SemiWiki – Page 723 – The Open Forum for Semiconductor Professionals

August 12, 2015September 1, 2024

Why Qualcomm Lost Samsung and Will Get Them Back!

Why Qualcomm Lost Samsung and Will Get Them Back!
by Daniel Nenni on 08-12-2015 at 12:00 pm
Categories: Foundries, Samsung Foundry

2016 will be a banner year for the System on Chip (SoC) industry. For the first time we will have leading edge SoCs (Apple, Qualcomm, Samsung) on the same manufacturing process enabling a true Apples to Apples comparison. Unfortunately, how we got there is being misrepresented by the media and analysts but that is Situation Normal for the semiconductor industry, absolutely.

It all started back in September of 2013 with the release of the Apple A7 SoC inside the Apple iPhone 5s which used the 64-bit ARMv8-A architecture versus the 32-bit ARMv7. A 64-bit CPU inside a smartphone? Surely you must be joking:

“I know there’s a lot of noise because Apple did [64-bit] on their A7. I think they are doing a marketing gimmick. There’s zero benefit a consumer gets from that” said Anand Chandrasekher, senior vice president and chief marketing officer at Qualcomm.

Prior to Qualcomm Mr. Chandrasekher spent his career at Intel including 5 years as Senior Vice President of the now defunct Intel Mobility group making SoCs. His comment was later retracted and Mr. Chandraskher was demoted (lost his CMO title):

“The comments made by Anand Chandrasekher, Qualcomm CMO, about 64-bit computing were inaccurate. The mobile hardware and software ecosystem is already moving in the direction of 64-bit; and, the evolution to 64-bit brings desktop class capabilities and user experiences to mobile, as well as enabling mobile processors and software to run new classes of computing devices.” – Qualcomm

At the time of this announcement Qualcomm was architecting their next 32-Bit SoC which was scrapped immediately in favor of a 64-Bit version. In a rush to market, QCOM had to use the ARM Cortex A57-A53 Big-Little cores in the Snapdragon 808 and 810 chips versus their own custom architecture. As a result the famed Snapdragon SoC, which had previously ruled the mobile industry, lost their most important customer in the number one mobile company Samsung.

A recent report, which is now being repeated by the Parrots of Wall Street, credited Samsung’s need to fill their own fabs as the reason for the switch from Snapdragon to the Samsung Exynos SoC. If you know Samsung (as I do) you will know that they do not work that way. If you know the SoC business (as I do) you will know that this makes no sense whatsoever.

The competing Samsung Exynos SoC was launched in 2010 using the ARM Cortex A8. Remember, QCOMM and Apple both license the ARM instruction set and build their own microarchitectures (Cores). Samsung, Mediatek, and other SoC vendors use off-the-shelf ARM cores. The differences are noticeable in regards to performance and power usage which is why Samsung continued to use the Snapdragon in the majority of their mobile devices up until this year.

Samsung is a fierce competitor both inside and out meaning that even the internal divisions of Samsung compete for business. Bottom line: If the Samsung semiconductor group can make a better SoC, Samsung mobile will use it, and that’s what makes Samsung a market leader. That is the real reason why the new Samsung mobile devices use the 14nm Exynos 7 versus the 20nm Snapdragon 810, it is simply a better chip. I bought a Samsung 6S Edge and have experienced firsthand the superior performance and power usage. Even my 20nm A8 based iPhone 6 significantly trails the Edge. The next QCOMM Snapdragon chip (820) uses a 64-Bit custom core (Hydra) manufactured on the Samsung 14nm LPE process and my guess is that it will again get the Samsung mobile business, absolutely.

The Snapdragon 820, Exynos 7, and the next Apple SoC (A9) will all use Samsung 14nm LPE so we will get to do a head-to-head comparison of the QCOMM and Apple custom architectures for the first time. We will also get to compare custom cores versus the ARM cores used in the Exynos. I will give you my bet on this race in the comments section. I will also tell you the REAL reason why QCOMM switched from longtime partner TSMC to Samsung for 14nm.

Why the comments section you ask? Because you have to register to see comments of course!

Also read: 3 Key Frontiers for Samsung’s Next Mobile SoC

August 12, 2015

The Alphabet Starts With G

The Alphabet Starts With G
by Paul McLellan on 08-12-2015 at 7:00 am
Categories: General

What is the second biggest tech company in the world? If you said Alphabet, you get bonus points. If you have never heard of Alphabet, then perhaps you have heard of Google.

