Bronco Webinar 800x100 1
Semiconductor Expert Forum

Posted on

Acceleration in a Heterogenous Compute Environment

Acceleration in a Heterogenous Compute Environment
by Bernard Murphy on 10-02-2019 at 5:00 am
Categories: eFPGA, FPGA
3 Comments

Acceleration

Heterogenous compute isn’t a new concept. We’ve had it in phones and datacenters for quite a while – CPUs complemented by GPUs, DSPs and perhaps other specialized processors. But each of these compute engines has a very specific role, each driven by its own software (or training in the case of AI accelerators). You write software for the CPU, you write different software for the GPU and so on. Which makes sense, but it’s not general-purpose acceleration for a unified code-set. Could there be an equivalent in heterogenous compute to the multi-threading we use every day in multi-core compute?

Of course we need to think outside the box on how this might work; you can’t just drop general code on a mixed architecture and expect acceleration, any more than you can drop general code on a multi-core system with similar expectations. But given sufficient imagination, it appears the answer is yes. I recently came across a company called CacheQ which claims to provide a compelling answer in this space.

The company was founded just last year and is headed by a couple of senior FPGA guys. Clay Johnson (CEO) was VP of a Xilinx BU for a long time before going on to lead security ventures, and Dave Bennett (CTO) has a similar background, leading software dev at Xilinx for many years before again joining Clay in the security biz and now in CacheQ. Funding comes from Social Capital (amount not disclosed).

Given their background, it’s not surprising they turned first to FPGAs as a resource for acceleration. FPGAs are becoming more common as resources in datacenters (just look at Microsoft Azure) and in a lot of edge applications, I’m guessing for flexibility and easy field update, also as an appealing option for relatively low volume applications. And they are also starting to appear as embedded IP inside SoCs.

Back to the goal. CacheQ’s objective is to let software developers start with C-code (or object code) and to be able to significantly accelerate that code by leveraging a combination of CPU and FPGA resources while speeding and simplifying implementation and partitioning between processor and FPGA. At this point I started to wonder if this was some kind of high-level synthesis (HLS) play. It isn’t (the object code option is perhaps a hint). They position their product as an ultravisor (think amped-up hypervisor). They build a virtual machine around an application which then goes through an optimization and partitioning phase, then into code generation and mapping across any of several possible targets: x86 servers or desktops, FPGA accelerators, embedded Arm devices or heterogenous SoCs.

Still, doesn’t mapping onto FPGAs require going through RTL with all its concomitant challenges? Here the company provides some detail though they are understandably cagey about providing too much. So here is what I can tell you. Part of what they’re doing is unrolling complex loops in the code and mapping these, with pipelining, into the FPGA. They also automatically create the stack to manage CPU to FPGA communication and they manage memory allocation transparently across these domains.

The “aha” here is that they’re providing a way for software developers to get acceleration, not a way for hardware developers to build a design. This is a quite different intent from HLS and a goal that many have been chasing for a while. They don’t have to map everything to the FPGA, they just have to provide significant net speedup in critical pieces of code. They show some impressive numbers for key functions on their website

I asked about active applications today. Clay mentioned use in weather simulation, industrial and government applications. I also asked about support for other potential accelerators (GPU, DSP, …). He said that these are in long-term planning; each can offer acceleration in its own way, I would guess for big matrix operations as an example.

This looks like an interesting challenge to the long-standing problem of making FPGAs (and ultimately other platforms) more accessible to the general-purpose programmer. Worth a closer look. The website is HERE.

Posted on

Webinar: OCV and Timing Closure Sign-off by Silvaco on Oct 10 at 10AM

Webinar: OCV and Timing Closure Sign-off by Silvaco on Oct 10 at 10AM
by Daniel Nenni on 10-01-2019 at 10:00 am
Categories: EDA, Events

The old adage that goes the one constant thing you can always count on is change, could easily be reworded for semiconductor design to say the one constant thing you can count on is variation. This is doubly true. Not only is variation, in all its forms, a constant factor in design, additionally the methods of analyzing and dealing with are continuously changing as well. For both of these reasons it is necessary to stay current and on top of the latest developments in the role of variation in timing closure.

Designers are often faced with a tradeoff between less pessimistic results and extraordinarily long runtimes and large data sets. There is a long history of innovation that has brought us to what are considered the best approaches for massive designs. The foundation of the entire process relies on library characterization to produce models that support effective chip level analysis. Advanced nodes have further complicated the process.

Fortunately, Silvaco is offering a free webinar that will provide a useful update on both the history and state of the art for how on-chip variation can affect the sign-off timing flow. Silvaco has longstanding expertise at the process and cell level, as well as flows for chip level timing verification.

The October 10th webinar will be presented by Bernardo Culau, Director of Library Characterization at Silvaco. He has in-depth experience developing tools for library characterization. The topics that will be covered include:

    • What variation means in the context of library characterization
    • What on-chip variation is and its different causes
      • Inter-chip variation
      • Intra-chip variation
    • Review of different industry approaches to account for on-chip variation
      • OCV, AOCV, LVF
      • Their advantages and limitations
    • Current industry standards for variation-aware libraries
    • The improvements needed to handle leading-edge technology nodes
    • The characterization challenges involved in creating variation-aware libraries
    • Silvaco solutions for variation-aware library characterization

The free webinar will be offered at 10AM on October 10th. This seems like a good opportunity to stay current with the latest trends in a critical area of chip design. It also looks like it might provide a glimpse of what is ahead in this area. Should be interesting.

