By Duncan Bremner, CTO SureCore Limited
Editor’s note: sureCore just announced availability of its 28nm FD-SOI memory compiler (press release here), which supports the company’s low-power, Single and Dual Port SRAM IP. Here, the company’s CTO explains why this IP is getting such impressive results.
~ ~ ~
Recently, sureCore announced results from a 28nm FD-SOI test chip that showed dynamic power savings exceeding 75% and static power cuts up to 35% (when compared against a number of current commercial offerings), while only incurring a 5-10% area penalty for its ultra-low power SRAM IP.
And while this data is easily substantiated as shown in Figure 1, the sceptical industry pundits have raised questions that fall into two camps: (a) That can’t be done; or (b) How did they manage that? In answer to both of these questions, here’s a quick look at the history and engineering strategy that we adopted to deliver these results.
Looking back to the early days of sureCore, SRAM fascinated us because despite many process iterations, the SRAM in use today bears a striking resemblance to the SRAM architectures that existed in the ’70s and ’80s. We concluded that no one had really taken a “blank-sheet-of-paper” look at the architecture for over 40 years. Recognising the growing importance of power efficiency for SoCs targeting forward-looking applications such as wearables, IoT, and other mobile devices, we examined power consumption in detail, and began by investigating how we could reduce SRAM power to a level attractive to the next generation of power critical, SoC designers.
Our starting point differed significantly from the traditional approach to SRAM R&D that typically starts at the bit cell. We recognised that the basic bit cell is fixed by the foundry; it’s a piece of electronics that is carefully optimised for fabrication. Modern bit cells are designed by the foundries who tend to put an emphasis on the broadest possible manufacturability drivers; yield and faster-time-to-volume as opposed to more performance-centric metrics. Their focus is on the front-end process optimisation, area and yield.
The basic rule of R&D fabless foundry engagement has been, “use the storage array – you won’t get a better packing density.” Consequently, the application use model had become separated from the technology — ‘faster or cheaper’ became the industry’s mantra instead of ‘faster and better’. This resulted in SRAM design teams focusing on how to build more sensitive read amplifiers to detect the signals, and better write amplifiers to drive the signal on to the bit cell. Not much time was spent looking at the fundamental architecture and asking: “Is this the best way?”
sureCore decided to take a more holistic view and stood back from the whole problem. We started with a clean sheet of paper and asked, “Where does the power go when you start storing data on SRAM?”
We discovered that a lot of the power is consumed hauling parasitic capacitance around. Our design strategy was therefore very simple; we developed a system architecture to optimize power while still retaining the area advantages of the standard foundry bit cell.
Simply stated, we architected the internal block architecture of SRAM by splitting the read amplifier function into a local and global read amplifier, thus dividing the capacitive load from the word-line, only driving the areas being addressed and not the whole array. This resulted in significant dynamic power savings during the read cycle. In a similar fashion, we reduced the write cycle power by a similar amount. Whilst hierarchical solutions are not new, the sureCore “secret sauce” is at circuit level developed by our engineering teams leading to not only significant power savings, but also comparable performance levels.
Our “blank sheet” approach delved deep; right down to the fundamental device physics level. Our strategic partners, Gold Standard Simulations — recognised world leaders in modelling devices at the atomic level and experts in nano-scale process nodes, helped us to understand the behaviour and limitations of processes at nodes below 28nm at a device level and bit cell level. Combining this fundamental device understanding with excellent circuit design and system analysis skills, we’ve identified where existing SRAM solutions waste power, and architected our solution to avoid this; we deliver power savings without the added complexity of write and read-assist.
At the outset, we determined it was important that our IP be process-independent. sureCore IP is based on architecture and circuit techniques rather than a reliance on process features. The result of this is technology that can reduce power in standard bulk CMOS, but is equally applicable to newer FinFET or FD-SOI processes and across all geometries, even down to 16nm and below. We believe our approach is paying off and, because we insisted in retaining the foundry optimised bit cell, sureCore’s technology can be retrofitted into existing designs enabling extended product life cycles.
This is our basic technology story… a start-up deciding to take a fresh look at an old technology and dramatically improving power performance over 75% compared with existing solutions. This is a new approach to SRAM power consumption for power sensitive applications and it delivers tangible battery life benefits to both the end user and the FD-SOI designers. Today’s FD-SOI technology is optimised for low power applications, bringing extended battery life to the nascent markets of wearables and IoT.