On Monday, Google announced that it was going to reorganize its corporate structure. This would usually provoke a big yawn but this could turn out to be significant. Google is creating a new holding company called Alphabet that Page and Brin will run. One of the subsidiaries is what you might think of as the old Google. It consists of search, search advertising, the datacenters, YouTube and Android. Alphabet will also own Calico, Fiber, Nest, Google Ventures, Google Capital, and Google X. One aspect of the reorganization is that these companies will be run largely independently of Google and, in particular, will have to establish their own brand identities in the way that Nest has done since Google acquired it (and ran it independently). I would be willing to bet that Google Capital will soon have a new name, and probably not Alphabet Capital (nor anything beginning with the letter G).

The other reason for making this change is that Google needs to really focus. Advertising is in transition as millennials cut the cord (and don’t read newspapers). So TV advertising and newspaper advertising is continuing to shift more and more online and still has a deficit in terms of time spent on the medium versus advertising spend (whereas newspapers in particular are the other way around: lots of advertising spend but declining eyeballs). Two huge competitors are also out there in Facebook and WeChat. If you don’t live in China, then WeChat doesn’t even seem that significant but it is simultaneously the Chinese WhatsApp (owned by Facebook, of course), Facebook itself, online portal for purchases and more. Its revenue per user is reckoned to be 7-8X what WhatsApp makes. Anyway, investors have been saying for ages that Google needs to focus on its core and not get distracted by these pie-in-the-sky things like autonomous vehicles or Google Glass. Or the glucose sensing contact lens that was the subject of one of the keynotes at DAC. This restructuring is a way to do both. Google (the search and advertising business) will now be run by Sundar Pichai who has no responsibility for any of these other businesses that will have to sink or swim depending on how successful they become.

In an interview with the Financial Times last year, Larry Page said:Looking forward 100 years from now at the possibilities that are opening up, we could probably solve a lot of the issues we have as humans.

As the FT says: Even Google’s famously far-reaching mission statement, to “organize the world’s information and make it universally accessible and useful”, is not big enough for what he now has in mind. The aim: to use the money that is spouting from its search advertising business to stake out positions in boom industries of the future, from biotech to robotics.

If you read between the lines, I think it is clear the Larry (and probably Sergey too) is pretty much bored with the search business and wants to spend his time focused on some of these other areas, using the money pump from traditional Google to do it. Google has a weird corporate structure with two tiers of shares so that they maintain control and, basically, nobody can stop them doing something like this whether they like it or not. They can only sell their Google stock if they choose to.

Before he died, Steve Jobs used to argue with Larry that Google was doing too much. Larry would push back that Apple was not doing enough:It’s unsatisfying to have all these people, and we have all these billions we should be investing to make people’s lives better. If we just do the same things we did before and don’t do something new, it seems like a crime to me.

Alphabet seems to be the way to do the “something new” parts.

Financial times interview with Larry Page from last year is here. Larry Page’s blog announcing Alphabet is here. Alphabet’s minimal website is here.

August 11, 2015June 14, 2019

Meeting Demand as Fab Capacity is Stretched Again

Meeting Demand as Fab Capacity is Stretched Again
by Tom Simon on 08-11-2015 at 8:00 pm
Categories: EDA, Siemens EDA

Global semiconductor production capacity and its utilization level are key elements of the technology economy. During a panel at DAC in June Mentor Graphics posited that we are entering into a period where leading edge processes will be in high demand and also older nodes are seeing increasing demand due to Internet of Things designs that are relying on low power and low cost silicon. All this could put a squeeze on wafer availability.

Without enough wafer fabrication capacity available, electronic product manufacturers who rely on semiconductor components will fall short on their own revenue targets. On the other hand, foundries need to carefully maximize their utilization to ensure adequate margins and profitability. Semiconductor fabs are notoriously expensive to build and their construction comes with extremely long lead times.

It’s interesting to look at hard data on utilization, but unfortunately the SIA stopped issuing their reports on wafer fab capacity and utilization in October 2012. Nevertheless, looking at their last report which covers up to 2011 a number of interesting things can be observed.

Prior to the 2008 downturn, utilization percentages were hovering in the high 80’s, occasionally reaching a peak of around 90%. During the 2008-2009 recession, they fell as low as 56%. But then quickly recovered. From 2010 to the end of the report period utilization exceeded 90%.

So what happens as capacity reaches its limits? The obvious answer is that it can lead to allocation, and all the unpleasantness that come with that. Of course certain processes will be in greater relative demand than others, leading to bottlenecks on particular nodes. Also, process portability affects how elastic the market will be if customers are able move to second sources.

It turns out that there are things that foundries and even their customers can do to help avoid these problems. Similarly, foundries and their customers, by working together, can also help avoid overcapacity which can be equally problematic. The key to this is accurate forecasting on the part of foundry customers.

