SemiWiki – Page 588 – The Open Forum for Semiconductor Professionals

January 11, 2017March 22, 2022

Tool Trends Highlight an Industry Trend for AMS designs

Tool Trends Highlight an Industry Trend for AMS designs
by Bernard Murphy on 01-11-2017 at 7:00 am
Categories: Automotive, Cliosoft, EDA

Archaeologists often use tools found in digs as a major indicator of trends in civilizations. The same could be said for trends in design, though we don’t have to reconstruct these design trends – we tend to see them ahead of us.

The trend in this case is the growing importance of sensors in designs and there’s no better example of that trend than in the evolution of smart cars. The use of sensor technology has now become an integral part of the nervous system of many automobiles. Almost all of these smart cars depend on sensing behavior and surroundings and have found application in demanding areas such as dashboards, brakes, vehicular response, increased safety, comfort etc. One common thread in all those systems is the usage of analog/mixed signal designs (AMS), often with a mechanical or micro-mechanical front-end.

The usage of AMS is what is driving this important tool trend. AMS designs typically have few overlaps with tools used in digital design but one area where design needs are starting to overlap is in design data management (DDM). Historically, small AMS design teams building more or less from scratch would see little value in DDM. But now even these small co-located design teams can see the merits of using design management for AMS designs. For medium to large design companies, with AMS designs becoming more complex and distributed, design flows are growing more involved, ignoring DM solutions is no longer an option. With analog design modules being integrated with digital and RF components, the need to have an underlying data management solution is inevitable. In the automotive space, quality and reliability is an exceedingly important criterion. With the sensors expected to last the life span of the vehicle, it becomes important for companies to know which revision of designs was used for the sensor along with the fixes made.

One can always argue that the data management for the design of sensors and SoCs could be done by using the traditional tools for digital SoCs. But the needs of the AMS designs, are quite different from digital designs. Digital design data prior to implementation is textual, and consequently rely on the same kinds of DDM tools used in software development. However AMS design data is done using schematics and layout in binary format. This has several consequences, which does not fit well with conventional software based DDM:

Since AMS designs are saved as binaries, branching and merging versions are very complex operations. Comparing two schematics or layout views require specialized tools which cannot be done by software based DDM often used for digital design as well. To ensure parallel development of designs, it becomes necessary to incorporate strict locking mechanisms. Checkout must lock an object for further checkouts, even by the designer who did the current checkout (until they have checked it back in).
Hierarchy and view dependencies are trickier. A cell view has multiple files, which need to be checked in and out as an atomic unit and hierarchical dependencies to other design objects (in other files) can’t be discovered easily without in-depth understanding of the implementation tool vendor’s database. This point alone explains why conventional DDM tools are probably never going to support AMS design.
The binary representation of design objects also requires more space on the file system than text based data. Since every user has his own personal copy of the project state this data must be stored and moved frequently, which creates additional costs and may become a performance bottle-neck for remote sites. A DDM should consider mechanisms that reduce the amount of traffic with remote sites.

But despite these restrictions, AMS designers still need the capabilities that digital designers expect: checkin and checkout control, ability to work with a local copy (isolating you from in-process changes in other areas), tagging at checkpoints and so on. An important factor for user acceptance is how well the DDM is integrated with the design tools. Unlike software developers, who are used to working with their favorite text editor command line tools, analog/mixed-signal designers work in a GUI environment and are expecting (a) a graphical visualization and (b) not having to switch between tools to handle and modify the design data.

ClioSoft’s SOS design management platform is one solution designed to meet these expanding needs. With over 200 customers developing a wide spectrum of SoCs, their robust solution is easily customizable to meet the requirements of any company. Moreover, with partnerships with all major EDA vendors, ClioSoft’s SOS is the only solution which can manage the design data for all types of designs – analog, digital and RF. What’s telling is their adoption stats, particularly in Europe which is the center of gravity for a pretty significant percentage of automotive electronic development. Among others, SOS has already been adopted in companies such as Infineon, Elmos, Creative Chip and Micronas.

Elmos recently drafted a very informative white paper on why they chose ClioSoft SOS for their DDM solution. They highlighted support for local cache servers as an important differentiator, allowing for very responsive access to all files being used at that site without requiring huge amounts of disk space. The white paper also has some pretty interesting discussion on how they managed co-development of analog and digital design in the same hierarchy, where the digital design is managed by SubVersion. They also talk about using SOS to manage tool configuration across all users and to manage tapeout.

Perhaps this is all ho-hum to digital designers but is has become essential in the world of sensor design and that’s driving a pretty significant shift in what is important for AMS design data management. You can request the Elmos white paper HERE.

Also Read

Managing International Design Collaboration

Making your AMS Simulators Faster (webinar)

3 small-team design productivity challenges managed

January 11, 2017January 25, 2021

Netspeed Gemini NoC Provides Coherent Fabric in Mobileye’s Next-generation EyeQ5 SoC

Netspeed Gemini NoC Provides Coherent Fabric in Mobileye’s Next-generation EyeQ5 SoC
by Mitch Heins on 01-11-2017 at 7:00 am
Categories: IP

Last week I wrote about NetSpeed’s network on chip (NoC) IP technology and design environment NocStudio. This week we see a real life application of this technology announced at CES by Imagination Technologies and NetSpeed. The companies have announced that Mobileye will use Imagination and NetSpeed IP in their next-generation ADAS and autonomous driving system-on-a-chip (SoC).

Mobileye is well known for its vision accelerators for deep-layered neural networks and they have plans to use Imagination’s I6500 MIPS CPUs along with NetSpeed’s Gemini NoC IP in their next-generation SoC dubbed EyeQ5®. Autonomous driving is very compute intensive as it must deal with a myriad of simultaneous inputs to make complex real-time decisions. In 2015, Audi used a MIPs-based Mobileye SoC to complete a fully autonomous 560-mile drive from San Francisco to Bakersfield. The SoC used was the Mobileye EyeQ4®.

