What is Next for GLOBALFOUNDRIES?

What is Next for GLOBALFOUNDRIES?
by Daniel Nenni on 04-04-2014 at 8:30 am

In response to changing industry dynamics, AMD announced in October 2008 a new strategy to focus exclusively on the design phase of semiconductor product development. To achieve that strategy, AMD partnered with Advanced Technology Investment Company (ATIC) of Abu Dhabi to create a new joint venture company designed to become the world’s first truly global contract manufacturer of semiconductors.

On March 4, 2009, GLOBALFOUNDRIES was launched as a new joint venture combining AMD’s leading-edge semiconductor manufacturing capabilities with the long-term financial backing of ATIC. This created a new global semiconductor manufacturing foundry with approximately 3,000 employees with AMD as its first customer.

In January 2010, the company announced the completion of its merger of operations with Chartered Semiconductor, a global semiconductor foundry based in Singapore. At the time, Chartered consisted of about 7,000 employees, mostly based at the company’s 6 fabs in Singapore.

Today, GLOBALFOUNDRIES is wholly owned by ATIC and is the world’s second largest independent semiconductor foundry. However, GF is still one fourth the size of number one foundry TSMC and faces stiff competition from the ever aggressive Samsung Foundry Division and the newly launched Intel Custom Foundry Business Unit. More recently, GF shuffled the executive staff and acquired a new CEO:

Santa Clara, Calif., January 6, 2014 —Building on the successful track record of its first five years in the semiconductor industry and its continued commitment to build out its global network of manufacturing facilities, GLOBALFOUNDRIES announced today, from its new offices in Silicon Valley, Sanjay Jha has been appointed as the company’s new Chief Executive Officer. Jha has served as CEO of Motorola Mobility Inc. and as the COO of Qualcomm Inc.

After spending 14 years at Qualcomm, Sanjay joined Motorola Mobility as CEO in 2008. Sanjay then sold Motorola Mobility to Google in 2011 for $12.5B with an exit package of more the $65M. Google then sold Motorola Mobility to Lenovo in 2014 for $2.91B. Yes Sanjay is a very clever man and he knows the fabless semiconductor ecosystem inside and out.

The $100B question is: What is next for GLOBALFOUNDRIES? The wild card here of course is Sanjay Jha. I do not know Sanjay personally (yet) but I do know people who know him and based on this, I’m wildly optimistic!

It is highly unlikely that Sanjay signed on to continue to stay the course at GLOBALFOUNDRIES. Being a second source or boutique foundry against the likes of TSMC, Samsung, and Intel makes no sense whatsoever. My guess is that Sanjay will go on an acquisition spree with the intention of building a major force in the semiconductor industry, absolutely. If I were Sanjay I would start with MediaTek and I will tell you why in the comments section. Acquiring the IBM semiconductor operations is also on the table I hear.

GF reads SemiWiki so offer your advice to Sanjay. This could be a real game changer, absolutely!

About GLOBALFOUNDRIES
GLOBALFOUNDRIES is the world’s first full-service semiconductor foundry with a truly global footprint. Launched in March 2009, the company has quickly achieved scale as the second largest foundry in the world, providing a unique combination of advanced technology and manufacturing to more than 160 customers. With operations in Singapore, Germany and the United States, GLOBALFOUNDRIES is the only foundry that offers the flexibility and security of manufacturing centers spanning three continents. The company’s three 300mm fabs and five 200mm fabs provide the full range of process technologies from mainstream to the leading edge. This global manufacturing footprint is supported by major facilities for research, development and design enablement located near hubs of semiconductor activity in the United States, Europe and Asia. GLOBALFOUNDRIES is owned by the Advanced Technology Investment Company (ATIC). For more information, visit http://www.globalfoundries.com.

More Articles by Daniel Nenni…..

lang: en_US


FinFET Custom Design

FinFET Custom Design
by Paul McLellan on 04-02-2014 at 8:30 pm

At CDNLive, Bob Mullen of TSMC gave a presentation on their new custom FinFET flow, doing design, and verifying designs. At 16nm there are all sorts of relatively new verification problems such as layout dependent effects (LDE) and voltage dependent design rules. We had some of this at 20nm but like most things in semiconductor, it gets harder with each process generation. I’m going to leave verification until another blog.