At DAC 2015 Mentor hosted a panel discussion on this topic. The panelists were Prasad Subramaniam – Vice President of Design Technology and R&D from eSilicon, Kelvin Low – Senior Director of Foundry Marketing with Samsung and UMC USA Vice President of Business Development Walter Ng. The discussion was moderated by Michael Buehler-Garcia – Senior Director Calibre Design Solutions at Mentor.

Walter Ng made the point that foundries really see allocation causing lost opportunity; in his words “it’s not fun for the foundry.” He added that it is essential that customers work with the foundry to ensure forecasts are accurate. In reality several customers might be competing for the same socket and only one will win, causing the other competing prospective orders to not materialize. This means that customers need to make a strong business case for their product to get access to valuable wafers during a period of allocation.

Samsung’s Kelvin made the point that silicon is not the only issue. Design for manufacturing affects yield and consequently the actual number of chips that can be marketed. Higher yield means fewer wafers are needed for a given number of finished chips. Another real limitation is tester time. Improved test vectors will speed production. All of the panelists agreed that customers play a significant role in enabling foundry capacity.

Kelvin also wants to see more data to back up the forecast numbers that are now being used for IoT chips. My own thought is that a lot of these chips might be on older nodes. This of course comes with a mixed blessing. It’s nice to have demand for older nodes, but if the demand is growing as new products are designed for older nodes, how do foundries fill this demand? Nobody is going make more 8” fabs. Walter Ng asked if it might make sense to actually build 12” fabs at older nodes.

Another consequence of significant new designs on older nodes are the questions raised about retrofitting the older flows for these nodes to add critical features like updated power management strategies. Not just the flow, but existing IP might need to be updated on these popular older nodes.

It seems that all the foundries are moving aggressively to increase supply. Kelvin cited Samsung’s outlay of $15B to build their new fab.

Likewise, Walter said that UMC is banking on increased demand for 28nm, and are investing in increasing their capacity.

Even in the distracting environment at DAC this panel drew quite a crowd, overflowing from the seating area of their booth out into the aisle. This alone indicates the level of interest in the topic of foundry capacity. Mentor did a good job of pulling this panel together, even though it could be argued that this topic is out of their wheel house. Of course only time will tell if the growing markets like IoT actually lead to capacity shortfalls for leading or training nodes. For further reading on why and how the IoT is putting higher demand on older process nodes I suggest this article.

August 11, 2015July 18, 2025

Design For Safety in Automotive Electronics

Design For Safety in Automotive Electronics
by Daniel Payne on 08-11-2015 at 12:00 pm
Categories: Automotive, EDA, Synopsys

Do you remember how auto maker Toyota had to pay a $1.2 billion settlement in 2014 because some of their automotive models experienced sudden, unintended acceleration? That scenario has to be an engineer’s worst nightmare because something was missed during the design and testing of an automotive electronics system that has to meet rigid safety standards. Prevention is always cheaper than a cure, especially when it comes to IC design, so I learned something new this week while watching an archived webinar called, “STMicroelectronics’Experience: Synopsys Logic BIST for Automotive and Safety-Critical Designs.”

A Toyota Camry that crashed in 2010. Source: NY Post

Safety Critical Applications
I already mentioned automotive as a safety-critical application, other industries include: medical devices, aviation, trains, bridges, power plants, etc. Just in the automotive space, stop and think about Advanced Driver Assistance Systems (ADAS) and how electronics control the feature and safety:

Air bags
Anti-lock Brake System
Electronic stability control
Adaptive cruise control
Emergency breaking assist
Blind-spot monitoring
Lane-departure warning
Rear cross-traffic detection
Pedestrian detection
Traffic sing recognition

Safety standards are defined for each industry: ISO 26262 for automotive, ISO 13485 for medical devices, DO-254 for aviation. Self-testing is a best practice for electronic systems to help meet each of these standard requirements. Synopsys recently added a synthesis-based in-system self-test product called Logic BIST (Built-In Self Test), and here’s where it fits into the overall design and test flow:

Logic BIST Flow

The required logic for BIST is automatically added to your gate-level design during logic synthesis, so you don’t have to modify the RTL source code in this approach. The Design Compiler tool meets the timing, area, power and test goals during the synthesis step shown in the first blue box above. The TetraMAX ATPG tool computes the seed and signature used by the logic BIST for self-testing purposes, which is different from the Synopsys manufacturing test flow where TetraMAX generates the test program for ATE and also provides silicon diagnostics capabilities.

Logic BIST adds controllability and observability to the scan flip-flops of your design, shown in grey below while the test logic is shown in blue.

Logic BIST Architecture

PRPG stands for Pseudo-Random Pattern Generation, and this is where stimulus is automatically created for self-testing of your logic. Test values are loaded into the grey flip-flops of you design, then results of your logic design are saved in the MISR (Multi-Input Shift Register) to be compared against a known-good value saved in the Signature, shown in green.

