics to deal with. Electricity has a finite speed - parallel buses and differential pairs need to have each conductor the same length and impedance. The closer your traces on the PCB, the lower the impedance you need to prevent interference. That means more current needs to be sent down the line.

      Wouldn't it be great if you could have simple, lower power, cheap processors at each peripheral to lower the bandwidth requirements, be able to respond to events as they happen without waiting for a shared bus or other tasks to complete and when you don't need them, you can switch them off.

      Oh wait. that's what is already done.

    4. Re:Already here, kinda by viperidaenz · · Score: 1

      If you've got the money you can spend $3,000,000 on a phone http://most-expensive.net/cell-phone-mobile

    5. Re:Already here, kinda by Anonymous Coward · · Score: 0

      The problem is not that this is hard. The only reason it's not a solved problem already is because this design has two show-stopping problems:

      The first problem is that it's very expensive. A single "gate" on an FPGA costs 10x-100x what a single "gate" on an ASIC costs. That's because a "gate" on an FPGA is actually 10-100 actual gates. For the same reason, an FPGA takes 10x-100x more power than the ASIC.

      The second problem is you don't want that kind of design. You want to play an MP3 while you play your GPS-based game over WiFi, not have to shut down some operations while you perform certain tasks.

  22. cores are the new Ghz by banbeans · · Score: 2

    Want to impress the pointy heads talk about more cores.
    Throwing more cores is not the solution to all problems and actually opens up a whole bunch of new ones.
    I wish that more attention would be spent to solve real problems rather than trying to be buzzword compliant.

    1. Re:cores are the new Ghz by Anonymous Coward · · Score: 0

      My chip has more Gigahertz than yours has cores!

    2. Re:cores are the new Ghz by Anonymous Coward · · Score: 1

      rather than trying to be buzzword compliant.

      Our hyper-core paradigm will reengineer the competative landscape, allowing us to leverage our time-to-market and maximize ROI.

    3. Re:cores are the new Ghz by viperidaenz · · Score: 1

      My chip has 8 cores, does yours have 8GHz?

  23. More cores != more power & efficiency unless.. by DontBlameCanada · · Score: 1

    The O/S supports the additional resources. Who has am O/S today that has a proven roadmap to get there?

  24. Ad-Driven Information by Anonymous Coward · · Score: 0

    Am I the only one who doesn't find "ad-driven information" to be a compelling reason for me to purchase a 48-core cell phone?

  25. not with today's coding methodology by Speare · · Score: 4, Interesting

    Until you revise the whole way people write software, adding cores is useful to a very limited point. Today's software can be split at one core per thread, or one core per process. If you try to get two cores to work on the same thread, you just increase serial contention, not decrease it.

    Even thread-happy Java is only working on maybe 3-5 threads at a time, the rest are sleeping until a device wakes it, or until a certain time has elapsed. A new compiler may be able to help a little bit, but it's just going to be creating very short-lived micro-threads when it detects those few opportunities for them.

    Graphics hardware is great for many parallel cores, because it's the same tight problem with different data, endlessly repeated. Multiply these 4x4s please. Fill these pixels please. Endlessly. Same goes for encryption, and maybe a few bits of video game AI logic. Not many other software naturally fits to using many cores.

    --
    [ .sig file not found ]
    1. Re:not with today's coding methodology by viperidaenz · · Score: 1

      Run these database queries, from 5,000 concurrent sessions...Join A and B and while you're doing that join C and D, when you're done, join the result and sort it. do that 5000 times on different data...

      Route those packets, inspect the content, filter them....

      Respond to this HTTP request....and these other ones from these 10,000 users.

      No... nothing much but graphics naturally fits to using many cores...

    2. Re:not with today's coding methodology by Anonymous Coward · · Score: 0

      Yes, parallel systems are hard to approach today, because we write software assuming serial systems. Of course that has to change. Take that as read and move on to figuring out how it has to change. Parallel systems are not going to get off your lawn. Adapt, or die.

      And please expand your conception of the kind of problems that are parallelizable. Believe it or not, there are more than three. And, there is more to parallel system design than GPUs. They are an example of a parallel system, not a definition of one.

      Or, prove you're right. Join Google and serialize their million-dollar clusters all into one box, since apparently none of the things they do naturally fit to using many cores. You'd save them a ton of money. Same goes for all the supercomputers in the top 100 list. Stupid waste of money, right?

  26. What about battery power? by who_stole_my_kidneys · · Score: 1

    unless there is some breakthrough in battery life in the next 5-10 years, having a 48 core cpu run on a mobile device for 5 minutes before depleting is worthless.

    1. Re:What about battery power? by Capt.DrumkenBum · · Score: 1

      Who says it has to run all 48 cores all the time?
      Lets assume this does become your primary computing device. It could run 2 or 4 cores on battery, rnen when you dock it and need some real power, it ramps up to all 48 cores.
      That is just off the top of my head. I am sure there are many other viable options. Also:
      Batteries will probably get better.
      Power needed by each core will probably reduce.