In this week’s announcement, we get a glimpse of Mobileye’s next-generation SoC, the EyeQ5®. The EyeQ5® is projected to be 8X faster than its predecessor the EyeQ4® and it is expected to produce more than 15 tera operations per second while consuming less than 5W of power. To do this the EyeQ5® will make use of a complex heterogeneous multi-core SoC that utilizes 8 configurable MIPS I6500 CPUs (the EyeQ4® used 4 MIPs CPUs) coherently combined with 18 of Mobileye’s next generation vision processors (the EyeQ4® used 6). As part of the heterogeneous I6500 clusters, NetSpeed’s Gemini NoC provides the fabric that lets Mobileye’s engineers coherently mix on-chip configurations of processing clusters for high system efficiency.

The ability to configure every component of the interconnect in a coherent heterogeneous environment is a requirement for ADAS applications. Mobileye’s designers will be able to use NetSpeed’s NocStudio software with integrated machine learning capabilities to accurately model and simulate their system configurations to optimize for the best performance, power and silicon area trade-offs and then produce fully synthesis-ready RTL for SoC implementation.

The combination of Imagination’s highly scalable MIPS I6500 CPUs with NetSpeed’s deadlock-free coherent NoC fabric enables designers to implement optimized configurations of CPU cores or clusters of CPUs. In a single cluster, designers can optimize power consumption and configure each CPU with different combinations of threads, different cache sizes, different frequencies, and even different voltage levels all while being cache coherent.

Mobileye’s use of IP from Imagination and NetSpeed IP is a valuable feather in the caps of both these IP providers as Mobileye is known to be a pioneer of heterogeneous SoC designs and they know how hard it is to get it right, especially in a coherent environment. Their SoCs are used by a majority of the world’s automakers including Audi, BMW, Fiat, Ford, General Motors, Honda, Nissan, Peugeot, Citroen, Renault, Volvo and Tesla. Mobileye’s use of these IPs is a testament to the strength of the offerings from Imagination and NetSpeed.

With 8X the computational performance of EyeQ4®, it’s easy to imagine that the EyeQ5® will take on even more data fusion than its predecessor which already simultaneously accepted inputs from 8 cameras as well as information from multiple radars and scanning beam lasers. The real power of the Imagination-NetSpeed IP collaboration, however, is that it enables Mobileye’s designers to be able to tune the additional MIPS clusters to take on more tasks while simultaneously optimizing the overall system for power, performance and cost. This could give the Mobileye team the ability to quickly configure and synthesize multiple versions of the EyeQ5® SoC architecture for different automotive markets which in turn could give them the opportunity to broaden their footprint in the automotive space and possibly take on more of the electronics functionality than just the image processing.

January 10, 2017September 30, 2020

Analog Bits and TSMC!

Analog Bits and TSMC!
by Daniel Nenni on 01-10-2017 at 12:00 pm
Categories: Analog Bits, FinFET, IP, TSMC

As a long time semiconductor IP professional I can tell you for a fact that it is one of the most challenging segments of semiconductor design. Given the growing criticality of semiconductor IP, the challenges of being a leading edge IP provider are increasing and may be at a breaking point. The question now is: What does it take to be a successful leading edge semiconductor IP company?

First and foremost, you must have a high tolerance for pain! Not only do you compete with other IP companies, big and small, you compete with internally developed IP which is like selling shoes to a shoemaker.

Second, you have to have a VERY close “silicon proven” relationship with the foundries. All was well in the Semiconductor IP business until FinFETs came about. Not only are FinFETs a significant design challenge requiring early access to leading edge processes, the foundries have locked down that early access. Do you remember back at 28nm and above when the foundry processes were all “T like”? IP companies developed products at TSMC and ported them to UMC, SMIC, and Chartered making it much easier to scale your IP development, right? That portability is now gone with FinFETs and as we move down the process path to 7nm and 5nm the design challenges and security restrictions are growing rapidly, absolutely.

Third, your business model had better be mean and lean with the ability to pivot at a moment’s notice. The good news is that silicon proven commercial IP is much more attractive now that design cycles are tight, the tightest I have ever seen actually. I am also seeing more systems companies making their own chips using commercial IP. Then there is the semiconductor company consolidation which is a double edged sword. It is good news if your customer takes over your competitor’s customer and not so good news if it is the other way around. So you had better be nimble, you had better be quick, and that brings us to the poster child for a successful leading edge IP company: Analog Bits.

Founded in 1995 here in Silicon Valley, Analog Bits has zero external funding and has enabled billions of chips from .25 micron down to FinFETs via more than 350 customers worldwide and more than 70 unique processes. They are experts (1st time right) at low power mixed signal IP and a pioneer in Multi-Protocal SERDES. Analog Bits is also an ardent TSMC supporter (which is where I know them from) and a member of the exclusive “TSMC Partner of the Year” club.

In fact, Analog Bits presented twice at the previous TSMC OIP Ecosystem Forum. The first presentation was Silicon-proven, low power IP for TSMC 16nm FFC for Automotive to Datacenter SOC’s and the second was Design and Verification of 16nm FFC Low Power SERDES for Datacenter and Automotive Applications. The theme of course is leading edge SoC design for two of the hottest semiconductor vertical markets. If you click on the links it will take you to the abstracts on the TSMC site. To see the full presentation you will need to have a TSMC account or you can contact Analog Bits and talk to them directly.