He wasn’t talking about digital SoC design, which broadly speaking is the same as before. You write RTL, synthesize the design, place and route it and then run verification. Mostly the tools take care of the hard stuff like double patterning. And whoever designed the standard cell libraries took care of all the complicated FinFET stuff. He was talking about custom and analog design where you do actual transistor level layout.


I’m sure you know what a FinFET transistor looks like these days. What you may be less aware of is that they have to be laid out in a sort of matrix. The fin is a fixed size and so the only thing you get to vary about a transistor is how many FinFETs you get to join up in parallel to build a wider transistor. This is often referred to as quantization. In planar devices we used to be able to vary the width and length how we wanted. Actually by the time we got to 20nm this wasn’t really true, the design rules were so restrictive that the transistors were also pretty much laid out in a matrix. To build them, for lithography reasons, the gate material needed to be laid out in parallel lines at the appropriate spacing and then a cut-mask used to split the gate lines up into transistors. So the length of the transistors was pretty much fixed but at least we got to pick the width. With FinFETs we used the same technique but we don’t get to pick the width arbitrarily, just how many fins are controlled by the same signal. So a FinFET design consists of rows of source/drain with rows of gate running orthogonally.


Above is a very simple representation. TSMC doesn’t let anyone see their layout except under NDA so this is actually Cadence’s generic FinFET process used for testing tools early in the whole process. And the planar transistor on the left is nothing like the last planar process at 20nm, it is more like how things looked at 90nm when lithography OPC constraints were a lot more forgiving.

On the right is a FinFET inverter. First thing to note is that the inverter seems to have three gates (red), which is true. Every transistor has to be terminated with dummy gates on either side. You can’t just cut off the diffusion by just ending the polygon like in the planar device on the left. You need to tie it off with a gate. This was actually true at 20nm too, which is one reason I said that the planar transistor was from an old process node. In the middle you can see the red hashed area, that is the cut mask that separates the P and N transistors.

The first thing TSMC did was build a capability into the PDK to build a “transistor” that took as input how many fins were to be used. It created the layout, including dummy gates and well boundaries.

Then they created a schematic migration methodology to automate much of the migration of designs from 20nm by picking appropriate fin-counts close to simply scaling a planar transistor to 16nm. The voltages are different, the PDKs are different, and the quantized nature of FinFETs needed to be taken into account. But when they were done they would have migrated:

  • Circuit symbols and schematics
  • Hierarchical design configuration view
  • Electrical & Physical design constraints
  • Functional behavioral modeling views
  • Testbench schematics and setups

However there is still no layout and the schematic is almost certainly going to need to be changed before the cell is finalized. The first step is thus to circuit simulate the schematic using estimated parasitics to get a starting point for getting to a layout.

The next step is rapid analog prototyping, to iterate between layout, extraction, circuit simulation and tweaking transistor sizes and layout constraints. The actual layout is automatically generated under the constraints. Every time the layout changes the parasitics change so hopefully the process converges reasonably fast.

Then onto verification. But that is a topic for another day.

If you have a Cadence account you should be able to find Bob’s CDNLive presentation here.


The Infamous Intel FPGA Slide!

The Infamous Intel FPGA Slide!
by Daniel Nenni on 03-11-2014 at 10:30 am

As I have mentioned before, I’m part of the Coleman Research Group so you can rent me by the hour to better understand the semiconductor industry. Most of the conversations are by phone but sometimes I do travel to the East Coast, Taiwan, Hong Kong, and China for face-to-face meetings. Generally the calls are the result of an event that needs further explanation or just a quarterly update. Again, as an active semiconductor professional I share my experiences, observations, and opinions so rarely will I agree with the analysts or journalists who rely on Google for information.

In 2003, Kevin Coleman founded Coleman Research to give investors a better way to access industry knowledge. Coleman helps thousands of clients get answers to their most critical questions, without leaving their desks. Rather than spending hours reading research reports, or traveling to meet people at conferences, we connect clients directly with industry experts, to hear immediate, relevant insights.


