Opportunities and Challenges for Smarter Mobile Devices

  • Published on
    07-Apr-2017

  • View
    214

  • Download
    2

Transcript

  • Opportunities andChallenges for SmarterMobile DevicesNamsung (Stephen) WooSamsung Electronics

    h FIRST OF ALL, I would like to congratulate allEDA professionals for the 50th anniversary of theDesign Automation Conference (DAC). Since thebeginning of DAC, both semiconductor industry andEDA industry have grown a lot. If we consider the

    revenue of last twenty years (19932012) alone, the semiconductor indus-try has grown 3.8 times (from $66B to$251B) and the EDA industry 5.2 times(from $1.3B to $6.7B).

    Looking back the papers presentedat DAC, we can easily find that DAC hasbeen leading and supporting the de-sign and manufacturing of semicon-

    ductor solution. In 1980s, the areas of physicaldesign, circuit simulation and testing saw manypapers. In 2000s, papers in the areas of low powerdesign, verification and embedded systems led DAC.In the last several years, we found strong interest inthe areas of thermal issues (for highly integratedchips) and variation tolerance (for advanced sili-con manufacturing). Interestingly, these two areas

    Editors notes:This article is based on a keynote address presented by the author at the50th DAC. It discusses the state-of-the-art in semiconductor technology andits interaction with smart mobile devices, wide I/O memory access andflexible displays.

    VYervant Zorian, Synopsys

    Namsung (Stephen) Woo gives his keynote address.

    IEEE Design & Test2168-2356/14 B 2014 IEEE Copublished by the IEEE CEDA, IEEE CASS, IEEE SSCS, and TTTC56

    50 Years of DAC: What Lies Ahead

    Digital Object Identifier 10.1109/MDAT.2014.2315955

    Date of current version: 19 May 2014.

  • address key technical challenges of semiconductorsolutions for smart mobile devices.

    This article will first examine recent trends ofsmart mobile devices and their influence on semi-conductor industry. Then, we will describe severalnew aspects of smarter devices (i.e., future wave ofsmart devices) and their impact on semiconductorand EDA technology.

    Mobile devices andsemiconductor industry

    The driving force of the semiconductor industryhas recently changed from PC to mobile device. Forinstance, the number of smart phones sold globallyin 2013 is expected to be 930 million units, and thenumber will grow to 1.1 billion units in 2014. Tabletsare also growing fast: from 210 million units in 2013to 290 million units in 2014.

    Along with the volume increase, smart deviceshave seen enhanced computing power for complexapplications. The mobile application processor(AP), which runs application software of smart de-vices, contains CPU core(s), graphics core(s), multi-media core(s) and other IP blocks in one silicon die.Since mobile devices are running on battery, mobileAP has to consume low energy including low stand-by current. As a result, mobile AP is typically built on

    a low-power (and, as a result, low-speed) siliconprocess.

    Then, how can we get higher performance ofmobile AP while consuming low energy? This diffi-cult goal has been achieved mostly by three areas ofdevelopment: advanced CPU/SOC architecture, so-phisticated circuit design (with the help of EDAtechnology) and advanced silicon process.

    Figure 1 shows a brief history of mobile AP forsmart phones. The X-axis is the year in which amobile AP was put in production, while the Y-axisshows the performance in terms of DMIPS (Dhrys-tone Million Instructions per Second). In 2009, forthe first time in mobile AP history, a mobile APrecorded 1.0 GHz clock speedwith low-power 45 nmsilicon process. In 2010 and 2011, mobile AP withdual CPU cores and quad CPU cores appeared, re-spectively. In 2012/13, mobile APs with eight CPUcores, four big cores and four little cores, were de-veloped and used in smart phones [1]. As shown inthe figure, the computing performance measured inDMIPS has been increasing every year.

    In order to offer high computing power at lowenergy consumption, many design techniques havebeen developed, and DVFS (Dynamic Voltage andFrequency Scaling) is one of them. Its goal is toadjust mobile APs supply voltage (Vdd) according

    Figure 1. Progress of mobile AP since 2008.

    March/April 2014 57

  • to its workload. Figure 2 shows an operationexample of DVFS, in which high Vdd is used whenAP deals with heavy workload and low Vdd whenworkload is light. In reality, APs firmware controls aPMIC (power management integrated circuit) thatprovides Vdd to AP.

    Silicon process plays an important role tosupport mobile AP. Figure 3 shows the progress ofCMOS process technology for mobile applicationsduring 20002013. The poly-SiOn technology hasbeen used for the 45 nm process, and since then,high-K/metal-gate (HK/MG) technology has beenadopted since 32/28 nm process. For 14 nm process(in some cases, 16 nm process), a 3D transistor,called FinFET, is being used.