You can hear a bit more about Analog Bits in the recruitment video below. They are hiring big time:

January 10, 2017June 14, 2019

Industrial IoT (IIoT) – Beyond Silicon Valley

Industrial IoT (IIoT) – Beyond Silicon Valley
by Brian Derrick on 01-10-2017 at 7:00 am
Categories: EDA, IoT, Siemens EDA

Industry 4.0, Smart Factory 1.0, and Internet of Manufacturing are industry initiatives aimed at accelerating the Industrial IoT. With current market forecast exceeding $40 billion and projected to approach $100 billion by 2020, IIoT has everyone’s attention. Well, almost everyone. Turning volumes of factory data into actionable information from the supply chain, to the floor, to operations, and up to management, and potentially to customers, is the key challenge of Industrial IoT deployment. IIoT has evolved just like we saw the integration of the back office, front office, and business intelligence evolve – point-to-point custom solutions built over decades.

Originally, equipment was connected to local Supervisory Control and Data Acquisition (SCADA) systems and clever plant managers discovered ways to use this data to manage the shop floor more effectively. As manufacturing became more complex, specialized software was developed to support a class of manufacturing management applications that allowed optimization across multiple lines, shop floors, and other locations connected to networks. With advances in sensor technology, networking communication, and computation capability, the IoT is accelerating with wild forecasted values of economic returns. Intelligent factories are one of the largest areas of return for IoT and IIoT. With hundreds of thousands of sensors already deployed, factories connected to the Internet, and suppliers and customers already communicating electronically, the industry is a great starting point for IIoT. But, unlike the software and Internet evolution, the IIoT center of excellence is not rooted in Silicon Valley.

Market Observations
With a market worth billions of dollars in the near future, the big industrial automation companies are already heavily invested in bringing industrial solutions to the market. According to an October 2016 Semicast Research report, General Electric, Siemens, and United Technologies are the largest IIoT OEMs (by revenue) but combined with 15 top OEMs, they all capture only 1/3 of the available market:

Taking a look at the headquarter locations of the top 15 IIoT industrial automation companies shows that the only company in Silicon Valley is Applied Materials, while the rest are mostly on the east coast of the US, and in Europe and Japan:

Industrial software that companies employ to control the plant or factory includes distributed control systems, human-machine interfaces, and SCADA infrastructure. The top industrial software companies (by revenue) are:

Large industrial automation companies are not the only entities chasing the IIoT market. Having long served the industrial automation and monitoring markets, sensor and actuator providers are positioned on the front lines of IIoT. As more computing and value moves to the edge of the IIoT, these critical sensor and actuator companies will become even more vital to realizing the intelligent factory. Here, Silicon Valley is represented by HP and Avago. The top sensor and actuator providers (by revenue) are:

Top cloud providers, like Amazon and Google, are highly-recognized companies that provide infrastructure, business processes, and application services to the industry. Here, Silicon Valley is represented by Google and VMWare. The top cloud services providers (by revenue) are:

Leveraging the Cloud and rapidly expanding internet connectivity, Siemens MindSphere Cloud for Industry allows for improved asset management and energy efficiency through data analysis and simulation by collecting and analyzing large volumes of factory data. Similarly, the GE Predix software platform also connects industrial equipment, analyzes data, and delivers real-time insights.

The IIoT is similar to its more highly-recognizable sibling IoT, in that the whole solution is based on connectivity. Long before the IoT, machines were connecting and communicating with each other. With the proliferation of WiFi, gateways have become a critical component of the IIoT. In fact, many smart sensors are gateways themselves. Here, Silicon Valley is represented by Cisco. The top IoT and intelligent gateway providers (by revenue):

Machine to Machine (M2M) communication (or sometimes Man to Machine) in factories is accomplished using wireless and cellular modules and terminals. M2M technology allows remote measurement, diagnostics, maintenance, monitoring, and reporting from the factory floor to a large audience within a company. The top M2M communication hardware companies (by revenue) are:

All the industrial giants in IIoT not only have relationships with every manufacturing company on the planet, they possess domain knowledge. Capturing, analyzing, and making decisions in real time from jet engines, offshore oil and gas rigs, or manufacturing plants around the world is vastly more difficult than analyzing ecommerce transactional data or tracking social media posts. This domain knowledge is captured over time, it is very different from one domain to another, and it is extremely valuable for analyzing business and making decisions.

Silicon Valley has its eye on the IIoT and companies there are not known to sit on the sidelines. Startups in Silicon Valley and around the world are already actively providing solutions:

The “big guys” are responding to the start-up pressure. For example, earlier this year Siemens set up a separate business unit called “Next47” to foster disruptive ideas more vigorously and to accelerate the development of new technologies. And, GE launched their “Fastworks” initiative to experiment rapidly like a startup and to discard ideas that do not gain traction.

The IIoT Payoff
It is clear that factory automation delivers products to market faster and cheaper, but companies need to make IIoT decisions based on how fast the return on investment (ROI) will take place. One way to do this is to look at public case studies in similar industries. For example:

A yogurt company implemented a fully-automated production line that increased capacity by 300 percent, lowered costs by 30 percent, and decreased lost batches of product by 95 percent.
An oil refinery installed wireless acoustic sensors and gas flow valves in flare stacks. This system paid for itself in five months with an ROI of 271 percent annualized over 20 years. Plus, the solution saved them over $3 million in hydrocarbon emission losses per year by quickly detecting and repairing faulty valves.
A company that makes sanitizing gel reduced production cost by 50 percent and achieved ROI in six weeks by adding a data analytics solution to their connected production machines.
A semiconductor foundry automated their customer quote system for custom ICs and achieved a 419% ROI in 3 months and achieved over $4 million in savings per year.

Looking Forward
The world of manufacturing is merging information technology, advanced IoT sensors, and analytics to create smart factories. Some companies are looking for fully-autonomous factories. Others see factory machines and manufacturing lines that automatically assemble themselves where customer orders turn into production data that drives the machine configurations. Others are exploring virtual reality, 3D printing, robot technology, and clever ways to exploit cheap sensor technology connected to the Internet. Manufacturing companies are on the leading edge of technology because small improvements to their systems add up to large savings over time. Consider this remarkable statistic provided by the oil industry: improving the productivity of existing oil well assets with IIoT solutions by 1 percent would increase the world’s output of oil by 80 billion barrels, which is the equivalent of three years of the global oil supply. Do you think that oil companies are motivated to join the IIoT evolution?