The Intel analyst meeting last November was full of surprises and resulted in a series of phone consultations. The Intel 14nm superior density claim slides were the most talked about and were absolutely crushed by TSMC, which I wrote about in “TSMC Responds to Intel 14nm Density Claims”. The other slide that caused a flurry of calls is the one above comparing Altera and Xilinx planar to FinFET. After talking to dozens of people (including current and former Altera, Intel, and Xilinx employees) I have concluded that this slide is an absolute fabrication. Get it? Fabrication? Hahahahaaaa….


I did a comparison of the Altera and Xilinx analyst meetings and found the slide above which supports my point. Clearly silicon does not lie so when the competing FPGA FinFET versions are released we will know for sure, but my bet is that Altera/Intel will lose this one. It also goes to my point that the transistor is not everything in modern semiconductor design and Intel’s claims of process superiority are a paper tiger when it comes to finished products.


There are thousands of FPGA and semiconductor process professionals reading SemiWiki so I’m hoping for a meaningful discussion in the comments section. If any of you would like to post a rebuttal blog I’m open to that as well. SemiWiki is an open forum for the greater good of the fabless semiconductor ecosystem, absolutely.

The most recent event that caused a flurry of calls was the JP Morgan Report: Meetings at MWC – Intel Mobile Effort Largely a Side Show, but Some Problems in Foundry a Concern. The press really had a field day with this one:

Some issues popping up with foundry business – we are concerned.Ourchecks indicate there have been some problems with Intel’s foundry effortscentered on design rules and service levels. It appears Intel is being inflexibleon design rules and having trouble adapting to a service (foundry) model. Our J.P. MorganFoundry analyst, Gokul Hariharan wrote today that Altera has re-engaged TSMC.

This resulted in a handful of tabloid worthy articles taking the JP Morgan report completely out of context:

Altera to switch 14nm chip orders back to TSMC, says paper Commercial Times, March 4; Steve Shen, DIGITIMES [Wednesday 5 March 2014]

While I appreciate the consulting business this generated I really do question the motives of Steve Shen. The first “Altera leaving Intel” rumor started HERE and I’m sure this won’t be the last but I’m still not buying it and neither should you.

More Articles by Daniel Nenni…..


Dr. Cliff Hou, TSMC VP of R&D, Keynote

Dr. Cliff Hou, TSMC VP of R&D, Keynote
by Daniel Nenni on 02-16-2014 at 9:00 am

This will be my 30[SUP]th[/SUP] Design Automation Conference. I know this because my first DAC was the same year I got married and forgetting how many years you have been married can cost you half your stuff. I have known Cliff Hou for half of that time and he has proven to be one of the most humble and honorable men I have worked with, definitely.

Cliff started at TSMC in the PDK group and produced the first TSMC Reference Flow which really was the starting point for the fabless semiconductor ecosystem (Grand Alliance) that we have today. Cliff then took over the TSMC IP group before becoming the Senior Director at Design and Platform which included the PDK, IP, and other design enablement Groups inside TSMC. In 2011 Cliff was appointed TSMC’s Vice President of Research and Development. Clearly Dr. Cliff Hou is rising star in the semiconductor industry and it has been an honor to work with him.

Cliff was our choice to write the foreword to the book, “Fabless: The Transformation of the Semiconductor Industry” as he and TSMC led this transformation. The foreword alone is worth the price of the book and I can’t wait to get Cliff to sign a copy for me at #51DAC where he will be keynoting:

Industry Opportunities in the Sub-10nm Era

The human thirst for connectivity and experience, as enabled by the electronics industry and the ongoing march of Moore’s Law, has already brought, and will bring even more, profound changes in way we interact with the world and each other. This profound enhancement of the human experience enabled by constant mobile connectivity, the Cloud, and sensors, brought to an ever widening worldwide audience, will bring untold opportunity to all of us here at DAC.

