Tuesday, July 31, 2007

Shifts in top 20 global semicon rankings

If the recent preliminary results released by IC Insights is anything to go by, there have been some movements among the top 20 semiconductor companies of the world during H1-2007. This is best illustrated by the table below.



While the top three -- Intel, Samsung and TI, retain their positions, ST and Toshiba have exchanged the next two positions, as have Hynix and TSMC, while Renesas remains at no. 8!

Freescale has taken a big drop from no. 9 to no. 16, while Sony, NXP and NEC gained one place each. Infineon has climbed back up to no. 12, from no. 16, while Qualcomm occupies the no. 13 position, up from no. 17. AMD dropped two positions, from no. 13 to no. 15.

Will the semicon industry see a tight year ahead? As per reports, IC Insights said that there should be a "noticeable seasonal rebound" in overall IC demand beginning in September 2007, which may cause "significant changes" in the top 20 semiconductor ranking in the second half of 2007. Wait and watch this space!

Thursday, July 26, 2007

Challenges for IC industry and Dr. Gargini's lessons

Those who were fortunate enough to attend the recently held SEMICON WEST in San Francisco had the pleasure of attending a great session on the past. present and future challenges for the IC industry by Dr. Paolo Gargini, Director of Technology Strategy at Intel.

For those who may not have the time to read this article, here's a snippet of what Dr. Gargini had to offer. The first lesson, he said, was that, "Something right may still happen even when everything seems to be going wrong.” However, challenges that were continuously posed also needed to be addressed at the earliest, in order to keep moving forward.

Many also predicted several times that some limit would be reached – that chip development would never get below ten nanometers, or below five nanometers. However, the industry was able to produce components with a gate oxide at about 1.2 nanometers.

Later, the game changed to scaling, which led to Gargini’s second lesson – “Predictors of engineering limits have always been proven wrong by the right improvements.” Shrinking silicon technology of the 1990s kept the industry moving forward.

His third lesson was, “It would be wrong to believe that the right fundamental limits don’t exist.” The fourth lesson was, “It is wise to look for the right solutions before things start going wrong.” Dr. Gargini recalled how Intel had announced that 45nm generation was ready. That it included high-k metal gates was only disclosed this January giving Intel the time to work on yields, enhancements, reliability problems, etc.

Gargini quoted Gordon Moore as saying that introduction of high-k metal gates was the single most important innovation in semiconductor manufacturing of the last 40 years!

Interestingly, a few weeks ago, the PULLNANO Consortium announced breakthrough results for 32nm/22nm. Among other things, the PULLNANO consortium has fabricated a functional SRAM using innovative MOS transistors whose device architecture differs significantly from that of transistors used in 45nm technology node.

These transistors are built using a low power consumption approach based on Fully Depleted Silicon On Insulator (FDSOI), coupled with a gate stack composed of a high-K gate dielectric and a single metal electrode stack. This is said to be the first time that such a compact SRAM cell has been fabricated using FDSOI, high-k dielectric and metal gate together.

Coming back to Dr. Gargini, his fifth lesson was, "It would be wrong to delay taking action and not do the right thing at the right time." According to him, there was a need for re-examining the opportunities for reviewing “old” theories and techniques that didn’t work on silicon.

Friday, July 13, 2007

ISA initiatives bode well for semicon

When the India Semiconductor Association (ISA) started functioning from way back in Nov. 2004, it had three main objectives. These were: bringing the Indian semicon industry together; ensure that the Indian government paid enough attention to this sector; and finally, putting together a semicon policy.

ISA has since come a long way. S. Janakiraman, president and CEO – R&D Services, Mindtree, and chairman, ISA, says that the ISA is now transitioning into putting systems and processes in place.

The ISA also has three special interest groups, focusing on market research, technology and talent. The market research aspect is more in terms of reference data and in terms of opportunities.

As far as technology aspect is concerned, mixed-signal is fast becoming a reality. The process geometry of semicon has been shrinking from 90nm to 65nm and even 32nm.