      --
      If I were God, wouldn't I protect my churches from acts of me?
  27. Hammer in Search of Nails by Revotron · · Score: 1

    Why does a cell phone need 48 processor cores? How about we force all the shitty mobile app developers to make shit work on one core before we throw 47 more at them to abuse with their terrible resource management skills.

  28. Why does everyone sill refer to it as a phone? by Anonymous Coward · · Score: 0

    When 99% of its use are for things that are not associated with phones. People are starting to sound out of touch and archaic. Your phone will not also be your computer rather your computer will also be your phone. It's not 1999 anymore.

  29. Nice heat sink! by Anonymous Coward · · Score: 0

    Nice heat sink http://www.computerworld.com/common/images/site/features/2012/10/chip_508.jpg! Too bad the cell phone will no longer fit into my pocket...

  30. Hope they'll have a bio-neural interface by CrowdedBrainzzzsand9 · · Score: 1

    Hope they'll have a bio-neural interface. I'll need one really bad by then.

  31. Voice and video analysis by crow · · Score: 4, Interesting

    Current phones do just fine with HD video and multitasking. Sure, some may glitch occasionally, but more due to software design than lack of CPU. This will do little to nothing for the things we use our phones for today.

    What it will enable is new classes of features, such as real-time video and voice processing. With that sort of CPU power, you can do voice recognition without sending the audio over the network for analysis. Who knows what people will think of doing for video analysis?

    Though for the most part, the added cores will be powered down, doing nothing but putting a good bragging number on the spec sheet.

  32. projected uses by Anonymous Coward · · Score: 5, Insightful

    Core 1-12 : DRM
      - these cores will check all audio/video/ebook files for copyright infringement

    Core 13-24 : TPM
      - these cores will implement TPM and secure the DRM portion

    Core 25-37 : Genuine Advantage Checking
      - these coes will check that the system state is valid, and all license keys are valid and updated

    Core 38-40 : Virus Checking
      - these cores will implement malware checks and virus checks

    Core 41-47 : OS and Sandboxing/Security
    - these cores will run the base os, and run all applications in sandbox mode

    Core 48 : User Application
    - this core will be available for running user applications in the performance reduced sandbox mode. Priority is given to cores 1-47, in order of decreasing priority.

    1. Re:projected uses by Anonymous Coward · · Score: 0

      Er .. I'm pretty sure this is meant to be a joke Joyce. Someone mod funny.

  33. We have that power now. by Anonymous Coward · · Score: 0

    If you can encode video or encrypt an email (encryption/decryption typically isn't that computationally tough), on a current smartphone, why wouldn't the most recent 2 or 4 core >1GHz CPU phones be able to do both at the same time?

    1. Re:We have that power now. by viperidaenz · · Score: 1

      Cellphones have dedicated video encode/decode hardware. If there is only one piece of hardware, they're only good at doing one at a time (although picture-in-picture never really took off on TV... I doubt it will be seen on cell phones.). When the codecs change, the hardware acceleration becomes useless too. The software decoder on my phone chews through battery compared to the hardware one, but the hardware decoder can't decode some video files.

  34. My phone != my computer by Todd+Knarr · · Score: 3, Insightful

    My phone won't become my computer because it's not about the CPU power. Hasn't been for ages now. The average phone already has more CPU power than the average desktop user makes use of. It's more about:

    • Memory. My computer has something like 16x as much RAM as my phone, which means it can run a lot more stuff simultaneously. It can run all those system-tray programs, without breaking a sweat. But RAM means power, and putting 8GB of RAM into a phone increases the power draw (modern memory requires not just constant power but constant refresh access) and decreases the battery life.
    • Size. My computer has 2 27" monitors on it, making for a lot more screen real-estate to work on. And I need it when I'm simultaneously writing a document, referring to a spreadsheet, several e-mail messages and a couple of Web sites, keeping track of a couple of IM conversations, and let's not forget Visual Studio with a dozen files open in it. You can't have that kind of screen space on a phone, you're physically limited by the size that can be conveniently held in one hand.
    • Peripherals. I mentioned monitors. There's also my good Model M keyboard (you don't appreciate a good keyboard unless you're working in a job where you're typing nearly constantly for hours on end), the high-end mouse, the small input tablet, the good 5.1 speaker set for the music, the scanner for getting paper documents into electronic form... You can't attach all that to a phone in any useful way.
    • Portability. Yes you can solve all the shortcomings of a phone with a dock and attached peripherals, but why? By the time you're done, you've removed the things you wanted from a phone: the ability to carry it in one hand, and to not have it tied down with wires. We're seeing with Win8 what the downsides are of trying to design a system for both phone/tablet and desktop uses, and you end up not being satisfactory for either so the idea of grabbing the phone out of the dock and going ends up hamstrung by that.
    • Price. One reason desktops are cheap is that they can use commodity parts and have enough open space in the case that they don't have to worry about power so much. Phones are almost 100% custom-built with a lot of work going in to designing parts that can be packed into that small a package with no ventilation at all in 100-degree ambient heat and not incinerate themselves. They're going to inevitably be more expensive than a desktop just because of that. And while Microsoft may be willing to hemorrhage money on hardware because it supports other revenue streams, phone manufacturers can't.