All of these changes demand continued chip and wafer-based scaling to deliver the power and performance necessary to enable wondrous, new applications. In less than two years we’ll be in production at 10nm, and shortly after 7nm, all made possible by a “Grand Alliance” of design ecosystem, equipment and material suppliers. At the same time, a new paradigm is being realized: heterogeneous silicon integration combining chips from multiple process technologies with 3D packaging to deliver compelling economics for a “System in a Si Superchip.”

New design techniques will be required for those applications becoming reality, including how 10nm, and 7nm will support those requirements, new manufacturing techniques, and the benefits they will provide. The introduction of 10nm and 7nm processes will alter today’s ecosystem while opening greater EDA and IP opportunities, and present new system and chip design challenges such as near threshold design, thermal and battery limitations, and 3D IC considerations.

IC designers, ecosystem providers and foundries have been committed to open innovation and mutually beneficial teamwork for many process technology generations, but success in the sub-10nm era will require unprecedented levels of collaboration and cooperation between all of us here at DAC. Our teamwork will drive industry progress, and the more we “collaborate to innovate,” the more successful our customers and all of us will become.

More Articles by Daniel Nenni…..

lang: en_US


Intel 14nm Delayed Again?

Intel 14nm Delayed Again?
by Daniel Nenni on 02-12-2014 at 9:00 am

From the sources in which I confirmed the last Intel 14nm delay, I just confirmed another. Intel 14nm is STILL having yield problems. Remember Intel bragging about 14nm being a full node and deriding TSMC because 16nm is “just” 20nm with FinFETs added? Judging by the graph, clearly FinFETs are not the problem here. Intel used a much more aggressive metal fabric to get better density which is challenging modern lithography methods.

“People in the trenches are usually in touch with impending changes early” ― Andrew S. Grove,Only the Paranoid Survive

Meanwhile, back at the fabless semiconductor ecosystem, 20nm is yielding ahead of schedule so TSMC will see revenue this quarter versus next. I would put the chances of TSMC realizing their forecast of 20nm providing 10% of 2014 revenue as being very good. Given the more cautious approach TSMC took to FinFETs, 16nm is also on track with tape-outs happening now. If all goes as planned, 16nm will ramp in 2015 as 20nm does in 2014.

TSMC expects 20nm to be 2% of Q2 2014 revenue so the ramp begins. Looking at the 28nm ramp, 20nm is expected to be 20-30% faster:

[LIST=1]

  • 28nm 2% Q4 2011
  • 28nm 5% Q1 2012
  • 28nm 7% Q2 2012
  • 28nm 13% Q3 2012

    Back to Intel; new Intel CEO Brian Krzanich committed 14nm for Q3 2013 which was later pushed out to Q1 2014 even though he held up a laptop at the Intel Developers Forum and boasted that 14nm was in fact on track. At an analyst meeting two months later he showed the slide above and said there were yield “challenges” that they are still working on. Well, from what I have heard, they are still working on it so the Intel 14nm ramp may be delayed yet again.

    The questions I have are: If this is true when will Intel disclose this new yield challenge? How much will it delay 14nm products? What about Altera? I’m sure delaying this type of bad news until the problem is fixed is best for damage control but I find this type of behavior not transparent and untrustworthy, just my opinion of course.

    Meanwhile the Intel pumping Seeking Alpha published an article, “Does Intel’s new CEO have what it takes?” This is pure entertainment. Thus far Intel management has made many mistakes that the author glossed over but have been covered in painful detail on SemiWiki. The lack of transparency started here with BK’s first conference call:

    Intel’s Q2 Conference Call
    Intel 14nm Delayed?
    Intel Is Continuing to Scale While Others Pause To Do FinFETs
    No Mention of 14nm at the 2013 Intel Developer Forum?
    Intel Really is Delaying 14nm Move-in. 450mm is Slipping Too. EUV, who knows?
    Intel Quark: Synthesizable Core But You Can’t Have It
    Intel Bay Trail Fail
    Yes, Intel 14nm Really is Delayed…And They Lost $600M on Mobile
    Intel’s Mea Culpa!
    Intel Bay Trail Fail II
    Intel Comes Clean on 14nm Yield!
    Intel is NOT Transparent Again!
    Why Intel 14nm is NOT a Game Changer!

