Silicon at the Core of Innovation - June 2007

On March 27, 2007, Sun announced the formation of a new microelectronics division, signaling an expanded focus on making Sun's silicon innovations available to a broader set of customers. Dr. David Yen, executive vice president of the new division, shares with Sun Executive Boardroom readers his view on why the timing is right for this move and what new possibilities it signals for customers.

Q: Why did Sun form a new microelectronics group?

A: This move is in alignment with Jonathan Schwartz's strategy of pushing Sun's intellectual property to a broader customer base. Our Solaris Operating System already enjoys much success beyond our own x86 server platform because of this strategy. We're doing the same thing now on the silicon side by pushing our processor, networking interfaces, and other silicon technologies to a customer base outside of Sun.

Q: What will the new group do and how does it address the needs of customers?

A: This team has been around for 16-18 years supporting the development of our SPARC processors. We have high-end microprocessor design expertise, which is the most demanding, as well as chip testing and tool development expertise. It's an engineering development powerhouse with well over 1000 engineers — a very significant competency in the industry as a fabless electronics house.

We have also made significant progress in the throughput computing market and are now leading the industry. Both our upcoming Niagara 2 processor and Neptune network throughput technology are unique and we've seen a lot of outside interest in embedding or licensing these technologies into various customers' product lines.

Customers will be able to leverage the breakthrough silicon multithreading from Sun to build their own highly threaded, networked products for the new Web 2.0 buildout.

Q: What is the business environment for microelectronics these days?

A: Very good. We're seeing two things happening. First, the number of network-attached devices — other than servers — is exploding. There are cell phones and "Blackberrys," but also things like portable game and music players, satellite radio, GPS, etc. At the same time, the companies making those devices are rethinking their approaches. They used to build on custom processors with custom OSs, but custom development takes time — and time to market is critical to their success.

So now, more of these companies are looking at general-purpose processors and general-purpose OSs, and that's perfect for us. We have chip multithreading, or CMT, processor technology, and the Solaris OS they need, plus the whole Solaris ISV ecosystem. That means they can focus on their value-add, rather than on custom processor and OS development — which is expensive and time-consuming. And of course, the kinds of services that appear on these devices are all inherently threaded, so that creates demand for our CMT-based servers and multithreaded network technology. So, we believe that Microelectronics will not only create new business for Sun, it will also create a pull for our existing businesses.

There is also demand for smaller, cooler technologies to alleviate datacenter constraints and address environmental concerns such as power and cooling. This is driving an effort to design CPUs that deliver higher throughput while minimizing power consumption and space. The ability to get more out of a piece of silicon real estate is limited by Moore's Law and the answer is chip multithreading.

Lastly, the fabless industry is growing rapidly. This trend is allowing companies to contribute their intellectual property for new designs and products. By setting up an independent microelectronics business unit, we can spread such innovation to a much broader market beyond Sun's own customers. In early April, we licensed our throughput networking intellectual property to Marvell, which gives them the ability to further develop, manufacture, and distribute all of its derivatives. Not only will Marvell's customers benefit from the improved features and costs, but Sun can become a customer for the chips Marvell develops as well.

Q: In what way do silicon investments provide competitive advantage?

A: Having our own silicon and hardware capability allows Sun to offer a more complete integrated solution — to maximize the benefits of all layers of technology, from silicon to system architecture to platform packaging to operating system to the software stack. This maximizes Sun's ability to optimize for such integration. For customers who just want an integrated solution, this is the best value.

However, in the spirit of openness, since we are contributing our IP and technologies to the marketplace, there is a much broader set of customers who can now utilize these technologies to create products that bring them competitive advantage. No other company offers a more complete and optimized multithreaded environment — from Java technology to the Solaris OS to our UltraSPARC CMT to our latest networking throughput silicon technology. Our silicon investments complete the multithreading throughput equation for our customers.

Q: How will this be different from the microelectronics business that Sun had in the mid-'90s?

A: The only thing in common between those two groups is the English word "microelectronics." Today, Sun is a much tighter, fully leveraged organization. To maximize the flexibility and availability of our technology to the outside world we have separate businesses, but we are all SUNW shareholders. Our whole goal now is to make our innovations available to any customer who wants them.

Q: How does this relate to the Sun/Intel announcement back in January?

A: With Intel endorsing Sun as the mission-critical OS for Xeon, it allows us to propagate our Solaris technology because it is the industry's best interest. With our strong interest in the x86 world, Intel took notice. And, in the spirit of our new tact with microelectronics, if Intel is interested in working with us on microprocessors, we're all ears.

Q: How does this tie to the OpenSPARC initiative?

A: The OpenSPARC initiative is a very important effort. With the open sourcing of Niagara and UltraSPARC T1, there have been various efforts outside of Sun to extract part of the eight-core, 32-thread Niagara. There are two companies who have extracted one of the eight cores out and offered it to the general marketplace — Simply RISC and Polaris Micro. Sun also released a single-core, single-thread design in March 2007.

The significance of this effort is that this implementation can comfortably fit into the commercial FPGAs (field-programmable logic) offered by a few companies. This allows people, particularly college students, to do rapid prototyping. Since it fits into the capacity of the commercially available FPGAs, there's additional capacity for students to do experiments, which provides a nice vehicle for professors to offer lecture or lab courses.

We intend to work actively in that regard to expose our innovative throughput computing technology to the next generation to broaden the understanding and appreciation of the SPARC architecture.

About David Yen
David Yen brings over 27 years of technology know-how, engineering vision, and strong business management expertise to Sun. As executive vice president of the Microelectronics group at Sun, Yen will oversee the developments in network, cryptography, and high performance computing. Prior to this role, Yen led the company's Storage Group. Other key roles at Sun have included vice president and general manager for Sun's enterprise server business, executive vice president of processor and network products, and executive vice president of scalable systems.

Yen was co-founder and director of hardware development of Cydrome, Inc., a mini-supercomputer startup, before joining Sun in October of 1988. Prior to Cydrome, Yen was employed by IBM Research for manufacturing automation and TRW, Inc. for advanced processor development. Yen earned a bachelor's degree in Electrical Engineering from National Taiwan University, received an M.S. and Ph.D. in Electrical and Computer Engineering from the University of Illinois at Urbana-Champaign, and completed a General Management Program at Stanford Business School.