    So while more cores may help phones do phone-like things better, especially combined with nifty ideas like Google Glass, but it's not going to help the phone replace my desktop.

    1. Re:My phone != my computer by oic0 · · Score: 1

      As an owner of a phone with 2GB of RAM, I think the RAM thing is covered.

    2. Re:My phone != my computer by aicrules · · Score: 1

      I really don't mind that they keep trying to add different new and improved widgets to every gadget we have. They clearly don't have a foolproof way to truly understand how something would be useful to an end user without stumbling on it anyway. I see some of the things they do in SciFi (minority report for example) as someone trying to take a stab at what people will want to do/benefit from in the not to distant future. While some of it ends up being mimicked in reality, that often results in the realization that while it looked cool, no one finds it useful and it dies out. However, I am fine with them doing this. The creative process for making ground breaking advancements is often trial and error, and a lot of times the advancements come in a form different than the original idea intended. If they want to push the envelop with this vague notion of "being able to do all the things we couldn't do now due to low processing power" then by all means...go go go!

    3. Re:My phone != my computer by Anonymous Coward · · Score: 1

      As an owner of a computer with 12GB of RAM, I think your 2GB RAM phone is seriously lacking if it were competing for my desktop.

    4. Re:My phone != my computer by hattig · · Score: 1

      Today's 2GB mobile phone will be 2014's 4GB mobile phone, and 2016's 8GB mobile phone (using TSV & WideIO for ultra-low-power connectivity to the SoC).

      Of course today's home computer is 16GB, and probably 64GB in 2016... and I'm not going to claim that it won't be needed because somehow computers continue to make use of all the RAM you ram inside the box. 8K monitors and 4K video will probably use a load of that memory just for OS imagery...

    5. Re:My phone != my computer by istartedi · · Score: 1

      My phone won't be my computer either, and the ergonomic issues you touched on dominate that.

      I can see us shoehorning all the memory, speed, and price into a tiny package some day. By definition though, they can never give it a full-sized keyboard and monitor without turning it into something other than a phone.

      The computer, memory, storage aspect of it will get so small that you can just integrate it into the keyboard (like C64s were!) or the monitor (like iMacs). You won't be able to tell the computer from the keyboard or monitor, except that it will have a few more ports on it and a slightly higher price tag. We might even see "has a computer" as a check-off feature for the monitor or keyboard. Millions of them will be hooked up to the preferred computer and go un-used, or just be there as a backup system in case the main computer starts acting flaky (yeah, I'm running on my monitor today. The keyboard crapped out)

      --
      For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?
    6. Re:My phone != my computer by gman003 · · Score: 1

      As an owner of a computer with 12GB of RAM (plus 2GB VRAM, but that doesn't count), I think 2GB is more than enough for a phone.

      As I type this, I have Firefox (two tabs), Chrome (nine tabs), Komodo (two open files), PuTTY, MySQL Workbench (two connections, fifteen tabs total), Google Talk, Windows Media Player, Notepad++ (six open files), Windows Explorer, and LibreOffice Writer open, plus Steam and a desktop background image cycler running in the background (probably plus some stupid DRM shit). This is covering three screens (two 1080p plus one smaller one). And this is on Windows 7, not exactly the most memory-lean OS there is.

      I am using precisely 4.40 GB of memory. Plus a bit more for the OS to cache things, since it's there. Let's liberally round that up to 6GB.

      So with merely three times the memory in his phone, I can run a *very* heavy desktop load without any memory issues. And if I close Firefox, my entire load can fit inside 4GB, so maybe twice the memory in his phone is more accurate. Since I've yet to want to run more than two programs at once on my phone, I can't exactly see 2GB being excessive for a phone.

    7. Re:My phone != my computer by Todd+Knarr · · Score: 1

      I'm not sure about integrated into the keyboard/monitor. I know my experience is that most keyboards (short of the type that cost $100+ for just the keyboard, no gaming features or LCDs, eg. Unicomp or WASD) have a working lifespan of less than a year. By that point either the keys are worn to the point they're illegible, or spills and gunk have gotten the keys gummed up to the point they won't work reliably. So I go to Fry's and buy another keyboard and plug it in and I'm good to go again. If the computer's integrated into the keyboard, replacing the keyboard means replacing the entire computer and then reinstalling everything from scratch. Not fun. I think the conventional desktop will remain a box for the chassis with separate keyboard, mouse, monitor etc. so we can replace the cheap subject-to-wear parts without having to replace the actual computer at the same time.