    We write these articles from the trenches to set the record straight. We also write these articles as research for an upcoming book on Intel to chronicle the rise and fall and hopefully the rise again of the number one semiconductor company.

    More Articles by Daniel Nenni…..

    lang: en_US


  • Designing an SoC with 16nm FinFET

    Designing an SoC with 16nm FinFET
    by Daniel Payne on 02-11-2014 at 9:35 pm

    IC designers contemplating the transition to 16nm FinFET technology for their next SoC need to be informed about design flow and IP changes, so TSMC teamed up with Cadence Design Systems today to present a webinar on that topic. I attended the webinar and will summarize my findings.

    Shown below is a 3D layout concept of an ideal FinFET transistor, followed by the actual manufactured device which is rotated 90 degrees from the layout:

    Continue reading “Designing an SoC with 16nm FinFET”


    The Great Wall of TSMC

    The Great Wall of TSMC
    by Paul McLellan on 02-03-2014 at 5:27 pm

    TSMC doesn’t just sell wafers, it sells trust. It’s the Colgate Ring of Confidence for fabless customers. This focus on trust started at the very beginning when Morris Chang founded TSMC over 25 years ago, and still today trust remains an essential part of their business.

    When TSMC started, the big thing it brought was that it was a pure play foundry. It had no product lines of its own. Foundry had existed before, but it was semiconductor companies selling excess capacity to each other. This meant that the buyer of wafers was always vulnerable to the seller company being successful and needing that capacity and they would get thrown out. And that was without even considering that companies might be buying wafers from a competitor, sending them masks of their crown-jewels and trusting that nobody would try and reverse engineer anything.

    So when TSMC started, it brought the confidence that TSMC wasn’t going to suddenly stop supplying wafers since they needed the capacity for themselves, nor that TSMC was competing with them in the same end-markets. That is not to say that there never have been capacity issues: TSMC cannot afford to build excess capacity “just in case” any more than anyone else, so when businesses take off better than forecast or some other event happens, wafers can end up on allocation just as has always been the case in semiconductor, an inherently cyclical business.

    Not competing with its customers remains the case today (as, to be fair, it does for GlobalFoundries, SMIC, Jazz and other pure-play foundries). But it is not the case for Samsung, which is in the slightly bizarre situation of having Apple as its largest foundry customer while competing with it as the volume leader in the mobile market (and never mind the lawsuits). Samsung is large diversified conglomerate, almost a lot of different companies all using the Samsung brand-name. Samsung makes all the retina displays for iPhone too, and doesn’t even use them themselves. They are a huge memory supplier. Apple is rumored to be moving from Samsung to TSMC for its next application processor (presumably to be called A8).

    Intel has made a lot of noise about entering the foundry business but the only significant company that has been announced is Altera. And there are even rumors that they are thinking of going back to TSMC. But a company like Altera using Intel for its high end FPGA products might need 1000 wafers a month when a fab has a capacity of 50-100K wafers a month. It won’t “fill the fab”. It needs to get an Apple or a Qualcomm or an nVidia for that. But at least Altera can be confident that no matter how successful Intel’s other businesses are, at those volumes they are unlikely to be squeezed out, the amount of capacity they need is in the noise.

    The other area that foundries have had to invest is to create an ecosystem around them of manufacturing equipment and material suppliers, IP and EDA companies. This grand alliance has made a huge investment in R&D. In aggregate, it has invested more than any single IDM. As a result the Grand Alliance has produced more innovation in high performance, lower power, lower cost than any single IDM.

    At a modern process node deep cooperation is required. It is not possible for everything to be done serially: get the process ready, get the tools working on a stable process, use the tools to build the IP, start customer designs using the IP and the tools, ramp to volume. Everything has to happen almost simultaneously. This requires an even greater sense of trust among everyone involved, and the fact that changing PDKs means changing IP means redoing designs means inevitably an increased investment too.