There is a blurry line between IoT and IIoT. Factory automation, commercial buildings, and in some cases healthcare fits nicely into the definition of industrial. Automotive manufacturing is an obvious industrial market segment with huge potential and when smart cars are deployed into smart cities, the transportation market looks more like an IIoT solution. The markets, applications, and opportunities seem endless.

As the larger systems of systems world begins to adopt design automation solutions and standards that semiconductor and electronic systems companies in Silicon Valley have employed for decades, the opportunity for design automation companies to create innovative solutions for new markets emerges.

January 9, 2017

CES: Carnival Corp Personifies Key to Monetizing IoT

CES: Carnival Corp Personifies Key to Monetizing IoT
by Mitch Heins on 01-09-2017 at 12:00 pm
Categories: Events, IoT

When one thinks of CES, one typically thinks of the latest in virtual reality or huge super high resolution televisions, sophisticated drones and robots. However, what caught my eye this year came from a company you don’t typically associate with high tech gadgets and that was Carnival Corporation. Yep, the company with all of the cruise lines. So what is the connection between Carnival and high tech?

Carnival is in the business of selling personalized vacation experiences that mostly happen on large cruise ships all over the world. They aren’t however selling a cruise, they are selling an ‘experience’ and there is a big difference. An experience implies something to which the consumer is emotionally connected before, during and after the event. In Carnival’s case, a ‘gadget’ is involved, but it goes much deeper than just a neat toy and it may very well be the key to how we all need to think when it comes to how we monetize the Internet of Things (IoT).

In a nut shell, what Carnival has done is to create a wearable device known as the OCEAN Medallion. The medallion, as its name infers, is a small waterproof device that can be worn as a pendant or watch that is personalized for each individual customer. The medallion makes use of both BLE (Bluetooth Low Energy) and NFC (Near Field Communication) technology to talk to thousands of sensors and portal devices that are placed throughout the cruise ship and port-of-call locations. A key difference however is that Carnival flips the typical BLE paradigm of fixed beacons and mobile readers around to have the medallions act as mobile beacons talking to fixed readers.

The term OCEAN in OCEAN Medallion is actually an acronym for One Cruise Experience Access Network. As the name implies, OCEAN is actually a vast network of devices all working together to help Carnival deliver a more memorable ‘experience’ to their customers. The medallion is used for multiple purposes including as an ID card that enables passengers to gain access to their cabin, a form of payment for on-board purchases in shops, restaurants, spas and gambling casinos, and as a location device that can be used with interactive portals to guide passengers around the ship and to track the locations of other members of a passenger’s traveling party.

So far this is all pretty standard stuff however, Carnival goes a step further. Carnival starts with a program they call OCEAN Ready. OCEAN Ready starts by shipping the medallion to the passengers before the cruise and it enables the passenger to update a profile which becomes known as a passenger genome. A passenger genome captures an individual passenger’s likes and preferences. Carnival uses these profiles in combination with advanced algorithms in an effort to anticipate what might make a passenger’s experience richer and more memorable. What makes the system unique is that the passenger genomes are not static. OCEAN actually learns on-the-fly by observing passengers’ behaviors and choices during the cruise and continues to use new information in real-time to anticipate what might make each passenger’s experience better.

The user interface for this is an application known as COMPASS. Passengers interact with COMPASS through interactive portals that are distributed throughout the ship including the TV in the passenger’s cabin. There are no buttons to push. Instead, portals are activated by the mere fact that the passenger approaches it with their medallion. Passengers can also interact with COMPASS using their smart phones. Ship’s crew also carry portable COMPASS portals that interact directly with passenger’s medallions letting the crew know with whom they are speaking and what interests that passenger may have.

Carnival combines COMPASS and the medallions with other apps like OCEAN Concierge. Concierge enables passengers to order food, drinks or schedule activities through a COMPASS portal. The crew finds the passenger (again using the medallion) to deliver the requested items and the system learns more about the passenger as it attempts to anticipate their next need or desire. The Concierge app also makes suggestions and sends passengers invitations to events and activities on-board and ashore based on preferences indicated in the passenger’s genome.

All of this goes to what Carnival calls ‘Experience Intelligence’. OCEAN is a learning network that customizes the experience for each individual passenger and this is what really caught my eye. Carnival didn’t set out to make another gadget. Instead they started out with the idea of customer centricity and trying to discover how to personalize and customize every passenger’s experience. They looked at the problem holistically and developed a system of hardware and software that would be simple to use but provide a way to truly exceed passengers’ expectations and provide an experience that would be remembered well after the cruise was over.

In the end, Carnival will buy a lot of ‘gadgets’ to make the OCEAN Medallion program work but the payback will be passengers who will each get a customized and personalized experience that they will not soon forget. That hopefully, will translate into repeat customers and more word of mouth references for Carnival that will encourage more people to cruise on their ships.

The lesson to be learned by all of us is that for IoT to be successful, it’s all about having in mind the ‘experience’ with which we want to leave our customers. All of the gadgets in between are a means to an end, not the end itself and Carnival has figured that out.

January 9, 2017

NVIDIA on a Tear at CES

NVIDIA on a Tear at CES
by Bernard Murphy on 01-09-2017 at 7:00 am
Categories: Automotive

Jen-Hsun Huang, CEO of NVIDIA, gave the opening keynote at CES this year. That’s hardly surprising. From a company that operated on the fringes of mainstream awareness (those guys that do gamer graphics), they finished 2016 as the top-performing company in the S&P 500, returning revenue growth of 35% (forecast). That’s startup growth and the same rate at which Amazon Web Services (the mighty Amazon cloud) is growing. Pretty impressive for a semiconductor company. And they are earning it. From the keynote alone, it’s obvious they are putting the same blistering level of innovation into their products that you’ll see at any of the FANG (Facebook, Amazon, Netflix, Google/Alphabet) companies.