    8. Re:My phone != my computer by Todd+Knarr · · Score: 1

      Run the Resource Monitor and watch it, because the in-use numbers are misleading. On the 2GB machine I deal with at work, opening Visual Studio, the Oracle DB client and a couple of terminal and shell windows only shows me using less than 1.5GB of memory, but I'm actually using a lot more. I can tell because every time I do anything or switch windows the paging activity shoots through the roof. That means that that 1.5GB in-use is not sufficient for the workload and memory's being paged out and in to satisfy current demands. That's a bad thing because on a phone there isn't anywhere to page out to. Phones don't have hard drives (and the SD card is way too slow for paging space).

    9. Re:My phone != my computer by viperidaenz · · Score: 1

      the good 5.1 speaker set for the music

      For your 2 channel music?

    10. Re:My phone != my computer by Todd+Knarr · · Score: 1

      Some of it, yes. The stuff I listen to a lot is all encoded 6-channel though. But even the plain stereo stuff goes through the normal audio processing. Bass goes to the subwoofer, mid-range and treble go to the front and rear pairs (so I get spatial fill, it's not just coming from in front of me). There's a major, major difference between what comes out of the 5.1 set vs. the plain no-subwoofer stereo speakers.

      A really good set would have 2 channels per front/rear speaker, with treble split from the mid-range and each routed to a driver tuned for that frequency range. But for ordinary listening that's overkill, the treble response of the regular drivers is good enough.

    11. Re:My phone != my computer by narcc · · Score: 1

      I remember walking home one night with 32mb of RAM in my pocket (4 8mb SIMMs). It was more RAM than I ever thought a computer would ever need. Heck, you couldn't buy a computer with 32mb of RAM -- 16 was a pretty high-end, iirc, with 8 being average.

      2gb wasn't even an amount of drive space -- it was science fiction. I had quite the monster: 512mb -- no way I could ever fill that.

      What did I use it for? Hitting the local BBSs, word-processing, spreadsheets, software development, and games. Replace "BBS" with "Internet" and I'm not really doing much that's new or different now than I was then.

      Maybe we need to start thinking about performance again. Doing more with less.

    12. Re:My phone != my computer by viperidaenz · · Score: 1
      So you're effectively using your 5.1 setup as a 2.1, by sending the same signal to front and rear.

      A really good set would have 2 channels per front/rear speaker, with treble split from the mid-range and each routed to a driver tuned for that frequency range. But for ordinary listening that's overkill, the treble response of the regular drivers is good enough.

      You mean a crossover, not extra audio channels?

    13. Re:My phone != my computer by istartedi · · Score: 1

      Yeah, good point. If we can get rid of the hard drive, then there are no moving parts except butttons which do tend to wear out quickly. Maybe the computer could be a module that slides into the computer, the way my CD drive slides into my laptop. In theory, I could replace the CD with another disk of some kind; but I've never had the need.

      --
      For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?
  35. Speed? by CCarrot · · Score: 2

    Sure, but how *fast* would these micro-cores run? Will we wind up with a single core not being able to handle anything on it's own?

    Plus, adding more cores for shared tasks increases the management overhead. It's like a project: more resources can help get the job done faster...if they're properly co-ordinated, communicating properly, and everyone isn't waiting for one particular resource to finish it's task before the others can proceed. All of this takes significant overhead time IRL, I can't imagine it would be much different for flocks of cores...

    I can see this allowing more simultaneous processes to run without bumping into one another, but basic speed for each process might go way, way down.

    --
    "I love animals! Some are cute, others are tasty, what's not to like?" - Betsy Schroeder, Jeopardy contestant
    1. Re:Speed? by arctus · · Score: 1

      This is kinda what I was thinking. I don't know enough about low level hardware interaction to answer this, but I would think its logical to assume trade-offs.

      I mean, could we really increase the number of cores, increase performance, and lower battery life all at once (lets assume battery technology stays the same for ten years and remains constant even though it wont)?

      I was thinking adding cores would do little in raw performance other than making the phone more adaptable to different use cases (multi-processing certain apps/services) and better at doing nothing more efficiently (idle). But like I said, not a hardware guy. I would like to know how a chip with 48 cores compares to a chip with say 16 cores in terms of benchmarks. Surely with more cores each core is less powerful right?

    2. Re:Speed? by CCarrot · · Score: 1

      I would like to know how a chip with 48 cores compares to a chip with say 16 cores in terms of benchmarks. Surely with more cores each core is less powerful right?

      I would think it would have to, unless we radically improve thermal dissipation technology first...or unless we find a way to break the first law of thermodynamics.

      Hmmm...now that would be interesting, what if we developed a way to allow the heat generated by the core to help power the processor? Something like wrapping them in these nano-antennae solar panels, so the infrared wavelengths of generated heat directly produces electrical charge -> battery -> processor? Very interesting...and I could see it being possible, in 5-10 years. Sure. We're pretty bright little monkeys, and we do love our toys :)

      --
      "I love animals! Some are cute, others are tasty, what's not to like?" - Betsy Schroeder, Jeopardy contestant
    3. Re:Speed? by viperidaenz · · Score: 1

      Why can't a single core not be sped up when the others can't used? Intel already do that in their other processors based on workload and thermal envelope.
      eg: 48 cores at 200MHz or 4 cores at 2GHz?