Design and manufacturing used to be fairly independent earlier. However, they are fairly coupled now. All of these changes need to be addressed, according to Jani Sir. The ISA SIG on technology would be studying and recommending how to implement, keep track and stay ahead of technology.

Talent is an extremely critical area! The depth of knowledge is in danger of drying out soon! ISA feels that there is a need for a lot of faculty retraining and change in curriculum. The industry also needs to collaborate with the academia.

To address this, ISA has collaborated with the Visvesvaraya Technological University (VTU), Karnataka’s nodal body for engineering education to form the the ISA-VTU understanding. This is a comprehensive initiative, which seeks to catalyze talent generation through enhanced industry-academia interaction. The VLSI Society of India (VSI) will partner and support ISA on this initiative in specific areas.

Established in 1998, Visvesvaraya Technological University (VTU) is headquartered at Belgaum. Presently there are 120 engineering colleges affiliated to VTU. Of these, five colleges offer a post graduate degree in VLSI design and embedded software.

Key aspects of the ISA-VTU initiative are as follows:

* Creation of a Semiconductor Research Consortium-like forum for India that would foster industry-oriented research;
* Facilitation of software acquisition in identified educational institutions;
* Create and support a repository of student projects;
* Faculty development through increased industry interaction and exposure; and
* Augmenting the curriculum evolved by SMDP/ VSI/ Universities in practicals and projects

The ISA also has another program, called the Si-Quest. The Si-Quest is a pioneering semiconductor campus awareness program aimed at talent generation from a broad-based quality talent pool in India. Its nationwide campus coverage, coupled with various awareness campaigns, will attract more and better candidates towards semiconductor industry. All of these bode well for the industry.

Wednesday, July 11, 2007

Paradigm shift indeed in semicon

Going through an article written by Dr. Wolfgang Ziebart, Member of the Management Board, President and CEO, Infineon Technologies, in Financial Times Deutschland, one cannot help but appreciate the great paradigm shift that has indeed taken place in the semiconductor industry.

The article titled: A paradigm shift in the semiconductor industry: Could this be the end of Moore’s Law? focuses on how changing technologies are indeed making life difficult for most market players to keep pace with all those changes.

The technical possibilities for shrinking chips have far from reached their limits at 65nm. There are preparations already on for 45nm and development work for 32nm has already started!

When PULLNANO announced breakthrough results for 32/22nm, did it surprise many? The PULLNANO consortium has fabricated a functional SRAM using innovative MOS transistors whose device architecture differs significantly from that of transistors used in 45nm technology node.

It talks about a compact SRAM cell that has been fabricated using FDSOI (Fully Depleted Silicon On Insulator), high-k dielectric and metal gate all together.

PULLNANO has demonstrated that the material and integration schemes used in 45nm generation can be modified to provide a robust solution at 32nm. It has also proposed an innovative new architecture that could provide even higher performance at 32nm and 22nm, using the so-called 'air gap' technique.

I do remember Dr Pradip Dutta of Synopsys India telling me about two months ago about 32nm, during a course of a telephonic conversation about ndia's moves in the semicon space.

After that conversation, I was wondering how quickly semicon technology had started to move. How quickly, from a has-been all these years, India was suddenly emerging as a semicon base to reckon with. How quickly, the geometry had moved from 90nm to, now, 22nm!

Dr Dutta also added that there would be more emphasis in India on doing high-end designs. In fact, whether it is frequency, number of gates, high complexity, etc., all of those would be driven by applications. India is now ready for doing high-end complex designs.

Indian companies are now definitely excelling in the design services sector. Some of them have also grown significantly. It is believed that they have also taken a load off the international design services company. Some of these Indian companies are now also developing their own IPs -- a paradigm shift in itself!