Logic BIST at STMicroelectronics
Cinzia Bartolemmei spoke about how her group is using logic BIST for both power-on test and in-system live test of safety critical cores. Requirements of this logic BIST approach for their designs are:

Small silicon area overhead
LBIST must be modular
Doesn’t require data from chip input pins
Simple to interface
Pass or fail response
Support both stuck-at and transition testing
Trade-off patten count and test coverage
Divide LBIST run into several timing intervals

They’ve been able to meet these requirements on IC design blocks ranging from thousands to millions of gates, and fulfill the automotive safety standards even on designs with multiple synchronous or asynchronous clocks. For a case study Cinzia talked about a macro cell used in automotive with about 120K flip-flops, scan chain length of 100 and two asynchronous clocks:

With this approach the area overhead for all DFT was 3% while LBIST required just 1.6%. For single stuck-at faults a test coverage of 91.76% was achieved using 20K patterns, while LBIST used just 2,300 patterns to reach 90% coverage. On transition faults a test coverage of 86.11% was reached using 20K patterns, and LBIST took just 12,400 patterns to get 85% coverage.

Summary
We live in a complex world where our very lives depend on electronics systems functioning perfectly in order to keep us safe. One method to address safety requirements is through Logic BIST, and companies like STMicroelectronics have used Synopsys tools to make their automotive chips adhere to stringent safety requirements. View the entire 21 minute archived webinar here.

August 11, 2015

What Does Legal Sea Foods Have to Do With EDA?

What Does Legal Sea Foods Have to Do With EDA?
by Paul McLellan on 08-11-2015 at 7:00 am
Categories: EDA, General

When I drive down to Silicon Valley I usually listen to podcasts rather than just listen to the radio. One that I especially like is Russ Robert’s EconTalk, which has an hour-long episode every Monday morning on a wide range of different aspects of Economics. Normally he interviews an economist. He has also interviewed the manager of a car dealership, and other people only tangentially connected to economics. This week it was Roger Berkowitz, the CEO of Legal Sea Foods. If you have been to Boston, even just the airport, you may well have eaten their clam chowder, but they actually have 34 fish restaurants on the East coast. They started just as a wholesale fish market, then went retail too, then a restaurant and gradually grew until they now have the portfolio of different seafood restaurants that they have today.

During the podcast at one point Russ asked Roger what was the difference between running a handful of restaurants versus 34. And how different it would be to run 68. They didn’t spend a lot of time discussing it but obviously there is a huge difference in scale from operating a fish wholesaler with a restaurant next door, to operating a fish operation in Boston that handles all the fish for three dozen restaurants, the furthest afield of which is in Atlanta.

That got me thinking about support in the EDA industry. Customer support in a successful EDA startup company goes through three phases, each of which actually provides poorer support than the previous phase (as seen by the long-term customer who has been there since the beginning) but which is at least scalable to the number of new customers. I think it is obvious that every designer at a Synopsys customer who has a problem with Design Compiler can’t simply call a developer directly, even though that might provide the answer the fastest.

There is actually a zeroth phase, which is when a startup company doesn’t have any customers. As a result, it doesn’t need to provide any support. It is really important for engineering management to realize that this is actually happening. Any startup engineering organization that hasn’t been through it before is completely unaware of what is going to hit them once the immature product gets into the hands of the first real customers who attempt to do some real work with it. They don’t realize that new development is about to grind to a complete halt for an extended period. “God built the world in six days and could rest on the seventh because he had no installed base.”

The first phase of customer support is to do it out of engineering. The bugs being discovered will often be so fundamental that it is hard for the customer to continue to test the product until they are fixed, so they must be fixed fast and new releases got to the customer every day or two at most. By fundamental I mean that the customer library data cannot be read, or the coding style is different from anything seen during development and brings the database to its knees. Adding other people between the customer engineer and the development engineer just reduces the speed of the cycle of finding a problem and fixing it, which means that it reduces the rate at which the product matures.

Eventually the product is mature enough for sales to start to ramp up the number of customers. Mature both in the sense that sales have a chance of selling it and the company has a chance of supporting it. It is no longer possible to support customers directly out of engineering. Best case, no engineering other than customer support would get done. Worst case, there wouldn’t even be enough bandwidth in engineering to do all the support. Engineering needs to focus on its part of the problem, fixing bugs in the code, and somebody else needs to handle creating test cases, seeing if bugs are fixed, getting releases to the customer and so on. That is the job of the application engineers.

During this second phase, a customer’s primary support contact is the application engineer who they work with anyway on a regular basis. But as the company scales further, each application engineer ends up covering too many customers to do anything other than support them. Since their primary function is pre-sales, to help sales close new business, this is a problem. So the third phase of customer support is to add a hotline.