    Other areas which have experienced dramaticgrowth in smart devices include camera sensors anddisplay driver chips. For instance, in the CMOS-based camera sensor chips for smart phones, thenumber of pixels in one sensor chip grew from1.3M pixels (in 2005) to 13M pixels (in 2012), whilethe size of camera sensor chip remains basicallythe same.

    Smarter devices: The next wave ofsmart mobile devices

    The next wave of smart mobile devices, smarterdevices, will offer much better user experience (e.g.,gesture recognition) and much broader solutions

    Figure 3. Progress of CMOS Process Technology.

    Figure 2. DVFS operation.

    IEEE Design & Test58

    50 Years of DAC: What Lies Ahead

  • (e.g., context-sensitive comput-ing). In this article, however, wewill look at only two areasrelated to technology.

    Wide IO for higher bandwidthbetween AP and memory

    Three elements of smarterdevices drive up the bandwidthrequirement between mobileAP and memory: 1) higherdata rate of air interface (e.g.,LTE-A), 2) higher graphics per-formance, and 3) higher displayresolution (e.g., UHD). TheWide IO technology, in which Memory isconnected to AP by TSV (Through Silicon Via), isthe best known approach for high bandwidthbetween AP and Memory.

    Figure 4 shows TSV in Wide IO and a way ofputting memory die on AP die. In addition to highbandwidth, the Wide IO technology provides low-energy memory access because of the closeness ofmemory to AP.

    The Wide IO technology has recently been real-ized in industry. The V system, which offers 512data lines between AP and mobile DRAM via TSV, isnow running real applications [2]. Experimentaldata show that the current implementation of WideIO offers 14% higher bandwidth than LPDDR3 andconsumes 60% less energy than LPDDR3.

    One challenge of the Wide IO technology iscomplex memory architecture. That is, the mobileAP has to deal with two types of memory (i.e., inter-nal memory connected by TSV and external mem-ory) with different access time. Mobile software hasto be intelligent if it wants to utilize internal memoryas much as possible. Some research groups havemade good progress in managing this complexmemory system in smarter devices.

    Flexible displayThe display resolution of smart mobile devices

    has been moving up, and current smart phones offerfull HD (FHD) resolution. In the near future, UHDdisplay will be used in high-end smarter devices.

    One disruptive display technology for smarterdevices is flexible display. As demonstrated at thisyears (2013) CES show [3], flexible display isworking well at laboratories. Recently, in 3Q13, an

    early form of flexible display, called curveddisplay,was adopted in two smart phone models from twocompanies. It is reasonable to expect smarterdevices with fully flexible display in the nextcouple of years.

    If we have fully flexible display at smarterdevices, the other electronics part (i.e., a PCB withmultiple chips on it) of smart devices must alsochange. That is, if the current form of PCB does notchange, the advantages of flexible display will notbe fully exploited.

    For the future path of electronics portion insmarter devices with flexible display, we can learnlessons from the current display driver IC. As shownin Figure 5, the current display driver ICs are put ona plastic film, which is connected to a display.

    If we can reduce the total number of chips insmarter devices from more than dozen (which is thecase with the current smart devices) to one or two,

    Figure 5. Display driver IC on a plastic connectedto a display.

    Figure 4. Wide IO technology.

    March/April 2014 59

  • and if we can put them on a plastic film, we wouldbe able to build smarter devices that fully utilizeflexible display. If we get this solution, we will haveSOP (System on Plastic) or SOF (System on Film)that works naturally with flexible display.

    SMART MOBILE DEVICES have contributed to thegrowth of semiconductor industry, and the trend willcontinue in the near future. In this article, we de-scribed an interaction between smart mobile de-vices and semiconductor solution; we showed thathigh-performance and low-energy mobile APsallowed smart phones to progress, and vice versa.

    We also touched smarter devices (i.e., the nextwave of smart mobile devices) and two technicalinitiatives, wide IO and flexible display, for them. Foreach initiative, we introduced the most recentresult(s) and discussed future paths.

    EDA technology has been helping semiconduc-tor industry (both in design and manufacturing)

    which, in turn, contributed to smart devices. Wehope this virtuous circle continue in the future. h

    h References[1] AP Data Book, System LSI, Samsung Semiconductor,

    20102013.

    [2] V System With Wide IO Solution, System LSI,

    Samsung Semiconductor, 2013.

    [3] Keynote Talk by This Author at CES 2013, Las Vegas,

    NV, USA, Jan. 2013.

    Namsung (Stephen) Woo is the president ofSystem LSI at Samsung Electronics, San Diego, CA,USA. Before joining Samsung, he worked at BellLaboratories, Murray Hill, NJ, USA and TexasInstruments, San Diego, CA, USA.

    h Direct questions and comments about this articleto Namsung (Stephen) Woo, System LSI, SamsungElectronics.

    IEEE Design & Test60

    50 Years of DAC: What Lies Ahead

    /ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages false /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False

    /Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ]>> setdistillerparams> setpagedevice