Jen-Hsun kicked off with the PC gaming sector which remains very important to NVIDIA. This business has doubled in the last 5 years to $31B and NVIDIA provides the dominant game platform today, as represented by GeForce. They’re obviously very proud of this but they’re looking to how they can grow it further. There are a few hundred million serious gamers today, but most PC/Mac users (around a billion) play games at some level, but can’t access the more advanced games and multi-player options because they don’t have the hardware. So NVIDIA has put GeForce for gaming in the cloud, called GeForce NOW, making it accessible to all users with an Internet connection. This apparently took some serious work to preserve the performance and low latency you expect in desktop gaming. Access will be available in March for early users and will be offered on-demand at $25 for 20 hours of play. Now you see part of why these guys are doing well – they’re expanding their market to casual users, from whom they’ll make money, and at least some of those casual users will like it so much they invest in their own GeForce-enabled desktop systems. 🙂

NVIDIA has also partnered with Google on Shield, their Android-based streaming device (same concept as Roku, AppleTV, etc). It serves up all the usual options – Netflix, Hulu and (unlike AppleTV) Amazon video and, of course, gaming – games can stream from their GeForce systems to the TV or from GeForce NOW in the cloud. More interestingly (for me, I’m not much of a gamer), Shield is tying in Google Assistant, providing natural speech-control of the TV but also home automation, so you have a central hub for voice-activated (including TV) control of any smart home device. To make this a through-house ambient capability they also are introducing the NIVDIA Spot, a small AI microphone (with lots of cool tech) which plugs into a wall socket and communicates with the hub, so from anywhere in the house you can say “OK Google …” and have the Google Assistant respond. (I have to believe NVIDIA is talking with Amazon about Echo integration, though that didn’t come up in the keynote.) Shield starts at $199 and SPOTs are separately priced ($50 each I hear).

Then of course there’s NVIDIA’s role in the automotive industry, which is already significant. This isn’t just about graphics, it’s also in a very big way about AI. Jen-Hsun makes the point that GPUs were a big part of what transformed AI from an academic backwater into a major industry, especially in deep learning. He calls GPUs the “big bang” of AI. Maybe I’d be more of a geek and call it the “Cambrian Explosion” (there was AI around before GPUs, it was just evolving slowly). Either way, NVIDIA saw this opportunity and ran with it – their solutions are a dominant platform in this field.

At the show, Jen-Hsun introduced Their Xavier AI Car Supercomputer – an 8 core ARM64 CPU, a 512 core Volta GPU, the board fuses sensor information, connects to CANs and to HD maps, is designed to ASIL-D and delivers 30 Tops in 30W. NVIDIA created a car they call BB8 (for Star Wars fans) which can drive autonomously given voice directions. The example they showed was “Take me to Starbucks in San Mateo”, from which it figured out the best direction and headed out. Interestingly, they see this more as a co-pilot (they call it AI CoPilot) than a fully autonomous intelligence – BB8 hands over control to the driver whenever it gets to situations it feels it can’t handle.

It also pays attention to the driver, looking for tiredness, inattention, perhaps having had a few too many drinks, and can warn the driver (or possibly take corrective action?). Even more interesting, it does this through facial recognition on the driver and gaze tracking. It also does lip-reading with 95% accuracy (they claim), much better than human experts. Why? Because cars can be noisy environments (music, traffic, passengers), so you want to pay special attention to driver commands, even when voice can’t get through.

Finally, Jen-Hsun announced new automotive partnerships. They have added ZF as partner (5[SUP]th[/SUP] largest automotive electronics supplier), and Bosch (#1 tech supplier to the car industry) has announced a production drive computer partnered with NVIDIA. And they have announced a new partnership with Audi (my favorite car) to build a next generation AI car by 2020. In fact, Audi was demoing a Q7 driving itself in a parking lot at CES after just 4 days of training. All of which reinforced that cars are still in many ways our favorite consumer devices, which is why CES is becoming as much of a car show as an electronics show.

There’s a lot of detail I skipped here, such as Shield supporting 4K and HDR. You’ll have to watch the video HERE to get the full keynote. I was really impressed. This is a semiconductor company that has reinvented itself to play right alongside the consumer technology leaders of today, not just as an “NVIDIA inside” but in many cases as a very visible part of our consumer experience. Other semis should take note. NVIDIA has shown that there is still a path to greatness in hardware.

Intelligent Vision in (almost) Every Application

Intelligent Vision in (almost) Every Application
by Eric Esteve on 01-06-2017 at 12:00 pm
Categories: Ceva, IP

Let’s take a look at the tremendous penetration of intelligent vision in so many and various applications. A few years ago, computer vision algorithms were implemented in applications directly linked with imaging, like computational photography for smartphones and cameras. We can mention today a bunch of segments like automotive, human machine interface or machine vision where computer vision is now the backbone of applications which have been created, thanks to the capabilities of the imaging technology.

CEVA has launched the 5[SUP]th[/SUP] generation architecture for imaging and computer vision, the CEVA-XM6 DSP, and offer a comprehensive vision platform built around the DSP. In a previous blog, we have explained how to build machine learning device implementing the Convolutional Deep Neural Network (CDNN) from CEVA. But let’s take a look at CEVA-XM6 platform, which is much more than a DSP core as it includes the CDNN toolkit comprised of hardware accelerators, neural network, software framework, software libraries, and a set of algorithms.