About Silvaco, Inc.
Silvaco Inc. is a leading EDA tools and semiconductor IP provider used for process and device development for advanced semiconductors, power IC, display and memory design. For over 30 years, Silvaco has enabled its customers to develop next generation semiconductor products in the shortest time with reduced cost. We are a technology company outpacing the EDA industry by delivering innovative smart silicon solutions to meet the world’s ever-growing demand for mobile intelligent computing. The company is headquartered in Santa Clara, California and has a global presence with offices located in North America, Europe, Japan and Asia.

Posted on

Webinar: Finding Your Way Through Formal Verification

Webinar: Finding Your Way Through Formal Verification
by Bernard Murphy on 10-01-2019 at 6:00 am
Categories: EDA, Events, Synopsys

Finding your way through formal book

Formal verification has always appeared daunting to me and I suspect to many other people also. Logic simulation feels like a “roll your sleeves up and get the job done” kind of verification, easily understood, accessible to everyone, little specialized training required. Formal methods for many years remained the domain of academics and one-time-academics performing an essentially black-box service to solve really hard problems unreachable in simulation. That’s changed, because those hard problems are becoming much more common and because verification tool providers have made it much easier to attack many cases without needing a PhD or induction into the formal priesthood. Better yet, there are excellent books to introduce novices to the domain and walk them through their first steps in using those tools.

REGISTER HERE to watch the recorded webinar.

One thing we felt was missing was a higher-level introduction, for a verification engineer, manager or director who is curious about formal but not yet ready to commit. They don’t want a tutorial; they want to know first “is this right for my organization?” “Is it really going to improve our verification quality or throughput?” “What are we going to have to do differently?” We wrote “Finding Your Way Through Formal Verification” for them. “We” here is myself (Bernard Murphy), Manish Pandey and Sean Safarpour. The book was published by SemiWiki and is available for download in the handouts section of this webinar.

Speakers:

Bernard Murphy – SemiWiki

Bernard Murphy is a freelance blogger and author, content marketing/messaging advisor for several companies and serves on the board of Mother Lode Wildlife Care in the California Gold Country. In a previous life he held down a real job as CTO at Atrenta. Earlier still, he held technical contributor, management, sales and marketing roles variously at Cadence, National Semiconductor, Fairchild and Harris Semiconductor. In his re-invention as a writer, Bernard has published well over 400 blogs between SemiWiki and EETimes. He received his BA in Physics and D. Phil in Nuclear Physics from the University of Oxford.

Manish Pandey – Synopsys

Manish Pandey is a Fellow at Synopsys, and an Adjunct Professor at Carnegie Mellon University. He completed his PhD in Computer Science from Carnegie Mellon University and a B. Tech. in Computer Science from the Indian Institute of Technology Kharagpur. He currently leads the R&D teams for formal and static technologies, and machine learning at Synopsys. He previously led the development of several static and formal verification technologies at Verplex and Cadence which are in widespread use in the industry. Manish has been the recipient of the IEEE Transaction in CAD Outstanding Young author award and holds over two dozen patents and refereed publications. 

Sean Safarpour – Synopsys

Sean Safarpour is the Director of Application Engineering at Synopsys, where his team of specialists support the development and deployment of products such as VC Formal, Hector and Assertion IPs. He works closely with customers and R&D to solve their current verification challenges as well as to define and realize the next generation of formal applications. Prior to Synopsys, Sean was Director of R&D at Atrenta focused on new technology, and VP of Engineering and CTO at Vennsa Technologies, a start-up focused on automated root-cause analysis using formal techniques. Sean received his PhD from the University of Toronto where he completed his thesis entitled “Formal Methods in Automated Design Debugging”.

 

 

Posted on

AI Hardware Summit, Report #2: Lowering Power at the Edge with HLS

AI Hardware Summit, Report #2: Lowering Power at the Edge with HLS
by Randy Smith on 09-30-2019 at 10:00 am
Categories: EDA, Events, Siemens EDA

I previously wrote a blog about a session from Day 1 of the AI Hardware Summit at the Computer History Museum in Mountain View, CA, held just last week. From Day 2, I want to delve into this presentation by Bryan Bowyer, Director of Engineering, Digital Design & Implementation Solutions Division at Mentor, a Siemens Business. This conference brought together many companies involved in building artificial Intelligent and machine learning hardware solutions. Naturally, there were several discussions around AI software and applications as well. Day 1 of the conference was more about solutions in the data center, whereas Day 2 was primarily around solutions at the Edge.

Most solutions at the Edge have power restrictions. These often are battery-powered or energy harvesting devices such as remote cameras, robots, cell phones, and many other sensor-carrying devices. A different class of edge devices represents those devices which are always on, such as smart appliances, leading to power concerns for a different reason – because it is always on. Of course, higher power typically leads to higher heat dissipation which again will lead us to prefer lower power. At the Edge, power is critically important.

When you look at designing for low power, we have been given many new tools and techniques over the past few years. Some are in process technology and the potential power saving we achieve from going to new process nodes. These savings have been getting less dramatic at each node. There are also circuit innovations that can occur, such as new memory techniques. The biggest savings will come from architectural decisions. This is for two important reasons: (1) if you can find a more efficient algorithm you can save energy; and (2) if you can implement the system in a set of blocks that give you the ability to turn off a large amount of the system resources when they are not in use, you can also save substantial power. The challenge is how best to achieve this.

To explore the algorithmic space, you need to be able to work at a sufficiently high level to make a difference. By far, the simplest way to do this is with high-level synthesis (HLS). Companies are doing this today, particularly in the areas of video and image processing, and in machine learning. HLS can be applied to ASIC or FPGA design. It can even make it possible to make late functional changes without severely impacting the project schedule. Mentor’s Catapult HLS has been deployed by many top companies in this area, as shown in the diagram above. You can get more information on Catapult HLS here.