    So TSMC has a competitive edge, the great wall of TSMC to keep out the barbarian hordes:

    • it sells confidence and trust, not just wafers
    • it does not compete with its customers
    • it has orchestrated a grand alliance to create an ecosystem around its factories that has made a bigger R&D investment than any single IDM.


    More articles by Paul McLellan…


    Why Intel 14nm is NOT a Game Changer!

    Why Intel 14nm is NOT a Game Changer!
    by Daniel Nenni on 02-02-2014 at 10:00 am

    On one hand the Motley Fool is saying, “Intel 14nm could change the game” and on the other hand the Wall Street Cheat Sheet is saying, “Intel should shut down mobile”. SemiWiki says Intel missed mobile and should look to the future and focus on wearables and in this blog I will argue why.

    Let’s look back to 2009 when Intel and TSMC signed an agreement to “collaborate on addressing technology platform, intellectual property (IP) infrastructure, and System-on-Chip (SoC) solutions.” Intel and TSMC ported the Atom Core to 40nm and offered it to more than 1,000 of TSMC’s customers:

    “We believe this effort will make it easier for customers with significant design expertise to take advantage of benefits of the Intel Architecture in a manner that allows them to customize the implementation precisely to their needs,” said Paul Otellini, Intel president and CEO. “The combination of the compelling benefits of our Atom processor combined with the experience and technology of TSMC is another step in our long-term strategic relationship.”

    Unfortunately this venture was a complete failure for business and technical reasons and was put on hold a year later. I was a frequent visitor to Taiwan at the time so I had a front row seat to this one. The excuse was that you can’t just flip a switch and be successful in the mobile market, meaning that Intel’s Atom effort will require patience and persevance. Fast forward to 2012:

    “We are moving Intel[SUP]®[/SUP] Atom[SUP]TM[/SUP] processors to our leading-edge manufacturing technologies at twice our normal cadence. We shipped 32nm versions in 2012, and we expect to launch the 22nm generation in 2013, and 14nm versions in 2014. With each new generation of technology, we can boost performance while reducing costs and power consumption—great attributes for any market, but particularly for mobile computing.”Our Mobile Edge by Paul Otellini, Intel 2012 Annual Report.

    Clearly that did not happen at 22nm with Intel literally GIVING AWAY 40 million 22nm SoCs to get “traction” in the mobile market. And Intel 14nm SoCs are delayed until 2015 which will be in lock step with the next generation of 14nm ARM based processors from QCOM, Apple, Samsung, and a handful of other fabless SoC companies.

    As a stopgap measure to fill their new 14nm fabs, Intel dipped its toe into the shark infested foundry business waters. Unfortunately the only taker was Altera and their 14nm wafer demand is 3+ years out and the volume is a fraction of what is needed to keep a fab open. Intel is lucky to have only lost a toe here as they also risked exposing the secret manufacturing sauce they are famous for. Intel then shuttered fab #42 which could have been filled by foundry customers.

    Let us not forget the other multi-billion dollar Intel forays away from their core competency: McAffee? Intel TV? Can someone help me complete this list in the comment section please? There are just too many for me to remember.

    That brings us to where we are today: Intel still does not have a competitive SoC offering and time is running out. I strongly suggest that Intel take note of Google’s recent move out of the Smartphone business selling Motorola Mobility to Lenovo:

    The smartphone market is super competitive, and to thrive it helps to be all-in when it comes to making mobile devices…..Larry Page Google CEO.

    If Intel is going to go all-in I strongly suggest Intel focus on Quark and the wearable (embedded) market. Mobile has hit commodity status and is moving way too fast for a semiconductor giant to keep up (TI already gave up their mobile SoC business). Intel has had a historically strong position in the embedded market and it is time for them to get back to a business they truly believe in, absolutely.