Automotive driver assistance systems (ADAS) is the most prominent example illustrating the penetration of computer vision completely shaking an automotive segment in which the electronic innovation was in a quiet mode. Now you will find DSP-based imaging in applications like traffic sign detection, free space, pedestrian detection, lane departure, forward collision warning and probably more. Why did it took so long for these types of application to be adopted in automotive? The answer is as usual linked with cost, performance (per dollar) and power consumption.

If we take a look at the CEVA-XM6 DSP architecture we can list the (four) scalar processors SPU0 to SPU3 and the three 512-bit vector processing units VPU-0 to VPU-2, all of which 128 single-cycle 16×16-bit MACs, bringing the total MAC count to 640. In fact, the most important enhancement may be the neural-network hardware accelerator (HWA) that offers 512 additional single-cycle 16×16 bits MACs, connecting to the DSP core’s processing cluster through an AXI4 interface. This HWA is one of the User-defined Coprocessors located in the bottom right box labelled TCE. Taking the example of the CDNN based machine learning, the convolutional layers consuming most neural processor cycles are implemented in the HWA, freeing the DSP core and providing a boost to the machine learning function. When the CEVA-XM6 DSP solution is implemented in a 16-nm chip, it offers unbeatable performance going with very decent power consumption and low footprint.

This performance/power efficiency, coupled with a reasonable chip price, is making CDNN based machine learning an affordable technology to be implemented in mass-market application today. A few years ago, such technology was only demonstrated in a lab, not implemented into a piece of silicon available at mass-market price.

The development of computer vision in Human Machine Interface applications is also opening new possibilities and new markets. The CEVA-XM6 DSP can be integrated to support gesture recognition, emotion sensing, eye tracking, face recognition or face detection. These applications are often linked with the need to provide more security in a world becoming more interconnected, not only thanks to faster communication but also due to higher flow of human moving across the planet. No doubt that these new markets will need more efficient algorithms and higher performance to develop the computer vision based applications increasing safety and security.

Deep learning and augmented reality are two segments, directly linked with machine vision, literally exploding and expected to generate innovation in the industry as well as in our future day to day life. Both are very demanding in term of raw performance and algorithm efficiency. Because the CEVA-XM6 platform is coming with imaging & vision SW libraries and CDNN network generator, it will help the developers to fasten their system time-to-market. CDNN support a variety of popular CNN technologies, including AlexNet and GoogleNet and CEVA offers software libraries for OpenCV, OpenVX and OpenCL support.

If you are interested by CDNN and deep learning solutions for ADAS applications, you should attend to this on demand webinar and download this product note:

On demand webinar “Challenges of Vision Based Autonomous Driving & Facilitation of an Embedded Neural Network Platform”:

http://go.ceva-dsp.com/Nov16-XM6-WebinarOnDemand.html

Learn how to use deep learning solutions for ADAS applications; How to run AdasWorks Free space detection neural network, while utilizing CEVA’s low power vision DSP combined with CEVA Deep Neural Network SW toolkit.

CDNN product note

You will find a complete description of CEVA-XM6 here:
CEVA-XM6 product note

By Eric Esteve

January 6, 2017

CES 2017, Semiconductors and Cycling

CES 2017, Semiconductors and Cycling
by Daniel Payne on 01-06-2017 at 7:00 am
Categories: IoT, Mobile

It’s back, that giant consumer electronics trade show CES 2017, held every January in Las Vegas with too many new product introductions to mention in one blog, so I’ll take a more focused look at what’s new for cycling.

Smart Bike
We all know what a smart phone is, but what could a smart bike be? The Chinese company LeEco has managed to integrate several pieces of separate technology into a single bicycle dubbed the LeEco Smart Road Bike:

Carbon frame road bike
4″ touchscreen (Android 6.0 BikeOS, Snapdragon 410 processor, 6,000mAh batter)
Turn-by-turn navigation
Music playback
Walkie-talkie communication
ANT+ support of heart rate sensor and power meter
On-board lighting
Security alarm
11 speeds
18.5 pounds

LeEco also has a Smart Mountain Bike with similar integrated features as their road bike.

This bike would appeal to someone that falls in love with the price, looks and features of an integrated bike, and is not concerned with big-name bike brands (Trek, Specialized, Fuji, Cannondale).

Another integrated, smart bike is called the SpeedX Unicorn, priced at $3,199 and being marketing on Kickstarter. When you straddle the bike and look at the stem you see an integrated, Android-power bike computer with a 2.2″ display. It looks pretty snazzy, however my Garmin 820 has even more programmable display fields:

Electric Bikes
You can upgrade any existing bike to an electric by getting the Rool’in, it’s an electric front wheelfor your bike and it comes in three sizes.

Swagtron has the SwagCycle Urban E-bike with a top speed of 40mph and range of 55 miles per charge, that’s fast and far for sure.

Exercise Bike
Toddlers need to work out too, right? So Fisher-Price has a Think & Learn Smart Cycle for your toddler that keeps them fit while they play an app on your Apple TV or Android TV.

Monitoring
I’ve never heard of measuring body temperature while cycling, however Bodytrak claims that there new device that fits in your ear can monitor:

Body temperature
Heart rate
Speed
Distance
Cadence

That’s a big claim, and right now I have separate sensors for heart rate, speed and cadence.

Power Meter
I ride with a power meter integrated into my left crank arm, however Leti from France has come up with a power meter in a pedal, called PUSH. The big news is that they plan to only charge $100 or so for this, while their competitors have priced much higher at $500-$1,500 range. Their product photo shows a conventional pedal, not a road bike pedal, so I’m not sure how big their market is going to be. The consumers with the most cash to spend on a power meter are road cyclists like me, or tri-athletes that compete.