One of the significant challenges of building custom hardware solutions is trying to explore multiple architectural choices to find the best combination of power, performance, and area (PPA). One aspect of this is in estimating the impact in different levels of precision in different stages of the design. Reducing accuracy in earlier stage may have little impact on the final result, yet greatly reduce power or area. Exploring these architectural choices in RTL is impractical. Instead, designers are turning to High-level Synthesis (HLS) to survey these custom solutions.  HLS provides several high-level optimizations such as automatic memory partitioning for complex memory architectures needed by the PE array, interface synthesis of AXI4 memory master interfaces for easily connecting to system memory, and synthesis of arbitrary precision data types for tuning the precision of the multiple hardware architectures.  Since the source language for HLS is typically C++, it can easily plug back into the deep-learning framework where the network was created, allowing verification of the architected and quantized network.

For those working on AI/ML accelerators, Mentor has made it even easier to get started by providing Catapult HLS Toolkits. There are currently four toolkits available, as described in the figure above. These toolkits seem especially well suited to AI/ML designs in Edge devices. Mentor’s participation in the AI Hardware Summit, the number of customers already using Catapult HLS in production, and the release of these toolkits specifically targeting designs people need for AI and ML have me convinced that Mentor is quite serious about this area and designers in this area need to consider them.

Below are some additional resources on the Mentor website about this topic:

Chips&Media: Design and Verification of Deep Learning Object Detection IP

Bosch Visiontec Case Study

NVIDIA Case Study on High-Level Synthesis (HLS)

Posted on

Crashing the Mars Rovers!!! Actel and Aerospace Corp

Crashing the Mars Rovers!!! Actel and Aerospace Corp
by John East on 09-30-2019 at 6:00 am
Categories: John East
8 Comments

In early 2003 Actel announced a new product family:  RTSX-A.  It was a family of antifuse FPGAs aimed at the satellite market.  Customers had known for a long time that it was coming and there had been prototypes available for many months.  Our space customers loved the product.  This was going to be a big win for us!  One of the first programs to use the product was the Mars Rover program. There were four Mars Rovers made:  two went to Mars and two lived in a huge sandbox in Pasadena which allowed scientists to emulate conditions on Mars.  The two that went to Mars were named the Spirit and the Opportunity.  NASA’s plans were that the Rovers would live for three months before succumbing to the treacherous Mars environment. The Spirit launch was on June 10,  2003.  It would take the Spirit about eight months to make the trip to Mars.  The Opportunity was launched a month later.

Shortly after the launch we got reports from potential customers of a few RTSX-A burn-in failures in their labs. What’s burn-in?  It’s a test that assures the user that a part that starts out good will stay that way after being in actual use for some time.  Testing parts for one week at extremely high temperatures is the normal way to assure that they’ll last a long time at normal temperatures. We hadn’t seen failures with our first tests, but when we got those reports from our customers, we looked harder.  When we looked harder, we saw some failures on occasion.  We would burn-in around 100 units at a time.  Sometimes we would get 1 or 2 failures.  Sometimes none.

A small number of failures is worse than it sounds! A typical satellite would cost in the neighborhood of 100 million dollars in those days.  The Mars Rover project cost much, much more than that. The Mars Rover project used the RTSX-A.  Uh oh!!

Satellites can’t be repaired. If one IC fails, the cost of the entire project may well be flushed down the toilet.  Worse, there wasn’t just one RTSX-A part in each Rover.  There were, as I recall, 38.  That meant if there was a 1% chance that one particular Actel part would fail, there was a 38% chance that one of our parts somewhere in the Rover would fail.  (My math isn’t quite correct here, but you get the point) It was clear something wasn’t quite right, but we couldn’t figure out what.

The Mars Rover program was by no means the only program planning to use our parts.  There were many others! Word got around the space community. Many customers weren’t sure if they dared to launch their satellites. They looked to us to tell them it was OK.  We couldn’t. We just didn’t know. We couldn’t figure it out. A few skeptics thought we were covering something up. We weren’t. It was a very tricky problem. We were working hard on it, but we just didn’t understand what was going on.

Then, I got a call from Bill Ballhaus, the CEO of Aerospace Corporation. Aerospace Corporation operates a federally funded research and development center. They provide technical guidance and advice on all aspects of space missions to military, civil, and commercial customers.  Dr Ballhaus asked me to come to the Aerospace headquarters in El Segundo to discuss “the reliability problem with Actel FPGAs”

I, of course, accepted the invitation, but if you had offered me a choice of going to this meeting or getting a root canal on every tooth, my oral surgeon would be a richer man today.  The invitation appeared to be for a one-on-one meeting between me and Dr Ballhaus.  I planned on going by myself, but our VP of Technology,  Esmat Hamdy, saw it differently.  He didn’t quite trust my technical savvy.  He thought that, if the meeting turned out to have a lot of technical content, I might not be able to answer all the questions well.  So —  Esmat insisted on coming with me. Bless his heart!