    More Articles by Daniel Nenni…..

    lang: en_US


    RTL Sign-off – At an Edge to become a Standard

    RTL Sign-off – At an Edge to become a Standard
    by Pawan Fangaria on 02-01-2014 at 10:00 am


    Ever since I have seen Atrenta’s SpyGlass platform providing a comprehensive set of tools across the semiconductor design paradigm, I felt the need for a common set of standards to evolve for sign-off at RTL level. Last December, when I read an EE Times articleof Piyush Sancheti, VP, Product Marketing at Atrenta, where he talks about a billion gate SoC design, shrinking market windows, and design cycles to the level of 3-6 months, I was looking for an opportunity to talk to him in a broader sense on how RTL level design paradigm is proliferating and what we can see in future. This week I had a nice opportunity talking to him face-to-face in Atrenta’s Noida office. Here is the conversation –

    Q: SpyGlass is primarily providing a platform for designs at RTL and for sign-off at that stage. What has been your experience so far?

    In today’s SoC design environment, you have size, scale and complexity of advanced nodes being the prime factors. Most of the SoCs use several soft IPs, configurable at different levels, and some hard IPs as well. Iterative design closures do not serve the purpose for such large designs. Add to it very short market windows; there is another level of market segment coming up for Internet-of-Things, that has very short turn-around-time in the order of 3 months. RTL sign-off has become a need today to answer this faster design closure with lesser cost.

    So, to answer in short, our leading edge customers are executing on RTL sign-off and are happy to see the value in it. Last year was the best year for us in terms of business and growth and we are looking at a bright future from here.

    Q: Considering the amount of IP re-use and sourcing from third party for SoC design, a standard RTL sign-off criteria can help in reliable IP exchange as most of the IPs are sourced at RTL level. Your comments?

    Yes, definitely, at the top level an SoC can have just connectivity between many IPs connected through glue logic. So, quality of the SoC will depend on the quality of IPs and therefore a standard criterion must be there for IPs, internal or external. We have been working with TSMCon a standard for soft IP qualification.

    Q: That’s quite encouraging. Looking at your talk in EE Times about billion gate SoCs becoming a reality, I can definitely see that RTL sign-off is a must. But do you see common standard RTL sign-off criteria or rather RTL coverage factors evolving across the industry for the overall semiconductor design?

    Yes, it’s required. Even if all IPs on an SoC are qualified, it doesn’t guarantee the quality of the SoC. What if there is a clocking scheme mismatch between IPs? Even at the connectivity level between IPs, we need to look at the common plane issues, consistency, synchronous versus asynchronous and the like. So, a standard at SoC level sign-off is again a must for the industry. And we are working at it, along with some of our leading customers; it depends on a majority of the design houses adopting this path. It will take time to break that inertia; people will realize that this change in methodology is needed when they are no longer able to continue with the same old methodology.

    We have talked about the problems so far, let’s talk about some solutions. We now offer a smart abstract model concept for blocks in SoC design. RTL sign-off can be done at a hierarchical level; this has very fast turnaround. This is now in use in some of the most complex SoC designs with multiple levels of hierarchy. We have seen amazing results in performance, capacity, memory utilization, number of violations etc. We are talking gains that are in one or two orders of magnitude. So, we definitely would be interested in evolving the common standard for SoC sign-off at RTL.

    Q: What all should get covered in RTL sign-off?

    It’s across various design domains; clocking, testability, physical, timing, area, and power. Rules to avoid congestion and ensure routing completion such as fan-in, fan-out, mux sizes and cell pin density. On the timing side, there is logic depth, CDC, clock gating etc. Similarly there are rules for power and area. We have about 300 rules of the first order. These have broad applicability across a wide range of the market segments.

    Q: RTL sign-off is a must at the beginning of an SoC design and a post layout sign-off at the end. Do you see the need for any intermediate level of sign-off such as post floorplan level?

    Yes, SoC design needs a continuous monitoring at each stage. Quality and sign-off is a culture which must be exercised at each stage as the SoC passes through the design phases such as floorplan, placement and so on. By doing sign-off at RTL, one can get to design closure much faster, more productively and at lesser cost. As we pass through lower levels of design, the cost and iteration time increases. The other advantage at RTL signoff is that it minimizes iterations at lower levels. Overall it can reduce the design schedule risk by 30-50%.

    Q: Do you see a possibility of leading organizations working at RTL, joining together to define a common standard for RTL sign-off of IPs and SoCs for the semiconductor industry? Can Atrenta take a lead? Who should own the standard?