Smart Lock
For folks that commute and need to lock up their bike while running errands should be interested in the Ellipse Smart Bike Lock which has a GPS device for location and an accelerometer to detect and report a crash by text to your contacts. The battery is charged through a solar panel, so no need for USB cords and wall charging. There’s an App included, and the price is $199.

Smart Helmet
Cars have turns signals and brake lights, so why not bikes? Well, now your smart bike helmet can have turn signals and brake lights thanks to Livall and their Smart Riding Helmet. You just connect a Bluetooth enabled button-set on your handlebars to control the lighting on the helmet, and then cars approaching from behind know when you are turning or braking.

Another smart helmet called the CLASSON is marketed on Kickstarter for just $99 and has turn signals, brake lights and even alerts you to cars in your blind spot.

Heads Up Display
Solos Cycling has a product that reminds me of Google Glass, except that it’s for cyclists so that they don’t have to glance down at their bike computer mounted on the handlebar or stem. It uses the ANT+ and Bluetooth Smart Sensor protocols so should connect to your sensors for: speed, power, cadence, heart rate.

10 Battery Road Bike
I wanted to update you on what I’m riding these days, it’s a Specialized SL4 frame with SRAM eTapwireless shifting, very cool, no derailleur cables and no electrical cables for shifting. Before a ride I have to ask myself, “Are all 10 batteries ready to go?” Here’s where the 10 batteries come in:
[LIST=1]

Left shifter – CR2032 battery, lasts a few years

Right shifter – CR2032 battery

Garmin 820 – bike computer, rechargeable battery, lasts about 200 miles per charge

Cygolite – front headlight, rechargeable battery, about 10 hours per charge

Garmin speed sensor – CR2032 battery, wireless, on the front hub

Stages Cycling Power meter – CR2032 battery, on the left crank arm, about 150 hours of riding

Front derailleur – SRAM eTap, rechargeable battery, about 1,000 miles per charge

Rear derailleur – SRAM eTap, rechargeable battery, interchangeable with front derailleur battery

Rear light – rechargeable battery, about 10 hours per charge

Heart rate monitor – CR2032 battery, about 1.5 years per battery

I did reach my mileage goal for 2016 of 16,000 miles, which included some 789,000 feet of climbing. Come and follow me on Strava, or better yet, come join me for a bike ride in the Portland, Oregon area.

January 6, 2017

Tesla’s (and Uber’s) Teflon to be Tested in 2017

Tesla’s (and Uber’s) Teflon to be Tested in 2017
by Roger C. Lanctot on 01-06-2017 at 7:00 am
Categories: Automotive

For the past two years the impression has been spreading that Tesla Motors can do no wrong. (I can’t really say the same for Uber after the recent San Francisco licensing debacle.) There is no question that Tesla’s legal department is growing by the month as fights persist over opening stores and forestalling liability judgments, but, so far, even fatal crashes of Tesla vehicles have failed to tarnish the Tesla brand.

This will change in 2017. Tesla is quietly and not so quietly shifting its strategy from bobbing and weaving and legal actions in order to open stores – toward strong-arm, Trump-like muscle flexing.

Sources in the industry indicate that Tesla has begun threatening to make sourcing decisions based on the level of local support for its marketing and sales activities. This kind of influence peddling is not new and is reflected in states modifying their autonomous vehicle laws to attract the likes of Google and Uber and their development dollars and investments. It is also reflected in Nevada’s scoring of two electric vehicle plants in 2016.

Competing car companies can only look on in awe, envy, disgust and anger at the result and the complete lack of a consumer backlash. Were a driver to be killed in a Toyota or a General Motors or an Audi with an autopilot-like system congressional hearings would be called, executives would be humiliated on C-SPAN and fines would be levied.

Tesla has the non-stick Teflon coating of the Silicon Valley startup widely regarded by consumers and legislators alike as the source of national pride and economic growth. Existing car companies are seen as passe impediments to progress, locked in the past and dragging their heels on safety advances – only contributing to the rising toll of highway fatalities.

Tesla has the halo of the innovator, not unlike Silicon Valley neighbors Apple and Google/Alphabet and Uber. We tend to treat these companies with kid gloves because we fear our economic future hinges on their success even if we find ourselves surrendering our privacy and … freedom?

European regulators are far more concerned with privacy and harbor no illusions about halos or innovation. This is why the rough treatment that Apple and Google have received from Brussels seems so odd from a distance. It’s worth noting that concerns for privacy have led to the barring of dashcams in Germany and Austria. The obsession with privacy does have its limits.

In the U.S. the steepest resistance to Tesla has come from state regulators standing in the path of Tesla opening retail stores. State legislators and regulators are vulnerable to the immense lobbying influence of automobile dealers who are tightly woven into their local communities and drive a substantial amount of economic activity including both tax revenues and employment – to say nothing of charitable and political donations.

But taking the economic argument to the next level to leverage production decisions to the advantage of product development has so-far eluded incumbent car makers. Incumbent car makers aren’t leveraging their sourcing decisions for economic advantage. Rather they are scurrying from the glare of the incoming Trump administration which is casting threats far and wide against car makers seeking to build plants in Mexico. The resulting negative impact on Ford, GM and Toyota stocks is manifest.

Perhaps in recognition of the might of the dealer lobby, Tesla has taken the gloves off. Work with us, say CEO Elon Musk’s minions, or we will take our business, our tax dollars, our employment contribution elsewhere. Uber, too, has taken this approach with mixed results. Austin, Tex., said: “No.” to Uber’s preference to not fingerprint its drivers. The state of Maryland said: “Okay.” to Uber’s demand.

Nowhere is Tesla’s threat more potent than Michigan, where Tesla is likely to achieve victory in its drive to open stores in the state in 2017. But the overt Tesla (and Uber) threats being made behind closed doors and, increasingly paralleling Trump’s more public Twitter-based efforts, will test the public’s patience.