We flew to LAX and then took a quick cab ride to Aerospace.  It’s about a mile from the airport. A secretary led us to a conference room  — except it was more like a sports arena.  There was a long rectangular table that probably sat 15-20 people.  Then there was an aisle circling those seats.  But on the other side of the aisle, there was an elevated set of chairs circling the table below.   There were maybe another 20-25 chairs in that set.  In total I would guess 30 or 40 chairs.    All but four were full.  Not full of just anybody   —  but full of PhDs.  Full of experts in any aspect of integrated circuits that you could think of.  Full of technical wizards who all had at least double my IQ. One of the empty chairs was for Dr Ballhaus.  One for Esmat.  One for me.  We took our chairs and were ready to start the meeting but Dr Ballhaus said that we’d have to wait.  The last chair was for a high ranking Air Force general who had invited himself to the meeting.   The general was running late.  When he got there, there was plenty of bowing and scraping done.  He was the head honcho!!

They hammered into me just how important this was.  The Rover program cost about one billion dollars.  The Rovers had been launched. There would be no calling them back.  No fixing them. If they went bad, that would be a billion dollars flushed.  And worse, the Mars Rover project was by no means the only satellite project with plans to use Actel.  There were several military satellite programs related to our national defense as well.  Those folks were even more worried than the Mars Rover people.  They were not happy campers!!!  As you would expect — the general had zero interest in putting up military satellites critical to the nation’s defense that were likely to fail!!!!!  (That unhappiness earned me an invitation to meet later with Peter Teets,  the Undersecretary of the Airforce, in the most secure area of the Pentagon.  Mr Teets wasn’t a happy camper either, but that’s another story.)

Back in Mountain View, we were breaking our picks.  We didn’t always see failures. When we saw them, there weren’t many.  That makes the problem harder to solve.  We suspected what we called the programming algorithm. The oversimplified explanation of programming an antifuse is this —  put a high voltage across it,  the dielectric will rupture and the antifuse will be a conductor for the rest of time. — In fact, it’s much trickier than that.  There are a lot of knobs to turn.  How high should the voltage be?  How much current should flow through the fuse?  How many times should you repeat what you’ve done?  How long should you apply the voltage?  How much should you “soak” the fuse? We would twist some of these knobs, come up with a new algorithm, and voila.  No failures!!!   Problem solved,  right?  Wrong!!!  The next time we’d run exactly the same test, there would be one or two failures.  We were completely perplexed!

The first Rover (the Spirit) was launched prior to suspicions that we might have a reliability problem. Then came the reliability worries. And then, around New Years when the reliability concerns had become rampant, the Spirit reached Mars. It was a big deal in the press.

The Spirit Lands on Mars!

It’s working fine!!

It was on the front page of every newspaper. Boy.  Did we ever feel good.  The Spirit was working perfectly!!!   …..……   But then  —  one week later —-

The Spirit Fails!!

The Spirit went bad.  That was on every front page too, but in bigger letters.  Was it the Actel parts? We didn’t know, but in my judgement … it could well have been. I was terrified!  I could picture the headlines when it was determined that Spirit failed because of an Actel part.  I could picture the lawyers lining up to file lawsuits against us.  I could picture the process servers skulking in the bushes waiting to spring out and serve me with subpoenas.  It wasn’t pretty!!  In fact, it was really, really ugly!! When I went home after work that night, I opened a bottle of wine and drank the whole thing by myself.  I like wine ……  but not a whole bottle.  My advice?   —  Don’t do that!!!  It didn’t work out well!!

Luckily, I was wrong. The Actel parts were not at fault.  After a week or so, NASA figured it out.  It was a software problem that was fixable by uploading new software.  They fixed the Rover, and independently we tracked down our problem and fixed it. To my knowledge Actel (Now part of Microchip) has never experienced a failure in space.

Scientists had planned for the Rovers to live for three months.  When did they actually die?  The Spirit lived for 6 years.  The Opportunity 14.

Epilogue

Somewhere around the year 2000 our board of directors asked the question, “John, how long do you plan to stay on as CEO?”.  My answer:  “I’ll retire no sooner than my 65th birthday and no later than my 66th”.  When you’re 55,  65 seems really old, doesn’t it?  Well, ten years later along came my 65th birthday.  January 20, 2010.  Funny thing.  By then 65 didn’t seem so old.  Still – all in all it was the right thing to do. We released an 8K (that’s the document that public companies use to disclose relevant information.) saying that we were beginning a search for a new CEO immediately and that we planned to complete the search and appoint a new CEO within a year.

We spent several months on the search.  There were some ups and downs – but bottom line, we made zero progress.  We were back to square one.  Then came a surprise. We received an unsolicited and unexpected offer to buy us out from Microsemi  — an Orange County based semiconductor company that I was barely aware of.  We discussed it at length!

Why would we be interested in selling?  We were in the middle of trying to transform ourselves from an antifuse company to a flash company.  I firmly believed that it could be done (In fact,  it was done!!!  See my week # 16, “From AMD to Actel”),  but it was obvious that it was going to take a long time before we finally started to see the results on our bottom line.  The bottom line matters!!  Shareholders want high stock prices!! And — for a company our age, the stock price is determined by the bottom line.

Dan McCranie (Who we had recently appointed chairman of the board) went to Orange County and met with the Microsemi CEO, James Peterson.  After some negotiations, Peterson offered a price that was higher than we believed we would be able to command consistently in the foreseeable future.  After a few vigorous board discussions,  we decided that we owed it to our shareholders to accept the offer.  We did. On November 3, 2010 Actel became part of Microsemi Corporation and I rode off into the high tech sunset  –  unemployed for the first time in 45 years.

This is the last episode in the series.  I hope you’ve enjoyed reading them as much as I’ve enjoyed writing them.

See the entire John East series HERE.