    4. Re:Speed? by CCarrot · · Score: 1

      Why can't a single core not be sped up when the others can't used? Intel already do that in their other processors based on workload and thermal envelope.

      eg: 48 cores at 200MHz or 4 cores at 2GHz?

      That's an interesting idea. Unfortunately, it looks like that technology currently only allows for about a 40% boost in speed per core (so from 200MHz to 280MHz, not 2GHz)

      Now, I'm not sure if that's a hardware limitation or a thermal one: it may be that they *could* do a much wider variation in processor speeds if there was a good reason to do so. But again, overhead. Tracking projected thermal profiles (so as not to accidentally fry the customer's hand...or the device), monitoring and continually adjusting performance based on load for all of the cores will inevitably cut into the overall efficiency of the combined unit.

      --
      "I love animals! Some are cute, others are tasty, what's not to like?" - Betsy Schroeder, Jeopardy contestant
    5. Re:Speed? by Anonymous Coward · · Score: 0

      Two cores at the same speed produce about the same amount of head as one core at that speed.

    6. Re:Speed? by viperidaenz · · Score: 1

      I would say that 40% increase in Ivy Bridge is based on the maximum speed of the CPU. If a 2GHz Ivy Bridge is clocked at 200MHz to save power, I don't see why it can't still get up to 2.8GHz.

      My PC at home sits around 1.1GHz at idle and turbo boost brings it to 3.6GHz.

      The upper limit seems like it is more a function of absolute speed (the technology producing the transistors and their voltage) and total power dissipation.

      Intel could probably set their i5/7 turbo boost frequencies to 6 or 7 GHz, parts of the old P4 architecture operated at twice the core frequency. There is no point though, if that means you blow the temperature limit in milliseconds or the socket/motherboard can't supply enough current.

  36. It's called "GPU" by Anonymous Coward · · Score: 0

    GPUs already have many cores and already deal with the problems which result from massively parallel computing. The Intel guys just need to get over themselves and buy NVidia and not have us wait 10 years for their own stuff.

  37. ownership by vlm · · Score: 3, Insightful

    The phone would be smart enough to not just be a computer but it could be my computer.

    As trends continue it will not be "my computer" it'll remain my service provider's computer which they graciously let me use a small fraction of its capabilities for a monthly fee.

    --
    "Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
  38. One thing for a phone-sized device by ibsteve2u · · Score: 1

    While on the surface it may seem that "Big Brother" (or the competition, or China - assuming that they're not one and the same) is less likely to have his nose in your PC given that the intent of a phone is to maintain external communications links, one would do well to remember that those teeny commo chips that fit onto a phone's dinky mobo would fit onto a massive PC mobo, too. Which is something that has probably occurred to anybody who remembers the Tandy 1000 TX - and so remembers anonymous custom chips.

    Which brings me to my point: It is a lot easier to build a Faraday cage for a phone-sized device to make sure that it truly is firewalled off...so a phone-sized PC with true computational power - assuming sufficient storage (and, ideally, a means of accessing external dumb storage) - is progress.

    (Yeah, I know: Paranoia, extraordinaire. lollll...but if you'd had the jobs I've had...)

    --
    Orwell: "In a Time of Universal Deceit, telling the Truth is a Revolutionary Act"
  39. Multiple uses, great! by Anonymous Coward · · Score: 0

    While moving around you use only one or two cores at ultra low power, but when you reach your desk you plug into that little connector or just put your phone on that special pad and you can do real work. Neat.

  40. Dedicated 64-bit. by Anonymous Coward · · Score: 0

    I did design a 64-bit processor and it's more cleaner than those Intel's things.

    It's designed for main programming languages where their speed is to extreme: C, Fortran,C++, Pascal, Modula-2, Java, OCaml, GHC, etc.

    Its model is dedicated to beowulf clusters of SMP computers.

    jcpm: i don't offer more details of my design, end.

    1. Re:Dedicated 64-bit. by Anonymous Coward · · Score: 0

      The difference between your clean, "speed to the extreme" 64-bit architecture and those slow, power-hogging Intel architecture chips is that Intel's end up actually being produced and sold.

  41. Ten years, really?! by cavok · · Score: 1

    Adapteva looks quite ready now.

  42. So when you lose your personal computer? by oic0 · · Score: 1

    Being able to carry your entire computer in your pocket and use it all day long is great, until you lose it lol. Unless of course this whole cloud thing actually catches on and carriers stop raping consumers on data prices.

  43. Really Intel? by TheSkepticalOptimist · · Score: 2

    I'm still waiting for the 80-core PC.

    How about realizing one of your predictions before creating another. You never achieved an 80-core PC and barely made a dent into the mobile CPU market so don't bother predicting anything for the mobile market until you are player.