An IBM paper on a 14nm SOI-FinFET SRAM functional down to 0.3V has garnered press attention. The paper, entitled 14nm FinFET Based Supply Voltage Boosting Techniques for Extreme Low Vmin Operation by R.V. Joshi et al, was presented during the Symposium on VLSI Circuits in Kyoto, Japan in June. According to the abstract, the authors presented a new, “… dynamic supply and interconnect boosting techniques for low voltage SRAMs and logic in deep 14nm FinFET technologies. The capacitive coupling in a FinFET device is used to dynamically boost the virtual logic and array supply voltage, improving Vmin. Hardware measurements show a 2.5-3x access time improvement at lower voltages and a functional Vmin down to 0.3V. Results are supported by novel physics-based capacitance extraction and novel superfast statistical circuit simulations.” EETimes reported on the paper in a piece entitled “IBM Slashes Next-Gen Power” (see it here), wherein the lead author confirmed that this work was based on a 14nm SOI-FinFET architecture.
SureCore’s ultra-low power SRAM technology on 28nm FD-SOI saves 70% in read/write power and reduces leakage by 30% compared to 40nm bulk implementations, writes SemiconductorEngineering Editor-In-Chief Ed Sperling (read the article here). Hitting the sweet spot for mobile, IoT and wearables, SureCore recently raised $1.6 million in funding.
The new IEEE S3S conference promised rich content, as it merged both The IEEE International SOI Conference and the IEEE Subthreshold Microelectronics Conference, completed by an additional track on 3D Integration.
The result was an excellent conference, with outstanding presentations from key players in each of the three topics covered. This quality was reflected in the increased attendance: almost 50% more than at the SOI conference last year.
The new triptych at the heart of the conference was well illustrated by the plenary session, which combined a presentation on ST’s FD-SOI technology by Laurent LePailleur (STMicroelectronics), one on Low Power Design, by Bob Bordersen (UC Berkeley), and one on monolithic integration by Zvi Or-Bach (MonolithIC 3D™).
Professor Bordersen’s presentation dealt with power efficiency, explaining how developing dedicated units with a high level of parallelism and a low frequency can boost the number of operations performed for 1nJ of expanded power. He illustrated his point by showing how an 802.11a Dedicated Design for Computational Photography can reach 50,000 OP/nJ while an advanced quadcore microprocessor will not even reach 1 OP/nJ. Such is the price of flexibility….but the speaker claims this can be overcome by using reconfigurable interconnects.
The “Best SOI Paper” award went to a GlobalFoundries/IBM paper entitled “FinWidth Scaling for Improved Short Channel Control and Performance in Aggressively Scaled Channel Length SOI FinFETs.” The presenter, Abhijeet Paul (GF) explained how narrower Fins can be used to improve short channel effects while actually giving more effective current without degrading the on-resistance. (See the DIBL and SS improvement on the chart.)
The”Best SOI Student Paper” award went to H. Niebojewski for a detailed theoretical investigation of the technical requirements enabling introduction of self-aligned contacts at the 10nm node without additional circuit delay. This work by ST, CEA-Leti and IEMN was presented during the extensive session on planar FD-SOI that started with Laurent Grenouillet’s (CEA-Leti) invited talk. Laurent first updated us on 14nm FD-SOI performance: Impressive static performance has been reported at 0.9V as well as ROs running at 11.5ps/stage at the very low IOFF=5nA/µm (0.9V & FO3). Then he presented potential boosters to reach the 10nm node targets (+20% speed or -25% power @ same speed). Those boosters include BOX thinning, possibly combined with dual STI integration, to improve electrostatics and take full advantage of back-biasing as well as strain introduction in the N channel (in-plane stressors or sSOI) combined with P-channel germanidation.
sSOI (strained SOI) was also the topic of Ali Khakifirooz’ (IBM) late news paper, who showed how this material enables more than 20% drive current enhancement in FinFETs scaled at a gate pitch of 64nm (at this pitch, conventional stressors usually become mostly inefficient).
An impressive hot topics session was dedicated to RF CMOS.
J. Young (Skyworks) explained the power management challenges as data rates increase (5x/3 years). Peak power to average power ratio has moved from 2:1 to 7:1 while going from 3G to LTE. Advanced power management techniques such as Envelope Tracking can be used to boost your system’s efficiency from 31% to 41% when transferring data (compared to Average Power Tracking techniques), thus saving battery life.
Paul Hurwitz (TowerJazz) showed how SOI has become the dominant RF switch technology, and is still on the rise, with predictions of close to 70% of market share in 2014.
The conference also had a strong educational track this year, with 2 short courses (SOI and 3DI) and 2 fundamentals classes (SOI and Sub-Vt).
The SOI short course was actually not SOI-restricted, since it was addressing the challenges of designing for a new device technology. P. Flatresse (ST) and T. Bednar (IBM) covered the SOI technology parts (FD-SOI and SOI FinFETs for ASICs respectively), while D. Somasekhar (Intel) gave concrete examples of how the change of N/P performance balance, the improvement of gate control or the introduction of Mandrels has affected design. Other aspects were also covered: Design for Manufacturing (PDF), IP librairies (ARM) and design tools (Cadence) for the 14nm node, to make this short course very comprehensive.