The hotline staff are typically not tool users, they are more akin to 911 dispatchers. Customers hate them since they are not as knowledgeable as they are themselves. Their job is to manage the support process, ensure that the problem is recorded, ensure that it eventually gets fixed, and that the fix gets back to the customer and so on. It is not to fix anything except the most trivial of problems themselves.

At each phase of support, the quality (and knowledge) of the engineer directly interfacing to the customer goes down but the bandwidth of available support increases. Engineering can only directly support a handful of customers themselves. Each AE can only directly support a handful of customers but more AEs can easily be added as sales increase. A hotline can scale to support a huge number of customers 24 hours per day, and it is easy to add more hotline engineers.

Econtalk has an episode every week going back to 2006 including Milton Friedman, Ronald Coase when he was already over 100 (he died recently) and many other famous, and not-so-famous, names. The website is here(or you can download from iTunes too). The Legal Seafood episode is here.

August 10, 2015August 22, 2024

Make American Semiconductor Great Again!

Make American Semiconductor Great Again!
by Daniel Nenni on 08-10-2015 at 4:00 pm
Categories: Foundries, GlobalFoundries, Intel Foundry, Samsung Foundry

As I watched the GOP debate between the top 10 candidates last week I asked myself which one of those men would I pick to help the United States stay competitive in the semiconductor industry. I’m saddened to say that the only candidate even remotely qualified for that conversation in my opinion is Donald Trump. Of course I backed Ross Perot in 1992 so I’m not what you would call a “politically correct” person.

My first political candidate of choice was Ronald Reagan in 1981 mainly because I thought it would be fun to have an actor in charge of our country, and it certainly was. He was also a Captain in the Air Force as was my father which I respected greatly. I remember a sound check prior to a radio address when Reagan made the following Cold War joke that went viral:

“My fellow Americans, I’m pleased to tell you today that I’ve signed legislation that will outlaw Russia forever. We begin bombing in five minutes.”

For the same reason Arnold Swarzenegger was my candidate for California Governor and Clint Eastwood for Mayor of Carmel but I digress…

Right before the debate I read the IC Insights top 20 semiconductor company sales report where Samsung cut Intel’s lead to 16% in the first half of 2015. During one of the many trips I made to South Korea I was told quite clearly by Samsung that their goal was to be the number one semiconductor supplier in the world so this did not surprise me at all. Based on my experience Samsung is a very deterministic company, much more so than Intel, and they have all of the tools necessary to lead the semiconductor industry, absolutely.

Another interesting development is that SK Hynix jumped both Qualcomm and Micron. Other than that the top ten did not change. The next big changes will be the Avago acquisition of Broadcom making them a Singapore based company. Avago already acquired Silicon Valley semiconductor legend LSI Logic so they are gone as well. NXP is acquiring U.S. based Freescale (Motorola) making it the largest European semiconductor company ahead of both ST and Infineon. The other big change to the semiconductor landscape that is not reflected in this chart is the GolbalFoundries acquisition of the IBM Semiconductor. It will be interesting to see what impact that will have on GF’s ranking in the second half of 2015.

Looking back 30 years, the advent of the personal computer brought semiconductors into our homes. The PC industry was controlled by three companies: IBM (system), Intel (semiconductor), and Microsoft (software). Samsung, the largest consumer electronics company, takes that a step further by providing both the systems and semiconductors. On the other side of Intel is Apple who I would argue is the most influential fabless semiconductor company in the world today. Apple of course controls the system, semiconductors, AND software.

Given the influence semiconductors have on modern day life one would think semiconductor design and manufacturing would be an integral part of the coming political platforms. As I said, Trump’s “Make America Great Again!” slogan resonates with me both personally and professionally. Unfortunately this seems to be Ross Perot déjà vu all over again… just my opinion of course.

August 10, 2015

GPS Chronicle: When Phone Met Location

GPS Chronicle: When Phone Met Location
by Majeed Ahmad on 08-10-2015 at 12:00 pm
Categories: Mobile

Benefon, one of the GSM pioneers, was the first handset maker to marry cellular with GPS in response to the European Union’s Mobile Rescue Phone (MORE) project during the mid-1990s. The result of this ambitious effort was the launch of the Benefon Esc! phone in late 1999 and Benefon Track device in 2000.

The Esc! phone was splash-proof and featured a large, grayscale LCD. It allowed users to load maps onto the phone to trace their position and movement, and even to call or send their coordinates via SMS to a list of set numbers by setting an “Emergency Key.” Interestingly, it also featured a “Friend Find” service, whereby users with Esc! handsets could track each other’s locations directly on their handset display.

The first GPS phone launched in 1999

It was evident by the late 1990s that by harnessing the power of location services in wireless handsets, GPS could radically alter the smartphone makeup. However, for that to actually happen, the industry had to overcome a few major stumbling blocks. For a start, GPS was a line-of-sight satellite technology while cellular was not. Then there were problems regarding indoor reception of GPS signals, which was inherent in satellite communications. A user couldn’t rely on the phone’s GPS to get around inside buildings.

Handset manufacturers like Ericsson and Nokia were initially reluctant to embed GPS circuitry into mobile phones, citing time-to-market issues and the added cost. A more crucial challenge related to lowering power consumption in the GPS circuitry in order to integrate it into mobile devices. The complexity and footprint of GPS chip, as well as the need for a separate antenna, further complicated a successful integration onto the mobile phone platform.

Then, there came this marvel of system integration that crystallized a new direction for cellular networks’ liaison with server-assisted location services. In May 2000, J-Phone, the first wireless operator to release a phone with a built-in camera, launched the world’s first location-based mapping service that displayed interactive maps within a web micro-browser. GPS met location met mobile Internet!

J-Phone’s J-Navi service was a pioneering effort to merge GPS into the handset

Japan’s second-largest mobile phone operator launched the J-Navi service, letting users in Tokyo enter a phone number, address or landmark, and then search the area within a 500-meter radius. This made it possible to find the subway station nearest to a particular shop, or a particular kind of restaurant within walking distance of an office building.

Most important, users of the service could download a full-color map. At the time of its launch, J-Navi was expected to handle around 100,000 hits per day, but on its third day of operation, it already had 1.6 million users. Searching was free, but users paid for the data transport costs, so in practice, it cost about 4 cents for a location search.

In the meantime, specialized chipmakers continued to improve the accuracy and availability of the GPS technology. The new circuitry was also able to gradually trim the GPS power for use in cellular phones. By late 2000s, smartphones had GPS systems on-board, and with location-aware Internet services, they were helping people to get from point A to point B.

Smartphone opened the floodgates of location-centric innovation

Google’s initiative of free turn-by-turn navigation for smartphones brought a lot of momentum to the smartphone industry. When used with a smartphone, the software sends coordinates to a server over the phone’s wireless Internet connection to grab mapping data. Maps are also stored on the handset’s SD memory card in some cases.

That way, directions keep coming when a user can’t get cellular reception, so long as he or she is still getting a GPS signal. Google’s Navigation app was probably one of the best in the lot; it took up almost no space on the phone because everything was in the cloud.

The Magnificent Seven of International IP Management

The Magnificent Seven of International IP Management
by Paul McLellan on 08-10-2015 at 7:00 am
Categories: EDA, Perforce

Almost all large projects these days are distributed across multiple geographic locations. As the world rotates underneath the sun, the focus of activity moves too: Europe, US, China, India, back to Europe. For this to work effectively requires a collaborative platform designed for multi-site design efforts, a platform that communicates the current state of the design, planned changes, history, and delivers what each site requires with minimal user intervention and maximum efficiency.

There is a famous aphorism, attributed to Willem van der Pohl, that there are only three numbers in computer science: 0, 1 and infinity. When providing a service like IP data management, there are some things that you want to be centralized (only one of them) and otherwise you want an unrestricted number of them (sites, IPs, projects, bugs…). This gives you the best mixture of integrity and efficiency, giving the illusion of a single central repository without the obvious efficiency issues of actually keeping all the data at a single central site.

Here are The Magnificent Seven of multi-site collaboration (or The Seven Samurai if you are more into Japanese cinema, Or 七人の侍 if you are really into Japanese cinema):
[LIST=1]

Centrally Defined Configuration Management: ProjectIC has the PiServer central database of IP and project metadata including the hierarchical resource tree for each project collected from user workspaces.

IP-centric Bug Tracking: Off-the-shelf bug-tracking tools such as Jira and Bugzilla don’t work well in an IP context since each IP is usually considered a separate “project” in the bug-tracking database. With hundreds of IPs in a typical SoC and perhaps thousands in a large semiconductor company this is impractical to use directly. ProjectIC will query the bug database for bugs associated with each IP version and present a consolidated hierarchical view of bugs found for the whole SoC.

Centralized IP Catalog: there needs to be a centrally defined database to discover the existing IP in the company, including its quality, the location of the files in the data management system, all the available versions of the IP, which versions are recommended for use by the IP owner, which projects are using these versions and so on. The ProjectIC PiWeb catalog has an easy to query searchable database with multiple levels of metadata for organizing IPs including labels, custom properties, project based etc. This catalog is auto-updating and self-regulating so that when IPs and IP versions are introduced to the system from with within a project they are automatically added to the catalog (this is important: if manual update is required then it can be guaranteed that the data is stale if not plain wrong).