# Mars Rovers, Spirit,  Opportunity, Actel, Microchip, Aerospace Corporation,  Bill Ballhaus, Esmat Hamdy  Microsemi, Dan McCranie

 

Posted on

Fossil Fuels in the Crosshairs

Fossil Fuels in the Crosshairs
by Roger C. Lanctot on 09-29-2019 at 11:00 am
Categories: Automotive
3 Comments

In the heat of a presidential campaign, especially one with 19 competing candidates, the contenders may get carried away in the interest of getting attention and, presumably, attracting supporters. Beto O’Rourke might be accused of such rhetorical excess for his call, in the third Democratic Party debate, for a mandatory Federal assault rifle buyback.

But fellow Democrat Senator Bernie Sanders may have beat Beto with his plan to pursue criminal charges against fossil fuel executives for, in the words of truthdig.com, “knowingly accelerating the ecological crisis while sowing doubt about the science to the American public.”

Sanders’ comments came during an MSNBC climate town hall hosted by news anchor Chris Hayes at Georgetown University last Thursday. Truthdig.com quotes Sanders:  “Duh, of course I would. They knew that it was real. Their own scientists told them that it was real. What do you do to people who lied in a very bold-faced way, lied to the American people, lied to the media? How do you hold them accountable?”

Over the subsequent din of shredding machines going into overdrive at the offices of the major oil companies could be heard the clacking of worry beads at the headquarters of the major car companies. Knowledge of climate change is one thing. Building a product collectively responsible for 1.2M highway fatalities globally and hundreds of thousands of premature deaths from emissions every year – with an aggregate estimated negative societal impact of more than $1T – is enough to give even Senator Sanders and former Congressman O’Rourke pause to consider their options.

President Donald Trump’s efforts to roll back emissions standards and fuel efficiency requirements are pulling back the curtain on some very unpleasant issues the automotive industry would rather not highlight. While the industry is wrestling with the challenges of connecting cars and the onset of autonomous driving and electrification, the specter of premature death and injury caused by motor vehicles with internal combustion engines is an ugly open secret the industry would prefer remain out of the spotlight.

President Trump has dragged the issue to the center of the stage in the interest, so he says, of helping to jumpstart automobile production. The subtleties of managing emissions and fuel efficiency while enhancing vehicle safety – a high wire act which the automotive industry has ably executed thus far – has clearly eluded the commander in chief.

Young people and old around the world have made their concerns known regarding climate change. Their ire is mainly focused, today, on legislators and politicians. It won’t be long before they turn on the fossil fuel producers. Car makers could well be next – as was clear from the presence of protesters demonstrating once again outside the 2019 Frankfurt Auto Show. The President isn’t helping.

Posted on

Micron Mired in Murky Memory Market – Cutting Capex 30%- 2020 Challenging

Micron Mired in Murky Memory Market – Cutting Capex 30%- 2020 Challenging
by Daniel Nenni on 09-29-2019 at 10:00 am
Categories: Semiconductor Advisors, Semiconductor Services
1 Comment

  • Solid Quarter but soft Outlook
  • Recovery Slow- Future Cost Downs Harder
  • Demand slightly ahead of supply-Shelf Stuffed?
  • Bouncing along the Bottom of the Cycle

Results ahead of expectation but guide behind expectation
Quarterly results were slightly better than street expectations at $0.56 EPS and $4.87B in revenues however guidance is for $5B +-$200M in revenues and $0.46 +-$0.07 in EPS which is well below expectations. While there may be some normal “sandbagging” of forward looking guidance even with that assumption its an unimpressive guide.

Disappointment that we are not yet at low tide
Investors are clearly not happy that the guidance does not yet indicate a bottoming of business. The stock was priced to perfection and a quarter guide that suggested being past the bottom of the cycle was also baked in to the high valuation.

We have been saying for a long time that this will be a longer, slower, shallower recovery. Investor and analyst hopes had gotten way ahead of reality. The reality is that we still have excess supply and demand is lukewarm with a number of potential risks in the market that add to uncertainty.

Cutting Capex 30% in 2020, front end cuts are deeper than back end
The company said what we have heard before and have been talking about for a while now……Capital spending will be down by 30% in 2020 versus 2019. Worse yet, the spending will be more focused on back end , assembly and test as well as buildings and less on front end equipment.

If we had to guess we would bet that front end equipment purchases are down at least 40% maybe as much as 50% while back end may only be down 20% or less.

This is obviously very negative for Lam and Applied and to a lesser extent ASML and KLAC. This additional data point of front end being cut more than 30% is obviously incrementally a lot more negative than some analysts and investors had been hoping for.

Wafer starts still being cut, bit growth will come from technology advancement
Micron made it clear that they are still idling capacity and wafer starts continue to come down as machines come off line as plugs are pulled. They said that bit growth will come from density (technology) increases not wafer start increases and that technology increases are enough to keep up with the needed teenage demand.

This is something we have been repeating for a while now, that bit growth can be met with Moore’s Law. So we can read this as selected technology only purchases, focused on pushing Moore’s law forward.

Further cost reductions will be more difficult in 2020
The company was also very clear that the aggressive cost of manufacturing reductions seen in 2019 will not be repeated in 2020. It sounds as if we are past the “easy” technological advancements are are now into more difficult technology changes that will be slower, harder and more costly.

We would read this as the company telegraphing that gross margins will be harder to come by in 2020 as costs will not come down as quickly as pricing.

Is Channel Stuffing going on?…yes
We have been warning of potential channel stuffing going on as Chinese buyers may be stocking up on fears of being cut off or buyers who frequent Samsung may be concerned about Japan cutting them off from critical materials.

Micron said this was going on but could not quantify how much of demand was this “mirage demand” due to stocking up.

Continued progress on 1Z and 96/128 NAND
Micron has made excellent progress in moving the technology ball forward and executing on all these new fronts.

These many advancements are clearly why Micron has been able to keep costs coming down ahead of falling prices. In past cycles, price drops always got ahead of cost drops but Micron has done a better job in this down cycle.

This has kept the company in much better shape on a competitive basis than in previous down cycles. Micron is likely a lot more competitive with the industry leader , Samsung, and claims to be ahead in some aspects.

Huawei is still “No Way”
Huawei is still on the “verboten” list even though Micron has applied for permission. News coming out today makes it seem as if the odds of Huawei being taken off the entity list or waivers being granted seem very low.

We would assume little to no business from Huawei going forward. If somehow this changes we would consider it a lucky break.

Still in wait and hope mode…maybe less hopeful
Micron’s stock has been in “high hopes” mode as expectations for a recovery got well ahead of themselves along with the stock price. We will obviously see a return to the reality of a slow recovery out of a murky bottom with an ill defined turning point.

We could see the run up in overall semi stocks reverse a bit as the wind comes out.

The stocks
There will likely be a significant correction in Micron’s stock price which was well ahead of where it should have been.

If we are at a run rate of $2 per year in EPS and were closing in on a $50 stock price we are at a 25 multiple which is obviously hard to support even at bottom numbers. There is likely support at the $40 level but we would not be interested in buying unless and until we got back to a “3” handle.

Collateral Damage
We don’t expect any better report coming out of Samsung as they are in the very same memory market with similar dynamics plus the additional worry of the Japanese embargo and now what looks like yield issues on their 7NM logic side as well. Samsung obviously gets the same pricing as Micron and even though Samsungs costs are generally lower there is less of a differential in the current down cycle.

Applied and Lam could see a 40-50% cut in business from Micron in 2020, making an equipment recovery all that much harder. While ASML and KLAC are more associated with technology advances than capacity advances there will also see weakness though less as KLA has always been more foundry/logic driven.

It would not be unreasonable to expect a 10% haircut in Micron’s stock price from its recent peak and AMAT and LRCX perhaps a 3-4% cut.

Posted on

GLOBALFOUNDRIES Ready for IPO in 2022?

GLOBALFOUNDRIES Ready for IPO in 2022?
by Daniel Nenni on 09-28-2019 at 6:00 am
Categories: Events, Foundries, GlobalFoundries
5 Comments

Hard to believe but it’s the 10th anniversary of Globalfoundries. What a journey this has been. It truly has been an honor to work with GF over the years as they invested many billions of dollars in the fabless semiconductor ecosystem and added a colorful chapter in semiconductor history, absolutely.

We have written hundreds of articles about GF that have been viewed by more than one million people. GF also has a chapter in our first book “Fabless: The Transformation of the Semiconductor Industry” which, in the 2019 update, documents the appropriately named GF pivot of 2018.

GF CEO Dr. Thomas Caulfield keynoted this year’s “Future of Innovation” event. Today GF has more than 16,000 employees and $6B in revenue making them the second largest pure-play foundry and the largest “specialty foundry”.

Tom made some interesting points in his opening:

  • World economy $85T
  • Electronics Industry $2T
  • Semiconductor Industry $475B
  • Semiconductor Foundry $63B

It really is interesting to know that five semiconductor foundries support the majority of an $85T world economy. Seriously, take away semiconductors and what do we have besides fire and the wheel?

It is also interesting to know that (according to LinkedIn) there are only 521,816 people worldwide who list themselves as “semiconductor related” professionals. So, congratulations to all of the hardworking semiconductor people like myself who made this miracle we call modern electronics possible.

Tom rightfully pointed out that 75% of the semiconductor devices shipping today are based on mature technologies (12nm and above) which is where the GF pivot has focused them. Tom also highlighted that the current GF output of 2.3M wafers ($6B in revenue) can be easily expanded to 3M wafers with an expected revenue of $8B. This is not a giant leap, in fact, I think that revenues will be even higher based on the platform strategy that was outlined in the presentation.

Please note that the equation in the figure above is a product (x) versus a sum (+) meaning that if any one of the variables is 0 the result is zero. This plays directly to the fabless systems companies which is the richest customer segment for the foundries.

Tom mentioned 15 different platforms utilizing 14 application features and 1,000s of silicon proven IP which will result in thousands of specialized application solutions. Again, the target here is electronics systems companies that are making their own chips.

The most interesting news out of the conference however was that GF is planning a public offering in 2022.  I’m a big fan of disruptive moves and while the GF pivot of 2018 was not what I would call disruptive, this IPO certainly will be.

A technology company IPO is definitely a rite of passage into corporate adulthood as it comes with a healthy level of transparency. Given the open media (fake news) we have today it is far too easy to become delusional from PR gone wild. Wall Street however is less easily fooled if you are playing by their rules, absolutely.

The big swing here is the legal action GLOBALFOUNDRIES filed against TSMC and some of their top customers. If the Wall Street bankers can make a silk purse out of this sow’s ear some serious IPO money will exchange hands and Abu Dhabi can finally put this one in the win column.

The semiconductor industry is full of incredibly smart people and it is an honor to work amongst them. One thing I can tell you is that the moves GF has made since Tom took the helm have been rock solid so I would not bet against him, not today.

Posted on

Chapter Twelve – The Future

Chapter Twelve – The Future
by Wally Rhines on 09-27-2019 at 6:00 am
Categories: Wally Rhines
1 Comment

Content of this book has focused upon predictability of trends in the semiconductor industry based upon past trends, experience and ratios.  What about newly emerging applications of semiconductors?  After all, the entire history of the semiconductor industry is driven by emergence of new applications.

Artificial Intelligence
One of the most exciting new applications affecting semiconductor technology is the broad adoption of AI related analytics. AI is not a new technology.  Figure 1 is the cover of High Technology magazine in July 1986.  I am the person on the left and George Heilmeier, former head of DARPA, is the one on the right.  We tried hard in the 1980s but the infrastructure had not developed to a level where AI would provide profitable opportunities.

Figure 1. Artificial Intelligence technology heavyweights of the 1980s

What’s different today? In the 1980s, we lacked the computing power to handle big data.   We didn’t have very much big data to analyze partly because there was no internet of things. More sophisticated algorithms were needed.  Most of all, there were no obvious near term ways to make money using AI.

Today we have overcome all these limitations.  AI and machine learning have taken on a life of their own.  They have become limited, however, by the processing power available.  Traditional von Neuman general purpose computing architectures are inadequate to handle the complex AI algorithms. The result: a new generation of computer architectures is evolving.

Figure 2 shows the trend in venture capital funded fabless semiconductor companies in recent years. In 2018, a new record of $3.4 billion total investment was set, far above the previous record of $2.5 billion in the year 2000.

Figure 2.  Venture capital funded fabless semiconductor startups

Venture capitalists have been focused on social media and software companies over the last twenty years.  Where is all this new money going?  The answer can be seen in Figure 3. AI and machine learning have dominated the fabless semiconductor industry investment by venture capitalists since 2012 with $1.9 billion invested.

What kind of chips are being funded?  The largest share is in the area of pattern recognition.  Chinese investments in facial recognition chips developed at companies like Sensetime and Face++ constitute a very large share.  There are seventy-five other companies developing chips for pattern recognition that have been funded by venture capital.  These include companies focused on pattern recognition for audio, smell and other applications.

Figure 3.  Venture funded startups since 2012 by application. AI and machine learning constitute the majority of applications

Beyond pattern recognition, the largest share of new fabless semiconductor companies are being funded for data center analytics or edge computing.

Edge Computing
Intelligence historically flows downhill (Figure 4). In the 1960s, mainframe computers dominated our computing capacity.  Dumb terminals connected us to our mainframe computing power.  By the 1980s, minicomputers were well established as an intermediate computing layer between the user and the mainframe.  Twenty years later, the personal computer became the local computing resource.  In another twenty years, the current environment has evolved.  Large cloud-based server resources handle the heavy computing but in between us and the cloud is the fog made up of gateways that collect, aggregate and locally process data.  Beneath that layer are the edge nodes in the mist, collecting and pre-processing the data.  As time passes, the lower layers will inevitably gain more intelligence as semiconductor technology allows us to build more intelligence into the local nodes. Those nodes will become increasingly complex as they incorporate disparate technologies – analog, digital, RF, MEMs, etc.  (Figure 5).  This creates major design and verification challenges.  Most of EDA history is focused upon digital logic and memory.  Edge nodes may require mixed technologies.  Simulating digital logic connected to analog, RF and other technologies is not easy.  A whole new family of EDA tools is required.

Figure 4.  Intelligence flows downhill

5G Wireless Technology
In the next decade, wireless communication will move to the next generation of technology, 5G.  This transition is more significant than past generations.  It affects a wide variety of the infrastructure beyond consumer communications.  Significant impact will be felt in applications involving industrial control, non-real time automotive analytics, urban infrastructure and much more.

One of the great opportunities for the semiconductor industry is the increased number of base stations required to support the infrastructure of 5G and the larger number of antennas in a phone. The number of semiconductor components required will grow dramatically as the world builds a 5G infrastructure. Connectivity to the cloud makes a wide variety of capabilities possible, especially in the factories of the world.  Gateways, which already generate more than three percent of worldwide semiconductor revenue, will be needed.

This connected world will be dependent upon more semiconductors for communications and computing.  For many years the semiconductor industry measured its revenue from the computing and communications industries which were each about 35% of the total.  Now the two are indistinguishable.  Seventy percent of the revenue in the semiconductor industry comes from one or the other or a unique combination of both.

Figure 5.  Diverse technologies like digital, analog, RF and MEMs will be required as edge nodes become more intelligent

Automotive Applications
During the last ten years, sales of semiconductors for automotive applications has increased to about 12% of the total semiconductor market as the electronic content of vehicles increased.  Some traditional electronic functions like engine control will not be needed in electric vehicles but there will be new requirements as well as the continued growth of infotainment and automotive driver assistance (ADAS) that require electronic controls.

Figure 6.  As of June 2019, 463 companies have announced intent to introduce electric cars or light trucks.  211 companies have announced autonomous drive programs

The number of companies planning to build electric cars or light trucks has now grown to 463, more than half of which are based in China (Figure 6).  Two hundred eleven companies have announced autonomous driving programs.  This number of suppliers is not needed and many, or even most, will drop out as we move closer to production. Meanwhile, one would expect an incredible bubble in demand for automotive ICs followed by a rapid decline.

It’s likely that no more than a dozen companies will lead the way in supplying the complex image processing subsystems required for autonomous vehicles.  It’s difficult to predict which ones will succeed but likely that companies that have not been traditional automotive OEMs will make up most of the total.

Other Predictable Futures
Lots of other technologies offer promise for growth.  Quantum computing is interesting because it has some capabilities like encryption that are not solved easily through other means.  The time lag for technologies like this tend to be longer than the evolutionary ones but they will eventually emerge in some form.