The rump session hosted a friendly discussion about expectations for the 7nm node. It was argued that future scaling could come from 3DI, either through the use of monolithic 3D integration or stacking and TSVs because traditional scaling is facing too many challenges. Of course, 3DI may not yet be economically viable for most applications, and since it is compatible with traditional scaling, we might well see both developed in parallel.
3D integration was also the topic of another joint hot topics session covering various fields of investigation, like co-integration of InGaAs and Ge devices (AIST), or 3D cache architectures (CEA-Leti & List). A nice example was given by P. Batra (IBM) of two stacked eDRAM cache cores, where the 16Mb cache on one layer is controlled by the BIST on the other layer and vice-versa with the same efficiency as in the 2D operation.
The first edition of this new conference was very successful, with a good attendance, two sessions running in parallel, extensive educational tracks, a large poster session and a lot of very high quality content. The two hot topics sessions generated a lot of enthusiasm in the audience.
Similar sessions will be repeated at the conference’s next edition, in the San Francisco area. It promises to offer outstanding content once more, and we already urge you to plan to submit papers and attend it.
Last May, we already let you know about the IEEE S3S conference, founded upon the co-location of The IEEE International SOI Conference and the IEEE Subthreshold Microelectronics Conference, completed by an additional track on 3D Integration.
Today, we would like let you know that the advance program is available, and to attract your attention on the incredibly rich content proposed within and around this conference.
The conference revolves around an appropriate mix of high level contributed talks from leading industries and research groups, and invited talks from world-renowned experts.
The complete list of posters and presentations can be seen in the technical program.
This year some additional features have been added, including a joint session about RF CMOS as well as one about 3D integration. Check the list of participants on those links, and you will see that major players in the field are joining us!
Our usual rump session will let us debate what the 7 nm node and beyond will look like, based on the vision presented by our high profile panelists.
There will be 2 short courses this year, and 2 fundamentals classes. Those educational tracks are available to you even if you do not register for the full conference.
On Monday October 7th, you can attend the short course on “14nm Node Design and Methodology for Migration to a New Transistor Technology“, that covers specificities of 14nm design stemming from the migration of classical bulk to bulk to FinFET/FDSOI technologies..
Alternatively, on the same day you can attend the “3D IC Technology” short course, introducing the fundamentals of 3D integrated circuit technology, system design for 3D, and stress effects due to the Through Silicon Via (TSV).
On the afternoon of Wednesday October 9th, you can opt to follow the Sub Vt Fundamentals Class on “Robust subthreshold ultra-low-voltage design of digital and analog/RF circuits” or the SOI Fundamentals Class “Beyond SOI CMOS: Devices, Circuits, and Materials “.
You could also prefer to take the opportunity to visit the Monterey area.
The conference has always encouraged friendly interactions between the participants, and because it covers the complete chain, from materials to circuits, you are sure to meet someone from a field of interest. The usual social events, welcome reception, banquet and cookout dinner, will provide you with more openings for networking, contemplating new project opportunities or getting into technical discussions that could shed new light on your research.
To take full advantage of this outstanding event, register now!
Please visit our Hotel Registration Information page to benefit from our special discounted room rates at the conference venue, The Hyatt Regency Monterey Hotel and Spa.
The latest conference updates are available on the S3S website (http://S3Sconference.org).
Targeting low-power SRAM for FD-SOI and FinFETs, UK physical IP start-up sureCore has received a £250K grant (about 292K Euros or $380.5K) from the Technology Strategy Board SMART. Working with the major foundries developing FD-SOI and FinFET technologies, the grant will be used in the development of a demonstrator chip to showcase sureCore’s patented array control and sensing scheme, which significantly lowers active power consumption. Through a combination of detailed analysis and using advanced statistical models, sureCore has designed an SRAM memory consuming less than half the power of existing solutions. SureCore is working closely with Gold Standard Simulations (GSS) Ltd. (GSS Founder/CEO Asen Asenov is a sureCore director).
As the SOI circuit switches, the body voltages of the switching transistors will change from their previous steady state condition. This is called the history effect. This is one of the most interesting circuit design issues in SOI but it is also a benefit of SOI that contributes to the SOI performance advantage over bulk CMOS. Leveraging tools to model and predict the body voltage of the SOI transistors will lead to a successful SOI circuit design. Read More
What is the best transistor structure to meet SRAM performance and yield requirements at the 22nm node? The semiconductor device research group at UC Berkeley pioneered the FinFET structure in 1998. Now SOI-based FinFETs lead the field of candidate structures to eventually replace the planar bulk MOSFET. In the near term, yield and manufacturability may trump performance for high-volume markets, however.
Our work recently presented at the 2009 IEEE SOI Conference indicates that a planar fully depleted (FD) structure on very thin-BOX (~10nm thick) is a compelling candidate. Specifically, we found that for 6T-SRAM cells at the 22nm node: