Key takeaway #2: If you need a Goldilocks process node – where you’ll get just the right balance between active power, unit cost and investment – look to FD-SOI. And, btw, the IP landscape has improved dramatically. Those were just some of the great points made by Huibert Verhoeven (shown above), GM/SVP of Synaptics’ IoT Division in his talk at the recent SOI Symposium in Silicon Valley.
BTW, if you missed part 1 of our coverage —Silicon Valley SOI Symposium a Huge Success. Key Takeaways (Part 1) Here. – you’ll want to be sure to read it, too. Almost all of the presentations are now posted on our website – click here to access them.
In this post here, we’ll cover presentations by Synaptics, GlobalFoundries, STMicroelectronics, Anokiwave and Dolphin Integration. It was a really full, day, so be sure to stay turned for Part 3 of our coverage to follow shortly: it will highlight the remaining presentations and panel discussions.
Synaptics’ Verhoeven’s presentation Revolutionizing User Experience Through Secure Neural Network Acceleration at the Edge was about Smart Home and using SOI. Synaptics is a human interface (HMI) company that’s been doing neural networks since 1986. They’ve always been on the leading edge, from their first shipment of PC touchpads to becoming a dominant force in all things HMI today: they now ship over a billion units annually.
They currently have SOI products shipping with dedicated neural networks for voice, he said. European [privacy] regulations have played a part in driving their use of SOI, as have challenges regarding power and heat. Things are getting smarter at the edge. For example, not only do users want their coffee machine to offer the usual morning espresso, Synaptics says that the next step is for your coffee machine to recognize you’re looking extra tired and ask if you might want a double?!
For them Smart Home and multi-modal applications are the primary area of interest, as well as some automotive. Although their biggest customers have resources, others need guidance. Voice is a critical component, but now you also need video and display.
Why SOI? Their HMI vision requires low power, significant computation and dedicated neural network hardware, explained Verhoeven, so FD-SOI with RF meets their needs. “22nm SOI is a Goldilocks IoT Process Node,” he proclaimed. It gets the combination of active power, unit cost and investment just right. What’s more, he said, “The IP landscape has improved dramatically. Our choice of SOI was not an accident.” Be on the lookout for more products leveraging FD-SOI over the next six months, he concluded.
At this point on SOI, they’ve got 1 TOPS products with dedicated NPU for speakers, soundbars, Wi-Fi mesh, appliances, STBs and smart displays. These products have voice and sensor real-time (RT) AI. Next up is >4 TOPS on SOI with dedicated NPU, targeting STBs and smart displays with voice, video, imaging and RT AI.
“Our clients are at the forefront of changing the world,” declared Mark Granger, VP of the Automotive Product Line at GlobalFoundries. His presentation, Capturing High Growth Market Opportunities with SOI, detailed how mobility, automotive and IoT are the growth markets for SOI. So not unsurprisingly, GF’s 22nm FD-SOI technology, 22FDX, is seeing particular traction in mobile, edge, wearables and automotive.
They’ve got twice as many tape-outs this year as they did a year ago, he noted. GF’s SOI portfolio includes 22FDX®, 45RFSOI and 8SW/7SW RF SOI for 5G/mobility; 22FDX for automotive (fully qualified for automotive Grade 2, with Grade 1 on the way); and 22FDX, 130RFSOI and 8SW/7SW RF SOI for IoT.
GF has announced a stream of good news recently:
You might have heard about the Dolphin Integration news, as we covered it recently here at ASN (if not, be sure to read it here). Dolphin’s IP and methodology solutions address energy efficiency challenges. Automated transistor body biasing adjustment can achieve up to 7x energy efficiency with power supply as low as 0.4V on 22FDX designs. At the Silicon Valley event, Dolphin Integration CEO Philippe Berger provided additional information in his talk, FD-SOI IP Platform for Energy-Efficient IoT SoC.
In another GF-related talk, Nitin Jain, the CTO of longtime GF RF-SOI customer Anokiwave presented Unleashing the mmWave Phase Array Using SOI for 5G & Satcom. Anokiwave is a fabless semi IC company (you’ll find a good technical discussion of mmWave phase array written by their Chief Architect here). They do active antennas (aka phased array), something the military’s done for a long time, but now Anokiwave is bringing it to new markets and applications including radar, satcom and 5G. What they’ve been able to do is planarize the active antennas. They use GF’s 45RFSOI process technology for phased array systems because of the cost, performance, scalability and system enhancements it enables. 45RFSOI, he explained, is ideal for beam-forming FEMs (including the switches, LNAs and PAs). The move to 5G/mmWave is going to require a lot of antennas, so these Anokiwave ICs are headed to high volumes, concluded Jain.
As Roger Forchhammer, Director of Business Development at STMicroelectronics pointed out in his presentation, Automotive FD-SOI Microcontrollers with Embedded PCM, ST pioneered FD-SOI (and that was almost a decade ago, btw). Then in February 2019, they announced a world first: they’d begun sampling 28nm FD-SOI microcontrollers (MCUs) with embedded non-volatile memory (eNVM) based on embedded Phase-Change Memory (ePCM) to 10 alpha customers. These MCUs target powertrain systems, advanced and secure gateways, safety/ADAS applications, and vehicle electrification.
(In case you want technical details, the breakthrough ePCM eNVM was first presented at IEDM in December 2018 – you can get the presentation that accompanied the paper, Truly Innovative 28nm FDSOI Technology for Automotive Microcontroller Applications embedding 16MB Phase Change Memory, from the ST website.)
In his Silicon Valley presentation, Forchhammer said they’re now doing Stellar, a whole family of automotive products on FD-SOI. To do it, they’d taken an existing device and moved it to 28nm FD-SOI with ePCM, which they manufacture at their fab in Crolles, France. A major advantage for automotive he cites is that in software updates it’s bit-level programmable. “ST is fully behind FD-SOI,” he concluded, adding that we’re see more automotive as well as IoT products coming soon.
Well folks, that’s all for this post. We’ll finish up our coverage of the SOI Consortium’s 2019 Silicon Valley Symposium in the next ASN post (there was so much to cover!). So please stay tuned.
Takeaway #1: As NXP VP Ron Martino noted in his opening keynote at the recent SOI Symposium in San Jose, FD-SOI is the technology platform for enabling edge computing, and ultra-low power is the sweet spot.
Organized by the SOI Consortium with support from our members, the recent SOI Symposium in Silicon Valley was an enormous success. Close to 300 decision makers signed up – more than double what we saw just a couple years ago. Attendees spanned the ecosystem: from end-users to design to foundries and right up to the investment community. The presentations and panel discussions were absolutely terrific, and almost all are now freely available – click here to get them.
The focus was heavily on FD-SOI this time, but some very interesting RF-SOI talks were given as well. This was a day packed with presentations by players from across the SOI ecosystem. In this post, we’ll only cover a few. But the others will follow quickly, so watch this page. And now without further ado, let’s dive in.
NXP is designing FD-SOI into many new products, said Martino, GM of the i.MX Processor Application Product Line. There’s a new wave of products – generically you could call them IoT but in fact they’re found throughout the industry. It’s about interacting with the cloud, so edge processing is critical. His presentation, Embedded Processors for Future Applications, is now freely available for downloading from our website.
The new i.MX7ULP is a great example of ULP in the sweet spot. From a design standpoint, it leverages IP, power optimization, and what he described as “starter biasing”. That gets them the long battery life with 2D & 3D graphics they need for wearables and portables in consumer and industrial applications.
Having deepened their expertise in biasing, NXP has now moved on to “advanced biasing” for the next generation of products. For example, the i.MX RT ULP (real-time, ultra-low-power) series are “cross-over” processors, which Martino says are the “new normal”. They deal with a high number of sensor inputs. The i.MX RT 1100 MCUs, which have been qualified for automotive and industrial applications, are breaking the gigahertz performance barrier with a low-power, 28nm FD-SOI process.
Another new product leveraging advanced biasing is the i.MX RT 600. They’ve done hardware acceleration on specific functions and optimized around visionand voice integration at low cost and power.
Likewise for the i.MX 8 and 8X subsystems for automotive and industrial applications. At Embedded World, they showed it driving advanced OLED screens, cameras (for parking, for example), V2X, audio, user monitoring (like driver pupil tracking), and integration into the windshield in a heads-up system. This is the high end of the capability of 28nm FD-SOI, he said. It’s a 6 CPU core system with multiple operating systems, about which he said: “It’s the dashboard…it’s amazing.”
FD-SOI enables a scalable solution for real-time and general compute with the lowest leakage memory, the best dynamic and static power, Martino concluded. NXP’s leadership in body biasing is enabling edge compute, and we can expect to see more content coming soon.
In another NXP presentation later in the day, Stefano Pietri, Technical Director of the company’s Microcontrollers Analog Design Team caught a lot of people’s attention. A wave of cameras went up to capture each of his slides in Analog Techniques for Low Power, High Performance MPU in FD-SOI – but you can get the whole thing now from our website. It’s a very technical presentation, in which he details the many ways FD-SOI makes the analog team’s job easier, enabling them to get performance not available from bulk technologies. They developed a lot of in-house expertise and IP (see slide 16 for a catalog of the IP).
Tim Dry, Director of Foundry Marketing: Edge and End Point presented Samsung’s FDS with MRAM: Enabling Today’s Innovative Low Power Endpoint Products. In a telling first, Samsung has made this presentation available on our website.
FD-SOI covers the wide range of requirements for intelligent IoT, he explained: from high to low processing loads; and active to dormant processing duty cycles. That includes chips that will last for ten years, and need to be able to wake up fast and kick right into high performance. These products are 50% analog, and packaging is part of the solution (especially for the RF component).
Samsung has been shipping 28nm FD-SOI (which they call 28FDS) since 2015, first in IoT/wearables, then in automotive/industrial and consumer. Yields are fully mature. In March 2019, they announced mass production of eMRAM on 28FDS. It’s a BEOL process, adding only 3 masks. It cuts chip-level power by 65% and RF power by 76% over 40nm bulk with external memory. Beyond the fact that it’s 1000x faster than eFlash, eMRAM also has other advantages that make it especially good for over-the-air updates, for example.
Samsung also has RF and 5G mmWave products shipping in 28FDS. The company has a fantastic ecosystem of partners helping here, said Dry. In AI at the endpoint, they’re shipping IoT products for video surveillance cameras: some are high speed, but some are also low speed – it depends on the detection use case. And most importantly for the design ecosystem, the IP is all ready.
Next up for Samsung is 18FDS, which will ship this year with RF, then in 2020 with eMRAM. 18FDS, Dry said, is optimized for power reduction. Compared to 28FDS, it’s got 55% lower power consumption, 25% less area and 17% better performance at the same power. You’ll hear more about it as well as their design services if you’re at the Samsung Foundry Forum in May (registration info here).
Kelvin Low, VP of Marketing for Arm’s Physical Design Group (PDG) gave a presentation entitled Biased Views on the Industry’s Broadest FDSOI Physical IP Solution. By way of background, Arm and Samsung Foundry recently announced a comprehensive, foundry-sponsored physical IP platform, including an eMRAM compiler for 18FDS. In case you missed it, at the time Arm Senior Product Marketing Manager Umang Doshi described the offering in an Arm Community / Developer physical IP blog, which Arm graciously agreed to share with ASN readers.
At the SOI Symposium, Low emphasized to the audience that Arm now has the broadest range of FD-SOI + IP solutions. It addresses mobile, consumer, IoT, automotive and AI/ML.
There are 18FDS POP (processor optimized pipe) packages for Arm Cortex-A55, Cortex-R52 and Cortex-M33 processors. IP integrates biasing and a number of standard PVTs (corners). And since the Samsung platform is foundry-sponsored, it’s free.
Arm did a test chip with eMRAM, which they’ve just gotten back. It’s functional (some details are available in slide 14 of their presentation), and the company is now preparing a demo board that they’ll be showing shortly. Watch this page!
That’s all for this post. The next post — part 2, covering presentations by Synaptics, GlobalFoundries, STMicroelectronics, Dolphin Integration and Anokiwave — is now available. Click here to read on.
GlobalFoundries and Dolphin Integration are collaborating on the development of a series of adaptive body bias (ABB) solutions to improve the energy efficiency and reliability of SoCs on GF’s 22nm FD-SOI (22FDX®) process technology for a wide range of high-growth applications such as 5G, IoT and automotive. The goal of the IP is to accelerate energy-efficient SoC designs and push the boundaries of single-chip integration. The design kits with turnkey ABB solutions will be available starting in Q2 2019.
As part of the collaboration, Dolphin and GF are working together to develop a series of off-the-shelf ABB solutions for accelerating and easing body bias* implementation on SoC designs. ABB is a unique feature of FD-SOI that enables designers to leverage forward and reverse body bias techniques to dynamically compensate for process, supply voltage, temperature (PVT) variations and aging effects to achieve additional performance, power, area and cost improvements beyond those from scaling alone.
The ABB solutions in development by GF and Dolphin consist of self-contained IPs embedding the body bias voltage regulation, PVT and aging monitors and control loop as well as complete design methodologies to fully leverage the benefits of corner tightening. GF says its 22FDX technology offers the industry’s lowest static and dynamic power consumption. With automated transistor body biasing adjustment, Dolphin Integration can achieve up to 7x energy efficiency with power supply as low as 0.4V on 22FDX designs.
“We have been working with GF for more than two years on advanced and configurable power management IPs for low power and energy efficient applications,” said Philippe Berger, CEO of Dolphin Integration. “Through our ongoing collaboration with GF, we are focused on creating turnkey IP solutions that allow designers to realize the full benefit of FD-SOI for any SoC design in 22FDX.”
“In order to simplify our client designs and shorten their time-to-market, GF and our ecosystem partners are helping to pave the way to future performance standards in 5G, IoT and automotive,” said Mark Ireland, vice president of ecosystem partnerships at GF. “With the support of silicon IP providers like Dolphin Integration, new power, performance and reliability design infrastructures will be available to customers to fully leverage the benefits of GF’s 22FDX technology.”
As STMicroelectronics Fellow and Professor Andreia Cathelin has beautifully noted, “Body biasing is not an obligation. It’s an opportunity.” And GF/Dolphin clearly aim to make that opportunity a much easier and more powerful one to take advantage of.
~ ~ ~
*A note on terminology: the terms back bias and body bias are used interchangeably. Likewise the terms adaptive and dynamic when used in the FD-SOI context. Here is a quick explanation of how it works, from an ST paper from several years ago:
Back-biasing consists of applying a voltage just under the BOX of target transistors. Doing so changes the electrostatic control of the transistors and shifts their threshold voltage VT, to either get more drive current (hence higher performance) at the expense of increased leakage current (forward back-bias, FBB) or cut leakage current at the expense of reduced performance. While back-bias in planar FD is somewhat similar to body-bias that can be implemented in bulk CMOS technology, it offers a number of key advantages in terms of level and efficiency of the bias that can be applied. Back-biasing can be utilized in a dynamic way, on a block-by-block basis. It can be used to boost performance during the limited periods of time when maximum peak performance is required from that block. It can also be used to cut leakage during the periods of time when limited performance is not an issue. In other words, back-bias offers a new and efficient knob on the speed/power trade-off.
For another good discussion of body biasing in FD-SOI, you might want to check out The Return Of Body Biasing by Semiconductor Engineering’s Ann Steffora Mutschler from a couple years ago.
Dolphin Integration, a partner in the ENIAC THINGS2DO European FD-SOI project, showcased its achievements with PowerStudio™ during the project final review. Power Studio is Dolphin’s cutting-edge EDA tool for safe Power Regulation Networks implementation.
THINGS2DO, which stands for THIN but Great Silicon to Design Objects, was a 4-year, >€120 million EU project (85% industry-funded) with over 40 partners that just finished up at the end of 2017. The goal was to build a design & development ecosystem for FD-SOI. The project funded and supported the development of major FD SOI-based IPs and ASICs as well as EDA tools. (Another recent THINGS2DO announcement was Dream Chips’ ADAS SoC fabbed in GlobalFoundries’ 22FDX technology — read about that here.)
“Being involved in the THINGS2DO project was an opportunity for Dolphin Integration to start introducing FD-SOI in its automatic design methodologies,” said Frederic Poullet, Dolphin Integration’s CTO (read the press release here). “Dolphin Integration plans to offer a full suite of tools allowing its customers to implement right-on-first-pass Power Regulation Networks.”
The company notes that THINGS2DO also proved that low power consumption makes FD-SOI a perfect fit for IoT and automotive applications. For instance, dynamic control of threshold voltage can be used to compensate for temperature variations, and to drive speed improvements by 200% in ultra-low voltage applications.
Dolphin Integration provides energy efficient IPs and ASIC services dedicated to the low-power application market and supports its internal teams with tailor-made software tools. To address the specific needs of its customers in low-power design, Dolphin developed PowerStudio™, a global solution for the optimization of Power Regulation Networks (PRNet) to be used at an early stage of the SoC design process. In particular, it addresses new design challenges in noise and power supply integrity.
The first module of PowerStudio™ will also embed architecture optimization features at the schematic level, in terms of FoM-based cost optimization, mode management, margin cuts and integrability rate-based risk optimization.
Btw, Dolphin Integration Director Frederic Renoux gave an excellent great presentation at an SOI Consortium event in Nanjing, China last year, entitled Embedding power regulation & activity control networks for best SoC PPA.
Dolphin Integration joined Global foundries’ FDXcelerator™ Program last year (read the press release here) to streamline design in 22FDX®. “Our comprehensive and robust library of voltage regulators, power gating cells and logic modules, enables to deal cost-effectively and securely with power distribution, power gating, power monitoring and power control of any SoC design in 22FDX,” Michel Depeyrot, Dolphin Integration’s Chairman, said at the time. “As connected devices sleep most of their time, users of 22FDX also benefit from our ultra-low power and accurate oscillators to design an always-on RTC which consumes as little as 60 nA.”