    Also Intel, its time to change your architecture. Packing in billions of transistors into a core to support legacy instructions is not an efficient design anymore. I don't want 4, 48 or 80 cores with defunct MMX and obsolete x86 instructions. Stop making CPU's where every core contains the entire history of Intel.

    --
    I haven't thought of anything clever to put here, but then again most of you haven't either.
    1. Re:Really Intel? by Anonymous Coward · · Score: 0

      One BILLION-core where each core is justly one transistor (a mosfet).

      Are you saying me that Intel can't it?

    2. Re:Really Intel? by TeknoHog · · Score: 1

      I'd probably be happy with just *pinky -> mouth* One Million Cores.

      --
      Escher was the first MC and Giger invented the HR department.
  44. Re:More cores != more power & efficiency unles by Cyko_01 · · Score: 1

    My money is on linux/android

  45. intel-lectual propaganda by Funk_dat69 · · Score: 2

    Intel like to throw claims like this out there to try and win mind-share.
    It means nothing, but sounds impressive in a vague, buzz-wordy way. It's just marketing.

    I'm not surprised their vague future predictions are aimed at mobile now. They desperately need mind-share in that segment.

    "Look! We're relavent in mobile! We'll have FORTY-EIGHT cores! All with frickin' laser beams attached to their heads!"

    --
    FUNK!
    1. Re:intel-lectual propaganda by WilliamBaughman · · Score: 1

      Intel spends a lot of money and manpower on good will, advertising, and mind share, but sometimes their strange and ridiculous announcements (Intel Pledges 80 Core Processor in 5 Years - six years ago) turn into products (Intel To Ship Xeon Phi For "Exascale" Computing This Year) that actually get used (TACC "Stampede" Supercomputer To Go Live In January). It may just be because I was so skeptical when I first heard of the processor that eventually became "Knights", but I'm hesitant to shout "that's impossible" when they come out with these announcements now.

  46. yesterday's arguement by Frontier+Owner · · Score: 0

    Yesterday, we discussed how the M$ surface couldn't do everything as well as dedicated devices. why would more cores change that argument? You are now sucking down more power. bigger or higher capacity batteries. screen sized tradeoffs, input device tradeoffs and to get around these you get the same junk to carry around now except you run you phone dead surfing the web when you really need your GPS to get you to see the client or the PHONE to tell him you're going to be late because your GPS is dead.

    1. Re:yesterday's arguement by m.dillon · · Score: 1

      The assumption that one is sucking down more power with more cores is not correct. The point is to suck down less power. Using simplified cores allows larger chip processes to be used (no need to shrink things down to 20nm), and you run them at a far lower frequency, resulting in far less leakage (in two dimensions). Having many independent cores also allows the chip to run at a far lower frequency and to actually shut the power off on a core-by-core basis based on load. Not just stop the clocks... actually kill the power in the core's circuitry. It's actually an easier solution than trying to implement low-power modes or frequency shifting for unloaded subsystems. If the frequency can be brought down enough on a large-feature chip one doesn't even need to shut the power off, leakage will be low enough that one can simply gate the clock and achieve almost the same efficiency.

      The power/performance tradeoff is definitely a win when you move to more, slower cores. That's been shown over and over again over the past decade with the opteron/xeon wars. The only potential lose occurs when you reduce the size of the on-chip caches since smaller caches will result in more power consumption accessing the dynamic ram. So tuning on-chip caches becomes very important. But that's about the only issue.

      -Matt

  47. I think as in Star Trek computer by SmallFurryCreature · · Score: 2

    Currently your PC is a rather dumb tool, a useful tool but dumb. The fantasy is that stuff like Siri makes computers actually useful other then for finding cat pictures. Because outside work related coding and some gaming, I don't use my PC for much that other devices couldn't also do. Yes, I watch movies on it but I can watch movies on a projector if they still sell them. I can play music on a wax cylinder.

    Yes, there are some small advantages but not nearly as many as you would like.

    Play romantic music computer... no not the safety dance... yes dancing is romantic but not the safety dance.

    Really, try to get a computer to do something as simple as understand mood and play the right music. That is what a MY computer could do, I think. Be more then a tool but an actual help. Not just enable me to with effort get things done (make manual playlist with the right mood music, then select play music when I am in the mood) but actually do stuff for me.

    But I don't see how more cores could help. Big Blue got cores coming out its nose and it can't make a playlist based on my mood anymore then a C64 could.

    --

    MMO Quests are like orgasms:

    You may solo them, I prefer them in a group.

  48. Divide and Conquer? by Anonymous Coward · · Score: 0

    So instead of say 1 core, running at 1.2Ghz, will we have 48 cores, each running at 25Mhz?

    1. Re:Divide and Conquer? by Anonymous Coward · · Score: 0

      48 x Apple II = luxury operating system inside of a smartphone.