Monday, July 2, 2007

Case for low-cost FPGAs

It is said that there is an increased adoption of programmable logic in more applications. And that, development costs of alternative technologies such as ASIC and ASSP continue to rise. Does this make a case for low-cost FPGAs? Altera seems to believe in this.

According to the company, when Altera created the low-cost FPGA market with the introduction of its Cyclone FPGAs, the consumer product market in Taiwan became a primary focus.

Its contention is worth noting. Needs of consumers change rapidly. Low-cost FPGAs are , in this scenario, said to be the best vehicle to reduce time-to-market and bring a product to market quickly, besides go in for high-volume production immediately, if required.

According to the company, changes to product specifications can occur via FPGA, even as a product is in the market. There's probably no need to to do a redesign. Now, this would save significant time and cost, and allow market advantage.

We have just done reports on Altera, who have introduced the Arria GX family of low-cost, transceiver-based FPGAs that support PCI Express (x1 and x4), Gigabit Ethernet and Serial RapidIO standards at speeds up to 2.5Gbps.

Close by, Xilinx allows expanding FPGA development options with its ISE 9.2i design tools. This is said to have easy-to-use, built-in tools and wizards that make I/O assignment, power analysis, timing-driven design closure, and HDL simulation quick and intuitive.

The ISE 9.2i release claims to reduce memory requirements by an average of 27 percent, while providing expanded support for MS Windows Vista, XP x64, and Red Hat Enterprise WS 5.0 32-bit and 64-bit OS, respectively.

Definitely, both are worth a look, as the low-cost FPGA scenario seems to hot up!

Sunday, June 10, 2007

Cogs in the wheel

In engineering, it is imperative that all cogs of the wheel come together, so that the wheel rolls smoothly. Similarly, it is imperative that all key IT processes in an organization gel together and work as one.

Imagine the nightmare that enterprises, small and large, would have to go through should this did not happen!

The first basic IT asset is your company's network, or the intranet. We have seen several times that a company's network's down for some reason and mails can't be sent or received.

In such cases, the organizations or the enterprises who are 'stuck' with this situation, are literally crippled. Mails can't be received, mails can't be sent out, important mails are missed, business-critical processes are waylaid, and so on and so forth. I've been part of this nightmare several times.

Once, the undersea cable snapped during my stay in a company. I don't need to add the problem we had to face for at least half a day, as service providers worked furiously to rectify the cable and restore normal service.

Some advocate satellite as the best medium for managing data transmission. Maybe! Some others cite wireless. Perhaps!! Then, I hear from many that there are issues related with security and storage. However, those would only come into play once your basic network is operational smoothly.

In telecom, they have something called five nines, or 99.9999, which means the network is up and running for this percentage of time! You've noticed how people go beserk and start cursing their phones or the network, should they fail to receive a network, or are unable to connect to the network!

What they don't know or realize is the hard work that's involved in setting up, maintaining and operating a network! It's similar to what sometimes happens in offices when the network breaks down and we are unable to send/receive mails.

Maybe, it would be prudent to first manage the internal network as best as possible, before moving on to bigger, better things. The cogs in the wheel got to move smoothly.

Wednesday, May 30, 2007

Extreme targeting Ethernet networks

Extreme Networks recently launched a host of new Ethernet solutions aimed at global enterprises and service providers at its Partner Conference in Bali.

According to Mark Canepa, CEO, the company provide secure, converged Ethernet networks for enterprises and global service providers. The positioning is in the Ethernet space and Extreme does not worry about any other evolving or legacy technology for that matter. The solutions aim solve tough network challenges with meaningful insight and unprecedented control that simplifies networking operations.

Extreme announced a range of new solutions as well. Its BlackDiamond 12802R carrier Ethernet switch enables a wider range of metro deployments. It also announced some other new solutions as well. These include:

The Summit X250e Series, which has powerful application support in a 10/100 package; ExtremeXOS 12.0, which is a consistent, easy-to-manage stacking on every ExtremeXOS fixed configuration switch; Universal Port Manager, which allows simple automation for the entire enterprise policy lifecycle; and Summit WM200, Summit WM2000, which promise to deliver the future of wireless convergence today with modular, integrated WLAN.