Both Tesla and Uber are placing multi-billion dollar bets on transformative transportation technology and business models. In the process jobs are both being created and destroyed. Tesla’s rise puts the entire internal combustion dealer network under threat. Uber is putting the jobs of millions of professional drivers of cars and trucks at risk.

Consumers have so far remained on the sidelines in the struggle – happy to benefit from subsidized cab rides (Uber) and subsidized EVs (Tesla). But these subsidized experiences have a cost (Uber – mistreated passengers, Tesla – fatal crashes) capable of bringing a reckoning in 2017.

Uber and Tesla appeal to our emotions and our pocketbooks. Let’s hope that in the end it isn’t all just a shakedown where we are surrendering both our freedom and our privacy – which represent core consumer value propositions that are carefully curated by the incumbent car makers. Both Tesla and Uber are out to narrow rather than expand our transportation choices. This is a battle where things will get very sticky indeed.

January 5, 2017

2017 Semiconductor Dead Pool

2017 Semiconductor Dead Pool
by Daniel Nenni on 01-05-2017 at 12:00 pm
Categories: EDA, Foundries

In 2015 we saw $85B in semiconductor acquisition activity and in 2016 there was more than $110B. Given 2015 and 2016 were relatively flat years for the $335B semiconductor industry and 2017 looks like more of the same we should expect consolidation to continue, absolutely.

So, let’s come up with a list of companies that may fall in 2017 and circle around at the end of the year to see how we did. I will be more than happy to defend my choices in detail in the comments section.

My first three picks will focus on companies in the growth markets of data center, IoT, and automotive chips. I also look at company leadership (strong or weak, new or old) and if an activist investor is involved that’s a bonus. But first let’s look at who was acquired in 2015, 2016, and who is up for grabs in 2017 (let me know who I missed).

Acquired in 2015:

[LIST=1]

Altera

Atmel

Broadcom

Emulex ISSI

Fairchild

Freescale

Micrel

Omnivision

PMC

Pericom

Richtek

Sand Disk

Silicon Image

Vitesse

Acquired in 2016:

[LIST=1]

Applied Micro

ARM

Brocade

Linear Technology

Mentor Graphics

NXP

Intersil

EZ Chip

Lattice

Qlogic

Invensense

Who’s left? (not exhaustive, just the ones I know)

Ambarella, AMD, Analog Devices, Broadcom, Cavium, Cirrus Logic, Cypress Semi, Dialog, IDT, Inphi, Infineon, Intel, MACOM, Marvell, Maxim, MediaTek, Melexis, Mellanox, Microchip, Micron, Micronas, Microsemi, Novatech, NVIDIA, On Semi, Qorvo, Qualcomm, Realtek, Renasas, Samsung, Semtech, Silicon Labs, Silicon Motion, SK Hynix, Skyworks,Sony Semi, STMicro, Synaptics, Texas Instruments, Toshiba, Xilinx.

The company names in bold are on the SOX Semiconductor Index which is up 36% this year, a big number considering the semiconductor industry on a whole was flat in 2016. Given we are looking at another growth challenged year, here are the first three companies that I feel are best positioned for Acquisition in 2017:

[LIST=1]

Marvell

Microsemi

Cypress Semiconductor

According to Marvell:

Marvell first revolutionized the digital storage industry by moving information at speeds never thought possible. Today, that same breakthrough innovation remains at the heart of the company’s storage, network infrastructure, and wireless connectivity solutions. With leading intellectual property and deep system-level knowledge, Marvell’s semiconductor solutions continue to transform the enterprise, cloud, automotive, industrial, and consumer markets. To learn more, visit: www.marvell.com.

While Marvell has always been viewed as a technology centric company with very controlling management resulting in a historically high executive turnover rate, that all changed in April of 2016 when founder/CEO Sehat Sutardja and his wife Weili Dai were ousted as a result of questionable management practices. Marvell now has a new CEO, executive staff, and board members.

According to Microsemi:

Microsemi Corporation (Nasdaq: MSCC) offers a comprehensive portfolio of semiconductor and system solutions for aerospace & defense, communications, data center and industrial markets. Products include high-performance and radiation-hardened analog mixed-signal integrated circuits, FPGAs, SoCs and ASICs; power management products; timing and synchronization devices and precise time solutions, setting the world’s standard for time; voice processing devices; RF solutions; discrete components; enterprise storage and communication solutions, security technologies and scalable anti-tamper products; Ethernet solutions; Power-over-Ethernet ICs and midspans; as well as custom design capabilities and services. Microsemi is headquartered in Aliso Viejo, California and has approximately 4,800 employees globally. Learn more at www.microsemi.com.

Microsemi has been reported to be looking at sale options after takeover interest from Skyworks. Microsemi is a well-known AMS expert inside the aerospace/defense, industrial markets, and communications including connectivity chips for data centers. The FPGA business is the big swing here for M&A after the $16.7B Intel acquisition of Altera and the recent $1.3B acquisition of Lattice Logic.

According to Cypress:
Founded in 1982, Cypress is the leader in advanced embedded system solutions for the world’s most innovative automotive, industrial, home automation and appliances, consumer electronics and medical products. Cypress’s programmable systems-on-chip, general-purpose microcontrollers, analog ICs, wireless and USB-based connectivity solutions and reliable, high-performance memories help engineers design differentiated products and get them to market first. Cypress is committed to providing customers with the best support and engineering resources on the planet enabling innovators and out-of-the-box thinkers to disrupt markets and create new product categories in record time. To learn more, go to www.cypress.com.

Last year Cypress founder and CEO TJ Rodgers stepped down naming Cypress insider Hassane El-Khoury President, CEO, and a member of the Board of Directors (TJ is no longer on the BoD). In 2015 Cypress cemented itself as a world class memory provider with the $5B Spansion merger and in April 2016 Cypress acquired Broadcom’s IoT business for $550M.