  49. Deja vu? by Grizzley9 · · Score: 1

    I'm pretty sure that if you told someone 10 years ago we'd have quad core processors running at over 1GHz with loads of graphics, ram, etc on a smart mobile phone, they'd think why can't you use it as your main computer? In 5-10 years by the time this 48 core would be feasible, desktop computing will have grown/changed as well. I'm not sure our mobile devices will ever be our main computers. A link to them perhaps, but its far easier to offload the work to a remote station than do it all onboard.

    1. Re:Deja vu? by Frontier+Owner · · Score: 0

      or we can just use our mobiles to access the cloud based apps and offload the processor work to them. Kinda like the old mainframe / terminal topology. You guys remember studying that in history right? And given how AT&T runs, you can do this once a month before you're over your limit on data and being charged another $30.

    2. Re:Deja vu? by omnichad · · Score: 1

      With even the cheap desktops coming with mobile CPU's now, we're just seeing a widening gap between power users and the mainstream computer owner. I upgraded my wife's computer to a wimpy Sempron 145 and she couldn't be happier with finally having 8GB of RAM (not to mention it being DDR3) and not really needing more CPU power. While I on the other hand am upgrading to an Ivy Bridge i7 this week and am not totally sure if I went high enough.

  50. I can see what they're thinking by gman003 · · Score: 1

    The issue they're trying to solve isn't "how do we make more powerful phones?", it's "how can we lower power usage without sacrificing power?".

    If you try to improve performance the same way desktops did (higher clocks, great gobs of cache, and more execution units inside the core), you'd raise performance, but you'd also massively increase your power draw without giving you an easy way to lower it when it's not in use. Underclocking only gets you so far - ideally, you'd need to be able to shut down execution units, or even some of your cache.

    With this number of cores, you can shut down entire cores while you idle, which is much simpler to manage. Sitting in your pocket, this thing would only be running on a core or two.

    I can see what Intel is thinking. It's a solid hardware idea, but it requires a massive shift in software that doesn't seem likely. Much like Itanium, in retrospect.

  51. Are they sure it's not 5 to ten years? by Anonymous Coward · · Score: 0

    Is it just me or is anyone else annoyed with quick, poor quality articles stating the obvious?

  52. Re:More cores != more power & efficiency unles by Anonymous Coward · · Score: 0

    What about QNX? They seem to have been doing multicore stuff for longer than anyone.

  53. The CPU is not the bottleneck anymore. by EmagGeek · · Score: 1

    9 women can't have a baby in a month. All the data being processed by those 48 cores still has to pass through the same vagina to get to RAM or Flash.

    Until each core in a multicore platform has its own private RAM interface, to its own private RAM, the point of diminishing returns will remain at some number of cores that is much smaller than 48.

    1. Re:The CPU is not the bottleneck anymore. by Anonymous Coward · · Score: 0

      (Or Vagina)

    2. Re:The CPU is not the bottleneck anymore. by fatphil · · Score: 1

      > 9 women can't have a baby in a month. All the data being processed by those 48 cores still has to pass through the same vagina to get to RAM or Flash.

      The biggest problem is that the programmers are dicks.

      --
      Also FatPhil on SoylentNews, id 863
  54. How about 64 cores now? by Anonymous Coward · · Score: 0

    Really 48 cores in five to ten years is news? I can get a processor with 64 cores right now. http://www.adapteva.com/introduction/

    1. Re:How about 64 cores now? by Anonymous Coward · · Score: 0

      Considering that AMD makes a 2048-core processor, 64 is hardly anything to get worked up over.

  55. Wut? by GigaBurglar · · Score: 1

    Interesting choice of words.. I still don't know what I'm buying - but it won't be for another ten years.. that's for sure.

  56. Get a Defy by Anonymous Coward · · Score: 0

    Seriously, get a Motorola Defy.

  57. Not powerful enough... by Anonymous Coward · · Score: 0

    until augmented reality can do this fluidly, not as a picture, but real time.
    http://4.bp.blogspot.com/-h3zgVdzcHqI/UIuobyqljkI/AAAAAAAAAmU/aPkT58aQwds/s640/Robotics;Notes+01.jpg

  58. Two issues by CBravo · · Score: 2

    There are two issues with all this multi-core speak we've been hearing:

    There is hardly any code, other than a few optimized libraries, that use multi core processors. They try to make it sound unimportant but it is the largest hurdle. There is a reason people don't think of a 3930k as being 50% faster than a 3820! Other than some failed libraries like OpenMP; there are no valid programming models to use either multithreading and/or networking transparantly (since it should not functionally make any difference). Ergo: The developers are still making their single-thread code.

    The advantage of multi core processors can be that you can lower you clockspeeds and/or memory speeds on secondary tasks. Hardly any research is done on that (I have myself but let's call that an 'academic' proof of concept).

    --
    nosig today
    1. Re:Two issues by Anonymous Coward · · Score: 0

      Don't you mean 3770(k)? The 3820 also has six cores...

