What would you do with extra 40%…?

What do you use your smartphone, tablet, or Ultrabook for if you had enough battery life so that you don’t have to worry about recharging for extra 40 minutes?  Maybe you’re a gamer, immersing yourself in 3D worlds and fast-paced action.  Perhaps you download and stream movies in high-definition.  Or maybe you take your own high-definition videos of the kids, uploading them to YouTube to share with the rest of the family. James Tschanz, Research Scientist at Intel Labs, tells that  “If you are using your computing devices for these purposes, chances are that much of the heavy lifting is done by the graphics processor – a small yet powerful computing engine that sits right next to the processor in the silicon chip ‘brain’ of your computer.” As our handheld devices have become powerful computers in their own right, Intel – along with others in industry – has recognized the importance of graphics and has dramatically improved graphics capabilities with each new design, enabling new usages that would have been unthinkable only a few years in the past.

With all of this exciting graphics capability, though, comes a major challenge – power.  Keeping up with advanced 3D games or computation required to drive an HD display takes energy, and this energy is provided by the tiny battery hidden in your phone. Besides driving the display on your phone, this battery also provides the power to the processor cores (the ‘brain’), radio features (the ‘ears’), camera (the ‘eyes’), bright screen, and all of the other essential features.  Therefore the engineers that design these graphics processors need to be very creative – creative with how much power to use, creative with how the computations are performed, and creative with how to deliver the absolute best performance while sipping minimal energy from the battery.

This year at International Solid State Circuits Conference 2014, Intel Labs is presenting an energy-efficient graphics processing core built on the latest 22nm tri-gate SoC process technology.  This graphics core incorporates several new features that allow it to improve energy efficiency by 40% – essentially giving longer battery life for the same performance or improving performance when you really need it.  This core also has the ability to very quickly go into standby mode where power savings is 10X compared to active – yet another way of saving energy for when it is absolutely needed without the need to worry about charging frequently.

blog GPU 

I asked James, where does this energy efficiency improvement come from? 

He says, “First, it is well known that the best way to reduce power is to reduce voltage, so Intel’s graphics cores are designed to take advantage of our 22nm tri-gate transistor technology which allows high-performance operation even as voltage is lowered.  However even with these advanced transistors, some specialized circuits (such as memory arrays) inside the graphics core can still limit the minimum voltage (called Vmin) which can be used.  Therefore this design demonstrates a new “selective boosting” technique which uses a slightly higher “boosted” voltage for key parts of Vmin-limiting memory arrays in the core.  What does this accomplish?  By dynamically boosting only small parts of the core, the rest of the core can go to even lower voltage – giving dramatic energy efficiency improvements.

Even with this voltage-reduction technique, however, there is still room for improvement.  When the graphics core is operating, it is switching units on and off very quickly to optimize power and performance.  All of this switching can cause ‘noise’ on the internal power supply – voltage droops and glitches which if not handled correctly could cause an error or failure inside the core.  Have you ever turned on a high-powered appliance in your house – maybe a vacuum cleaner or space heater – and noticed that the lights slightly dim?  A similar effect can happen inside the processor chip, and although there are advanced voltage regulators to minimize this issue, a small voltage ‘safety margin’ is added to ensure that everything still works 100% correctly.  The only problem with this safety margin is that it increases voltage, which – as we just discussed – is the opposite of what we are trying to do to save energy.   In this graphics core we therefore include a technique – called adaptive clocking – to reduce this safety margin by detecting voltage droop events and slowing down the clock frequency of the core to prevent failures.  These droops are infrequent and very fast, so slowing down the clock is not perceptible to the user.  The user will notice, however, the extra battery life achieved by this technique!

Finally, this core includes special circuits that allow it to go very quickly into a ‘sleep’ state, saving all of the important data but consuming very little power.  Just as in selective boosting, this requires a separate voltage supply which is kept ‘on’ even in sleep mode, allowing the voltage to all of the other circuits in the core to be completely turned off.  In fact, even the separate ‘retention’ supply can be lowered in sleep mode through an on-die sleep voltage regulator, giving further power savings.”

These techniques are currently found in a testchip within Intel Labs. They demonstrate how they have the potential to greatly improve experience on future mobile devices and may provide a truly wire-free experience.

Divya Kolar

About Divya Kolar

Divya Kolar holds a M.S in Computer Science conferred in 2006 from Portland State University. She joined Intel in 2005 and has previously worked as a Software Engineer where she was an active researcher in various security and manageability technologies like Intel® Active Management Technology. Today she is a Vision Strategist in the Intel’s largest research group and is responsible to promote Intel technologies to external media partners besides performing ecosystem enabling and competitive technology analysis for Intel Labs’ microprocessor research. Besides her responsibilities at Intel she has always been enthusiastic in promoting and encouraging young adults to stay in computing. She is an active board member for the largest women employee group at Intel and has been an active member in Anita Borg Institute and local SWE chapters since 2007 and has conducted multiple presentations at these conferences for over 5 years.

3 Responses to What would you do with extra 40%…?

  1. I have learn many fantastic products in this article. Unquestionably worthy of book-marking for returning to. I wonder simply how much try you set to make the sort of great useful website.

  2. Robert Novak says:

    This article fails to disclose if Intel is placing the GPU cores across the PCIe buss away from the CPUs or if they are on a faster buss that allows the GPUs and CPUs to share the memory buss. Not all GPU tasks are offloaded graphics processing (even in tablets and smartphones). Other tasks include encryption/decryption, compression/decompression, accelerated pattern matching (load regular expressions in texture memories). For storage processing, GPUs that can be quickly accessed by Atom CPUs would make them suitable for storage, especially for computing the cryptographic hash digest of data stored on disk or transmitted over networks to insure the integrity of the data since newer Atom CPUs have included RAID instructions (vulnerable to bitrot) but not SHA256/SHA512/SHA3.

  3. xcore fan says:

    “What do you use your smartphone, tablet, or Ultrabook for if you had enough battery life so that you don’t have to worry about recharging for extra 40 minutes? ”

    id probably use the time to write a complaint that Intel are not providing generic affordable Si Photonics Optical PCI Express* (OPCIe*) in 2014 for desktop i5/i7 mass home/SOHO consumer use on all their quad 14nm cores.

    that they are not it seems even providing generic AVX2 SIMD or better for their mobile SOC’s so we cant finish that high visual quality x264 encode in real time.

    and would they PLEASE provide the option to actually buy quad core/octa thread WITH an updated Iris™ Pro E Graphics core as standard for increased data decode throughput etc… and some x264/x265 patches from your dev’s would be so nice too…

    a nice cheap Home/SOHO 10GbE card x2 + router/switch pack would be nice too so we can the consumer masses can finally go fast over our home/SOHO LAN networks to our freeNAS boxs please….