48-Core Chips Could Redefine Mobile Devices 285
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."
Maybe...but not soon (Score:5, 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.
Re:Desktop (Score:5, 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...
Already here, kinda (Score:5, 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).
not with today's coding methodology (Score:5, 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.
Voice and video analysis (Score:5, 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.
Re:Desktop (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).
Re:Desktop (Score:5, Interesting)
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.
Re:Desktop (Score:4, Interesting)
From Rob Pike [usesthis.com]:
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.