    2. Re:Two issues by CBravo · · Score: 1
      --
      nosig today
  59. Ha, what about battery life? by Rexdude · · Score: 2

    Reminds me of an old Soviet Russia joke(more here):

    A Polish tourist comes back home after visiting the USSR. He carries two very large and heavy suitcases. On his wrist is a new Soviet-made watch. He tells the customs man: "This is a new Soviet watch. It's a wonder unknown in the capitalist countries. You see, it shows time, the rate of your pulse beats, the phases of the Moon, the weather in Warsaw, Moscow, and New York, and more and more!"

    "Yes, it's a wonder," the customs man agrees. "And what is it you have in these big suitcases?"

    "Oh, it's just the batteries for that watch."

    --
    "..One hosts to look them up, one DNS to find them, and in the darkness BIND them."
  60. Another way to pay for the Marketeers BMW X5 by Anonymous Coward · · Score: 0

    Great, now we have yet another way of producing mountains of wasted precious earth resources. I don't want my 50 core smartphone any more as there is a 100 core device being released next month, and the month after that there will be a 200 core device.

    Where does it end?
    Have we not screwed our planet up enough yet?

    Of course in addition to upgrading the smart phone we'll feel compelled to upgrade the desktop, get a pad that we don't have a use for, upgrade the TV set etc etc.....

    I'm all for scientific advancement but this is going to turn into yet another greedy, selfish marketing exercise.

  61. 48 cores? by sgunhouse · · Score: 1

    I always wondered why someone doesn't make a single chip using lots of cores that are along the lines of the early 8-bit processors. I mean, given the transistor count on modern chips, how many 8080/6502/1802 processors with their own memory could be put on a single chip? Okay, you'd need some sort of controller and so on - I'm sure even a 48-core (or 12 core) chip has some dedicated circuitry so the cores can communicate. It doesn't absolutely have to be 8-bit of course, but we did get work done on the old 8-bit machines. So, 1000 cores anyone?

  62. Time for "fibers" by Anonymous Coward · · Score: 0

    Beyond threads as the atomic unit of execution (fibers = threads of threads) plus strong use of vectorization will improve future computing 100 fold in 10 years time. The hardware is here. Time for compilers to catch up.

  63. whole new world of possibilities by nurb432 · · Score: 2

    Ya, with nanobattery life these things will have, people might actually go out and expirence the real world again, while their phone/tablet charges.

    --
    ---- Booth was a patriot ----
  64. 80 Cores in 5 years? by owlman17 · · Score: 2

    Speaking of 5 years ago, we're still waiting for this to go commercial.

  65. How the world moves and turns by stifler9999 · · Score: 1

    Once we had dumb terminals on our desks, and we connected to a mainframe in a floor of a building.
    Then we moved to having all the power on our desks, got rid of the mainframes.
    Now the mainframe will be in your pocket, and you will use a dumb terminal on any desk to interact with it!

  66. CONSPIRACY TO KEEP THE BROWN MAN DOWN!!! by Anonymous Coward · · Score: 0

    I cannot believe, as a non East Indian, you racists fuckers choose to ignore the achievements of http://en.wikipedia.org/wiki/Tilera and its hard working Indian immigrants founders.
    This fabless vaporware manufacturer is generations ahead of Intel (on paper), and has 100-core processors (on paper) ready to impress any clueless idiot.

  67. What a load of crap by Anonymous Coward · · Score: 0

    Really? You need 48 cores to encrypt an email while doing something else? I think you're doing something wrong....

  68. Really? Mobile? by highphilosopher · · Score: 1

    Is that 48 cores in your pocket, or are you jsut happy to see me?

  69. Why? by Anonymous Coward · · Score: 1

    48-Core Chips Could Redefine Mobile Devices

    Why? its a friggin phone. Do I really need 48 cores to get a phone call?

    How about 48 cores in my workstation. Now there I could get some use out of 48 cores.

  70. 48-core Wheeee by agrisea · · Score: 1

    Oh boy, Oh boy, Oh boy, 48-cores in a phone! It is a National Security nightmare, who needs a portable Cray?! Plug n Pray USB with a power/data/array port and woohoo, spatial physics is no longer in the the realm of data centers.

    "Enric Herrero, a research scientist at Intel Labs in Barcelona, explained that with the prototype chip someone could, for instance, be encrypting an email while also working on other power-intensive apps at the same time — without hiccups."

    Encrypt an email? Ah, so that is what a "smartphone" is supposed to be doing. "Without hiccups," guy must be new at Intel - Have you seen the OS being used on phones right now? Go back to basics and speeds increase without all the GUI crap - why isn't that being made on the fly?

    Everyone is being so upbeat and optimistic, you got 50 days left. :P

    --
    Agrisea Tsunami - Epyc Servers... https://agrisea.net/products
  71. Single thread on multiple cores by Ottibus · · Score: 1

    Having a thread move between cores is less efficient that keeping it on one core, so if I am running a single-threaded program I would expect to see one core busy and the other cores mostly idle.