Extreme's target markets for the year include WLAN, carrier Ethernet switches, security and L2+L3 Ethernet switches. The company's mission has been to provide secure, converged Ethernet networks for enterprises and global service providers who have complex networking needs.

Partnering with the channel is an essential part of Extreme Network's plan for success in Asia-Pacific. While Extreme builds great products and solutions, it also understands the network's big picture and the future path of technology. It extends the value of the infrastructure by developing key technology partnerships and helping customers learn how to use them.

The channels build great customer relationships and deep understanding. They, in turn, are able to understand the customer's big picture, and their future aspirations and plans. The channel extends the value of the customer's network by assembling the key components and delivering valuable services.

Monday, May 28, 2007

Next-gen PCB design with Allegro

Cadence was kind enough to discuss its Allegro solution with me recently. The solution aims at enhancing the productivity for next-generation PCB designs.

According to Steve Kamin, Group Director, Cadence's Allegro systems division, customers face constraints such as higher frequency, power consumption, pin counts, packages, etc. The PCB designers also face challenges such as decreasing hole sizes, hole diameter, etc.

Features of the Allegro include GUI (graphical user interface) modernization, context sensitive editing paradigm, color and visibility improvements, integration of physical and spacing constraints into the Constraint Manager, and an interactive planning and global routing, respectively.

In a PCB design flow, Cadence has focused on usability and productivity. The Allegro has a modernized GUI. All of the tool bars have been completely refreshed. Cadence has also added foldaway windows.

Cadence has added context-sensitive editing and included an open GL graphics engine that improves the visibility of objects and components on a board. This release can handle all sorts of intricate designs.

It has also addressed the physical and spacing constraints, which has been included that into the Constraint Manager. This is a cockpit, a spreadsheet-based tool, which manages all of the properties. Cadence is selling this solution to OEMs, ODMs, PCB design services companies, etc.

The Allegro solution has an interactive planning and global routing facility. Cadence has also added an entry-level product in the OrCAD PCB Designer Basics. Its GRE is a new, next-generation technology, which has two new elements -- the interactive flow planner and the Global Route Environment.

The GRE cuts to 1-2 iteration of routing. Customers can control the direction of the routes. Fabricators sometimes have had to kill the design as they could not control the routing or it was not possible. Hence, Allegro takes care of this issue. It saves significant time, from 15 months to about three months.

Allegro facilitates an improved design creation and simulation as well. The Allegro System Architect has differential pairs support and improved schematic generation. Other features include physical and spacing constraints in Allegro Design Entry HDL, performance and convergence improvements in Cadence Pspice and Allegro AMS Simulator, etc. Cadence has improved the automatic schematic generator. Also, in the analog simulation product line, it has added the automated convergence capability.

Finally, the company has added Cadence Help across all of its products. This allows cross-linking across all Cadence's tools. Customers can also add their own content. Cadence integrates third-party thermal integration tools into the products.

Monday, May 21, 2007

Chipping away at wireless USB

I happened to meet up with MosChip Semiconductor Technology, a well known fabless semiconductor company, which is listed on the BSE (No. 532407). What attracted me was the fact that MosChip is said to be focusing on wireless USB. as well.

It has been concentrating more on the digital side and is talking with a company for providing the radio (analog) chip, for a licensing agreement. As and when the market for wireless USB picks up, MosChip would be ready and waiting! Now, this is great! An Indian company, chipping away at the wireless USB market, waiting for its time to pounce and capture it!

Its engineering team is capable of designing end products that work across different operating systems and multiple platforms of various hardware configurations. MosChip has engineering expertise in PCI, USB, IEEE 1394, serial–parallel, SoC, device drivers, embedded firmware, engineering and reference platforms, PLL, transceivers, and A/D and D/A.