Multi-Site Data Replication: Although the IP metadata needs to be centralized, the data management repositories themselves need to be replicated to reduce the time to deliver files to user workspaces. For example, one underlying data management environment is Perforce (with its Edge and Replica servers) where metadata queries, commits and syncs can happen locally without the need for WAN access, all without any user interaction.

Decentralized Data Management: Another method for reducing multi-site data latencies is to maintain the master Subversion or Perforce repository at the remote site if that is where the majority of the development activity is taking place. ProjectIC allows the repository location to be defined on a pre-IP basis so that workspace creation will query the local server and reduce WAN delays

Multi-site IP caches: An important part of the Methodics multi-site solution is the use of IP Caches to maintain local read-only versions of popular IPs for consumption at remote sites. These are updated and propagated automatically as part of the IP release process and mean that users who only need read access to a particular IP can set that IP to “refer” in their workspace configuration and ProjectIC will manage the reference automatically.

Large Dataset Block-level Replication: A slightly different technique for reducing multi-site delays is to use Warpstor to maintain master versions of the important project workspaces and deliver changes multi-site between these masters incrementally at the file-system block level, as new releases are made. This reduces the need for large DM checkouts into a workspace since the user workspaces can use lightweight clones using these replicated masters.

See also WarpStor, the Data Tardis: Small on the Outside, Large on the Inside

The white paper Methodics—Architected for Multi-Site Collaboration is here.

August 9, 2015

Virtual Reality is Ready to Rocket

Virtual Reality is Ready to Rocket
by Daniel Payne on 08-09-2015 at 7:00 am
Categories: General, Mobile

Virtual Reality (VR) is such a hot technology concept right now that the topic has made the cover of Time, Wired and Forbes magazines this year, along with countless online articles. What really captured my attention was that moment in 2014 when Facebook acquired VR startup Oculus for $2B, yes that is billions of dollars. The last time that I saw this much excitement was the advent of personal computers back in the late 1970’s.

Palmer Luckey, founder of Oculus VR

To make VR work you need several components:

A headset to immerse viewers with 3D content
A video processing engine capable of quickly rendering 3D content
3D content that is compelling for entertainment, education or exploration

There are a few dozen companies all clamoring for position in this brave, new, 3D, VR world. VR headsets can range from a simple, folded piece of cardboard like Google Cardboard that uses your existing cell phone, to commercial headsets with integrated headphones.

Google Cardboard

Video processing may be supplied by your cellphone, tablet, laptop, desktop or dedicated hardware. If you visit the local BestBuy store, the only VR product for sale are headsets that use either the Samsung Galaxy S6 or Samsung Galaxy Note 4cell phones as the video processing and software from Oculus:

Samsung Gear VR

Let’s take a quick survey of other VR headsets that have been announced to get a feel for the variety to choose from:

Crowd-funded on Kickstarter, owned now by Facebook
Both audio and video, wired
Available Q1 2016

|-
| Project StarVR
|

Hardware by InfinitEyes
Software by Starbreeze
Dual Quad HD screens
Still in development

|-
| AirVR
|

iOS devices only
Kickstarter funded
In development

|-
| Avegant Glyph
|

Audio and Video
120 Hz refresh rate
In development

|-
| Cmoar
|

Powered by Android or iOS devices
Includes Augmented Reality
Kickstarter funded

|-
| Dior Eyes VR
|

Dior fashion brand
Uses Samsung Galaxy Note 4 device
Fashion viewing market

|-
| Emax X1
|

Compatible with Oculus Rift
Made in China
Soon to launch

|-
| Fove
|

Eye-tracking technology
Full 360 degree experience
Kickstarter funded

|-
| Google Cardboard
|

Use most any smart phone
Lowest price
Available now

|-
| Homiod
|

Smart phone powered
Wireless
69.99 Euros

|-
| HTC Vive
|

HTC and Valve team
Available end of 2015
Content from HBO, Lionsgate, Google

|-
| ImmersiON BlueSky Pro
|

Dual 1920 x 1080 displays
Multiple game engine support
From Silicon Valley and Spain

|-
| Impression PI
|

Senses hand gestures
Versions for smart phones and stand-alone
Kickstarter funded

|-
| MindMaze NeuroGoggles
|

Brainwave controlled
Tracks hand motions
From China

|-
| Pinch VR
|

Uses your smart phone
Includes finger sensors
From Canada

|-
| Razer OSVR
|

Software is Open Source
Well-known gaming company
Coming soon

|-
| Samsung Gear VR
|

Available at BestBuy
Uses Samsung Galaxy Note 4 or S6 devices
$199.00

Source: Virtual Reality Times

What an amazing array of VR devices that are soon to be unleashed into the consumer and commercial marketplace. If I were to guess which of these VR headsets are still around in 18 months it would: Oculus Rift, Samsung Gear VR and Razer OSVR. Let’s see what Facebook does with their Oculus brand and if Samsung can create early interest in VR. I was surprised to see some 114 favorable product reviews on BestBuy about the Samsung Gear VR device, yielding a 4.5 out of 5 star rating.