    As I said earlier, we are already working with TSMCand some of our other leading customers on this. We would be very interested in a common standard evolution which can benefit the whole semiconductor design industry. However, it needs about 10-12 major players from the design community, foundry and EDA to get the ball rolling. Eventually it will become a success only when majority of the semiconductor design community embraces it, as we have seen in other spaces. At this moment, we are not limited by capability; we are limited by the number of users which need to be large enough to provide that kind of momentum.

    So, yes we can give it a start, mature it, but going forward some standard body should own it. It may be a new standard body or any of the existing one, we have to see.

    Q: How far from now do you see that standard evolving?

    I guess it should take minimum 18-24 months from now. It will not fly until we have a critical mass of the community starting to use it.

    I felt extremely happy after talking to Piyush, especially on learning that what I was thinking is already in progress. This was one of my best conversations with industry leads. I really admire Piyush’s thought process when he said, “we are not doing it on our own. We continuously learn from our customers and partners who provide us the right direction to do things better in this challenging environment and change the ways that can lead to better productivity.” Let’s watch what’s there in store for future.

    More Articles by Pawan Fangaria…..

    lang: en_US


    The Changing Semiconductor Foundry Landscape!

    The Changing Semiconductor Foundry Landscape!
    by Daniel Nenni on 01-29-2014 at 8:00 am

    The foundry landscape is changing again and it is definitely something that should be discussed. There are some people, mostly influenced by Intel, that feel the foundry business has hit the wall at 20nm which couldn’t be further from the truth. After spending 30 years working in Silicon Valley, I have experienced a lot of change which is why I founded SemiWiki.com and co-authored a book on the fabless semiconductor revolution. Chronicling this change and looking towards the future is for the greater good of the semiconductor industry, absolutely.


    A big change happened at 28nm, when TSMC was the only foundry to yield, which resulted in wafer shortages and fab capacity issues. Of course TSMC did not initially build capacity for 90% market share. What semiconductor company would (with the exception of Intel)? Fabless companies such as Qualcomm, Broadcom, and Marvell that were used to multiple manufacturing sources were limited to a single source at 28nm which was not a comfortable position for them at all. Pricing and delivery is everything in this business thus the multiple manufacturing source business model. As it stands today, 20nm looks to be the same with TSMC in a dominant market position.

    The top fabless companies will make a correction at 14nm and use both TSMC and Samsung for competitive pricing and delivery. There really was no other choice since GlobalFoundries does not have the capacity yet to source a QCOM or Apple and Intel 14nm failed to make a passing foundry grade. With the exception of Altera, NONE of the top fabless semiconductor companies will use Intel at 14nm, which is one of the reasons why the Intel fab #42 in Arizona is being shuttered, my opinion. If fabless companies had the choice between Samsung/Intel and GlobalFoundries they would chose GF without a doubt. Working with an IDM/foundry that competes with you is a last resort for sure.

    This change is of great help to the fabless semiconductor ecosystem in regards to jobs and design enablement (EDA and IP for example). Due to ultra-strict security measures and process differences it will require many more engineers, tools, and IP to manufacture at both TSMC and Samsung at 14nm. This cost of course will be offset by cheaper wafers due to the pricing pressure competition brings.

    If you want a more detailed understanding of the changing foundry landscape there are three very good sources of information:

    [LIST=1]

  • IC Insights’ McLean Report
  • GSA 2014 Foundry Almanac
  • Me

    Why me? Because pound for pound I have access to more reports, attend more conferences, and talk to more semiconductor people than anyone else in this industry, believe it. I am connected to 17,962 semiconductor professionals on LinkedIn so if I don’t know the answer to your question I most certainly know someone that does. Generally I make people buy me lunch for a discussion on the foundry business but now that my book “Fabless: The Transformation of the Semiconductor Industry” is out, if you buy the book I would be happy to take your call or email and answer whatever questions you may have. Connect with me on LinkedIn, if you haven’t already, and let’s talk.

    More Articles by Daniel Nenni…..

    lang: en_US