It is also developing custom products and IP for customers. Although Moschip is currently working on USB, PCI and IPSec, it is now going toward the SoC, according to its CEO, Ram Reddy.

The company adds application packages as software platforms are required for people to plug in the board. MosChip has 45+ software engineers who work on both embedded and platform to support such activities.

MosChip can give the complete object code to companies and source code to bigger companies with an agreement in place. It focuses mostly on OEM and ODM companies in the Far East. MosChip has strong distribution ties in place in China, Hong Kong, Japan and Taiwan.

Moschip, being a fabless company, contracts work to UMC, TSMC, etc. As and when, proposed fabs comes up in Hyderabad, Moschip would partner with any one of them. MosChip also announced the MCS8140, a highly integrated network processor. This network processor can be used with USB, multiple servers, etc., rather, with everything that is on IP. A maximum of 16 devices can be supported on the USB ports via IP. It is a single-chip solution.

Friday, May 18, 2007

Indian semicon industry: Give it time, you will see the results

Continuing from my previous blog, I'd like to mention here that Gartner, as well as some others, do point out that the Indian semiconductor industry is still in a nascent stage. We are all well aware of that!

A semicon policy has recently been announced to attract investments in semiconductor manufacturing. About five players announced plans and some have submitted proposals to the government. It will be at least another two years before things start rolling.

S. Janakiraman, president, ISA, and president & CEO - R&D Services, Mindtree, elaborated that the Indian government had, for the first time, come out with an exclusive policy for the semiconductor industry. This policy should provide significant boost to the manufacturing sector.

The ISA had closely worked with the Indian government to structure the policy to meet the expectations of the industry, while protecting the countries interest in terms of the right size of investments as well as right technologies being deployed.

The impact of the policy on the manufacturing sector will take time, but when it does becomes reality, it will create significant positive impact for the semiconductor sector.

There is also a clear interest by all leading IDMs as well as fabless semiconductor companies in looking at India as a destination for complex design sourcing, as well as increase their market share in a growing market.

Manufacturing of electronics has seen remarkable progress in the last 18 months with majors such as Nokia, Motorola, Flextronics and Foxcon now investing in India. Similar thrust in investments in semiconductor fabrication, test and packaging is not far away.

Technologies are fast evolving and mixed signal is becoming reality. Geometries of semiconductor technology is also shrinking, from 90nm to 55nm and even 35nm. Not only that, India is now in a position to talk about a multilayered chip or a 3D chip.

Pradip Dutta, president, Synopsys, believes all local and global majors based in India are poised for growth. Global majors will likely continue to leverage on the talent available. There will likely be even more emphasis on high-end design as well as determining the go-to-market strategy.

Whether it is the frequency, number of gates, high complexity, etc., all of those would be driven by applications. India is now ready for doing high-end
complex designs. All of this really augurs well for the Indian industry.

Kamal Aggarwal, VP - Marketing and Strategy, SoftJin Technologies, told me that the semiconductor market, globally, is likely to grow close to 10 percent during 2007. In India, it is likely to grow much faster due to the high growth in electronics consumption, maybe, around 30 percent.

Some observers feel that fabless is the best way to go for India, in order to tap our engineering talent in chip design and let the foundries in Mainland China and Taiwan give India the best wafer prices and help us keep up with ever reducing geometries.

These countries have the infrastructure for this R&D and we (India) have the engineering talent. Yes, I agree that this is a separate issue. There may not be several fabs as many think it would be. It'd probably be a mixture of maybe four or five fabs (my estimate) and lots of fabless companies, along with the ancillaries.

India should also encourage fabless semiconductor chip companies (product companies), give them concessions and tap into the US $300 billion semiconductor component market as well. All of this will happen, just give it some time folks!

Indian semicon industry: Good start's there for all to see, ain't it?

Our telecom and IT minister, Hon'ble Dayanidhi Maran resigned late last week. It prompted a friend to ask me whether this move would derail India's semicon policy and the path the industry had taken.