Economically speaking, I think that the real winners are the 3D content creators, because they should command bigger revenues than just the hardware providers of VR. It’s the same business model as Game Consoles, sell the console for a low or subsidized price, then make all of your revenue and profits on each new gaming title.

Who knows, there may even be some use of VR for EDA companies as IC designers may benefit from taking a 3D tour of their FinFET chips, packages and boards.

August 9, 2015June 14, 2019

Never Imagined So Easy Class-based Testbench Debugging

Never Imagined So Easy Class-based Testbench Debugging
by Pawan Fangaria on 08-09-2015 at 7:00 am
Categories: EDA, Siemens EDA

When it comes to debugging a design testbench organized in object-oriented style with objects, component hierarchies, macros, transactions and so on, it becomes an onerous, tasteless, and thankless task for RTL verification engineers who generally lag in software expertise. Moreover, class-based debugging tools have lagged simulators ability to simulate these testbenches. However, modern testbenches are most likely to be class-based using object-oriented programming concepts. They are generally developed using SystemVerilog and UVM. In such a situation, how about having an automated, GUI driven, post-simulation debugging environment which becomes a fun to operate with for RTL verification engineers without needing software expertise? Yes, such class-based debugging tools and environment have come up to make debugging of SoC design testbenches easy, interesting and productive.

Mentor’sVisualizer along with Questa simulation provides an excellent class-based debugging environment. In the above diagram, there is a simple UVM schematic with a DUT and four interfaces. Each interface is connected to an agent. Traversing down the hierarchy, we can see that each agent has a sequencer, a driver, and a monitor. Although the UVM-based testbench can always be explored through schematics, a faster and convenient approach for debugging large circuits can be to debug directly through objects in the testbench.

The UVM component hierarchy can be traversed to any level. Any object selected in the component hierarchy window will have its corresponding source code displayed in the source code window. In the source code, the value of any member variable inside an instance of an object can be displayed by just hovering over the variable. In the post-simulation mode, all objects that existed during simulation and recorded by Questa are accessible for investigation into their class member variables. Similarly virtual interfaces of the objects can be explored. For post-simulation exploration, there are easier ways to display all member variables of an object at once at any instance of time.

By pressing the right-mouse-button in the source code of an object, selecting ‘Browse This’, and then selecting ‘t’, one can see the complete object pointed to by ‘t’ at the current time. The ‘t’ points at the object ‘@sequence_item_A@4841’ in the above example. A still shorter way to see the object could be by selecting ‘t’, pressing right-mouse-button, and then selecting ‘Browse (t)’.

By pressing the right-mouse-button and selecting ‘Add ‘this’ To Wave’, one can add the current object to the waveform window and display all the transactions that the object has created. That’s really powerful.

The Class Instance Window provides another way of exploration for UVM testbenches. It is organized by base class. By expanding and selecting particular objects we can change the current context to a required object. The creation times of different objects are displayed in this window. A search can be created to find objects created between two times. Also, instances can be searched based on a regular expression.

In the above window, a search is created based on any object ending in ‘4839’. The regular expression “*4839” has been entered in the search field. A selection of the first sequence_item_A has also displayed the source code for that object. Hovering over the ‘addr’ field shows the address for that transaction (that sequence item). Similarly, the other sequence_item_A objects can be investigated.

Also, the expressions can be created on particular address values for checking transactions on those addresses.

A StripeViewer is opened above and expression ‘32’h000006eb’ entered in the ‘addr’ field. A search then shows the above transactions.

On a click on ‘Add to Wave’, the transaction stream enters the wave window. A selection of any transaction in the StripeViewer highlights the same in the wave window. For example, the transaction selected, as shown in green in the StripeViewer, moves the cursor to the particular transaction in the wave window as shown above.

This kind of post-simulation class-based debugging brings a new dimension to modern SoC verification and debugging. The designers can have full visibility into the design with little effort and RTL verification engineers can easily debug the design. Read a whitepaperwritten by Rich Edelman at Mentor Graphics for a detailed and interesting tour over these methods. They have applied this approach for debugging real customer designs with great success!

Pawan Kumar Fangaria
Founder & President at www.fangarias.com