The history of the semiconductor industry is driven by major new applications.  Waves of growth were initiated by defense electronics in the 1950s, mainframe computers in the 60s, minicomputers in the 70s, personal computers in the 80s, laptops in the 90s and wireless communications in the most recent two years. Each wave has been accompanied by the emergence of new semiconductor competitors followed by a shakeout that leaves one supplier dominant and shuffles the top ten ranking of companies by revenue (Figure 10 in Chapter 5).

At the same time, the semiconductor industry, like most industries, moves back and forth from standardized versus customized solutions.  This has been referred to as “Makimoto’s Wave” after Tsugio Makimoto, former CEO of Hitachi Semiconductor, who popularized the phenomenon.  As we move into the third decade of the twenty-first century, the semiconductor industry is moving into a customization wave.  Standard von Neuman computer architectures that operate on a string of standard instructions have dominated the computer and semiconductor industries.  Architectures like the Intel 808X and ARM RISC will continue.  Domain-specific architectures tailored for specific tasks like facial recognition are emerging. There will be dozens more as AI and machine learning usher in new opportunities.

What should we consider as the future possibilities for the semiconductor industry? As we showed in Chapter 4, the semiconductor industry is likely to grow through evolutionary means through about 2040 or when demands for lower power or higher performance usher in a new technology for information “switching”.  Carbon nanotubes, bio-switches, or many other possibilities could fill in the switching learning curve of Figure 5, Chapter 3.  Chances are that this “switch” will happen gradually as the need arises for a new application.  In addition, non-silicon materials like Gallium Nitride, Silicon Carbide and other materials will take on increasingly important roles driven by need for characteristics like larger band gaps, i.e. roles like power switching, microwave communications and existing ones like solid state lighting.

Just as steel is still a primary material for construction one hundred fifty years after the booming growth of the steel industry, semiconductors will be at the foundation of business and technology growth for a long time.  Those of us who participated in the last fifty years of exciting growth of semiconductors are still surprised when we see our “mature” industry generate another wave of growth to accompany an emerging application.  I’m confident that there will be many more to come.

Posted on

AI Hardware Summit, Report #1: Doing More to Cost Less

AI Hardware Summit, Report #1: Doing More to Cost Less
by Randy Smith on 09-26-2019 at 10:00 am
Categories: eSilicon, Events, Semiconductor Services

I recently had the pleasure of attending the AI Hardware Summit at the Computer History Museum in Mountain View, CA. This two-day conference brought together many companies involved in building artificial intelligence solutions. Though the focus was on building the hardware for this area, there was naturally much discussion around software and applications as well. The first session I want to summarize was presented by Dr. Carlos Macián, Senior Director, AI Strategy and Products at eSilicon.

When I saw an eSilicon presentation on the agenda, naturally I assumed it would be about their recently announced neuASIC™ IP platform. If you don’t know about that yet, you may want to read about their AI IP platform first. Instead, we were treated to a much broader presentation on controlling the total cost of ownership (TCO) of an AI hardware solution. The presentation was quite insightful and showcased just how much depth and experience eSilicon has when it comes to building these types of ASIC products.

TCO is an important concept. When deciding how to address the challenges of building a hardware solution for a specific AI application, one needs to understand how each decision affects the total cost of the product. Some decisions carry more cost in area (die cost), yield (die cost), effort (person-hours), quality (sales, reputation, returns, etc.), power (packaging and other costs) and so many other factors. The list of traits and their associated costs is quite long. Given that most companies should have a grasp of the common TCO drivers, this presentation focused on the key items to consider for state-of-the-art AI products.

From the slide above, you can see that AI designs for data centers have some familiar drivers that are exacerbated by the need to move to massive parallelism – hyperscale. Hyperscale computing refers to the systems and architecture in distributed computing environments that must efficiently scale from a few servers to thousands of servers. Hyperscale computing is used in environments such as big data and cloud computing – today’s massive data centers.

Carlos clearly explained the biggest challenges to AI hyperscale implementation, along with the enabling technologies that have been rolled out at several companies now. Recent announcements, such as Intel’s announcement at HotChips of their Lakefield processor built using Foveros 3D technology, are a clear sign that these technologies are available now. The challenge is to find a partner who understands all of these enabling technologies, something that eSilicon has already demonstrated.

The presentation then went on to focus on an example of solving these AI design challenges by utilizing one of the enabling technologies – 3D memory overlays. The presentation demonstrated if you stack parts of the solution vertically (e.g., xRAM, SRAM+IO, and compute) on different die in the same package that there are huge efficiencies to be gained. One dramatic gain is yield. Manufacturing several smaller die that can be stacked increases yield dramatically. In the example shown at the event, yield improved from 15.7% to 68.6%. This yield improvement provides a tremendous decrease in the cost of production and therefore a dramatic improvement in the TCO.

Despite the difficulties some will encounter in getting these hyperscale AI designs to function at a reasonable cost, I think eSilicon has shown it has the expertise to get them across the finish line. They also disclosed that they are already working with suppliers on the next set of challenges as the degree of scaling increases – new die bonding technologies, vertical signal density, thermal density, combined yield, and many others. I will be anxious to hear more on these items when eSilicon is ready to discuss them.

eSilicon seems well prepared to deliver AI hardware designs. You can learn more about their NeuASIC AI capabilities here. You can learn more about their 2.5D/HBM2 packaging solutions here. As I have mentioned before, as an IP vendor, I referred my licensees to eSilicon before where their success lead to us getting our clients to volume quickly. That is why I recommend them highly.