Of course not! We've made a good start, and there's still some way to go. The former minister has put us on a good path and the Indian semiconductor industry has been taking the right steps to figure among the top nations soon.

Take a look at some statistics. According to the India Semiconductor Association (ISA), India’s total semiconductor consumption in 2005 was about US $2.8 billion, met chiefly through imports. This demand is likely to exceed US $36 billion by 2015.

Electronics -- whether it is office automation, consumer electronics, medical electronics, telecom or industrial automation, will require more of the semiconductor industry in India.

The major end user segments have been communications, IT and consumer electronics. Together, the important product sub-categories that would drive the semiconductor market in India are mobile handsets, wireless equipment, especially BTS equipment, which is gaining momentum, set-top boxes, and smart card terminals.

With its growing middle class population of nearly 400 million people which will only increase over time, India’s electronic equipment consumption which was estimated at around US $28.2 billion in 2005, is likely to reach US $363 billion by 2015 growing at around CAGR 30 percent. The Indian electronics equipment domestic production was US $10.99 billion in 2005 and projects an opportunity to touch US $155 billion in 2015.

Statistically speaking, the Indian semiconductor design industry, comprising of VLSI design, board design and embedded software companies, has design companies across Bangalore, NCR Delhi, Hyderabad, Chennai, Pune, Ahmedabad and Goa. All of the global top ten fabless design companies also have India operations and 19 of the top 25 semiconductor companies are already present here as of early May 2005.

The Indian semiconductor design industry (i.e. VLSI, hardware/board, embedded software) had a turnover of US $3.2 billion in 2005 with an engineering workforce of around 75,000. It is estimated to reach US $43 billion by 2015 and provide jobs to 780,000 professionals with a CAGR of around 30 percent for this period.

The ISA-EY Benchmarking Study 2007 benchmarks the semiconductor design sector in India with peer countries on factors of importance to the sector and identifies important high level focus areas and actions to facilitate sustainable growth. Peer countries are Canada, China, Czech Republic, India, Israel, Taiwan, UK and the USA.

Some of the main findings that clearly position as India as an attractive design hub are:

* Availability and scalability of talent
* Quality of talent
* Quality of technical education
* Talent cost advantage

The increased flow of international companies setting up their offshore design and development base and the ramping up of engineering human resource in the existing set up, including the non-captive companies is expected to drive the revenues for the total design market in India.

Second part follows in the next blog.

Sunday, May 6, 2007

Trends in fixed-mobile convergence

Kudos to my colleague, Idhries Ahmad, for having put together a wonderful piece on fixed-mobile convergence (FMC). I would just like to highlight the trends on FMC in India. FMC is definitely at a very early stage in the Asia-Pacific region itself, and probably nascent in India.

Akshay Agarwal, Ecosystem Partner Manager - India & Korea, Texas Instruments, India forecasts that service providers in India may start with pre-IMS, skip UMA and may jump to IMS directly. Besides, the successful adoption of FMC would also depend on the deployment of 3G and Wi-Fi. Pre-IMS services could mainly be applications such as instant messaging and photo sharing.

The contribution of value-added services (VAS) has been steadily increasing and is likely to form a major portion of FMC revenues in the near future. Even as the FMC services market gains traction, the go-to-market strategy of service providers should be to provide a variety of FMC-compliant dual-mode handsets to customers.

At this point, the Indian telcos are still focused on meeting the high demand for basic voice services, and are gradually opening up to generating demand for converged service offerings.

Rising income, declining handset prices, and an increased awareness of viable mobile services among masses has led to the growing adoption of mobility solutions. The major areas of growth will be the youth segment and the rural areas.

FMC could be the solution to provide rich media services to the end customers. It may facilitate both fixed and mobile operators a magical solution to provide high-end services to subscribers, further entail the convergence of access networks, and help the operators to integrate their forces to tap the huge potential of rural India.