Saturday, January 27, 2007

Chips Grab Mainstream Headlines

I was surprised and excited to see a chip-process development featured on the front page, above the fold, in the San Jose Mercury News. This is "the newspaper of silicon valley", as they say. The headline gushed Chip advance biggest in 40 years.

It's about successes announced by Intel and IBM to fabricate metal-gate transistors with "high-K" dielectric. This is good news, which hopefully will enable us to continue to follow Moore's Law.

But it doesn't hit me that this is the biggest development in 40 years. Is this triumph over materials significantly different from figuring out how to to use Copper interconnect, or "low-K" interconnect isolation? Those were both hard manufacturing problems that the process wizards solved in the last decade. Maybe it was a slow news day. ;-)

Thursday, January 25, 2007

Is Texas Instruments Exiting the Semiconductor Technology Rat Race?

SeekingAlpha is a financial web site that I just discovered. It's unique in that it focuses on analysis rather than news. And I've started following semiconductor company analyses, such as Is Texas Instruments Exiting the Semiconductor Technology Rat Race?
Technologists live for working on leading edge technology. TI killing off 32nm development is a major change in how it views itself and whether it will be able to attract the best and the brightest scientists in the future -- not to mention how the ones who do this there today must feel.

Now that is significant! Another IDM "throws in the towel" on doing all their own process and fab development. Will we be catching our breath at 45nm for awhile?

Yahoo! EDA!

Does Yahoo! understand the EDA industry? I have the headlines from their EDA News News Archive - Yahoo! Finance in "My Yahoo!". But most of the time, even a very liberal definition of "EDA" doesn't explain why these articles are here. I can understand including foundries, IP vendors, and even programmable logic companies. But how to understand the inclusion of U.S. Department Of Commerce Makes Annual Adjustments to Seasonal Factors for ISM Manufacturing PMI and Diffusion Indexes and ISM Non-Manufacturing Diffusion Indexes or Roche Diagnostics and Protedyne Sign Global Agreement?

I thought these industry news groupings have a human editor, but it makes you wonder ...

Wednesday, January 24, 2007

Intel Builds 80-Core Chip

This Intel 80-Core Chip is pretty far off in Research-land, but it does pose intriguing concepts, such as
  • Better performance/power ratio
  • Finer granularity of compute power and power consumption. If you need more performance, it wakes up more cores, and when you're done, they go back to sleep.
  • "Core hopping". If one part of the chip gets hot, the work that those particular cores are doing is moved to other cores on another part of the chip.
  • On-chip communications network. Of course, all multi-core chips have some sort of network. I'd like to see how sophisticated the Intel one is.

Prof. Richard Newton's Influence on the EDA Industry

Lamentation & Loss is a enlightening eulogy for Professor Richard Newton. By interviewing several EDA industry legends, Peggy Aycinena succeeds in recognizing Prof. Newton's profound impact on EDA, and reveals interesting history about Cadence and Synopsys, among others.

Wednesday, January 17, 2007

Hewlett-Packard's FPGA research, and replacing transistors

You may have seen the story earlier this week about a "breakthrough" announced by HP researchers, which involved applying nano technology to semiconductors.

I was intrigued, and Googled around to find this analysis at the "ars technica" web site: Hewlett-Packard's FPGA research, and replacing transistors. This excellent article explains that the HP announcement is for a very specific (and limited) application: FPGA interconnect. And it points out that the more promising HP announcement actually came out in 2005, which pointed to use of nanotechnology for circuit switching, not just interconnection.

Gotta love this web site's tag line: Serving the PC enthusiast for over 6x10-2 centuries!

Tuesday, January 09, 2007

Best Companies to Work For ASIC & EDA?

In the news today is the much-talked-about list of 100 Best Companies to Work For 2007 in Fortune magazine.

Hmmm, the companies that we typically work at didn't fare too well. I only see a couple that would typically have ASIC or EDA engineers:

Cisco Systems
Texas Instruments

As far as high-tech, there are a lot more software companies listed, and many related to health care.

It would great if EE Times did an industry-specific version -- Best Places for an EE to Work For!

Thursday, January 04, 2007

DAAMIT vs. ChipZilla, Past, Present and Future

I've mentioned this tech enthusiast web site before. Geek News - Is AMD's vision enough? is a great summary of the multi-core strategies being pursued by AMD and Intel. Be sure to check out the User Comments, which are one of the best resources of this site. "GoatGuy" is the real deal, always good for expert commentary. And this lexington_steele guy is pretty provocative, too. I wonder where he got that name?

Wednesday, January 03, 2007

R.I.P. Dean Newton

The EDA industry and the University of California lost an influential leader with the passing of the College of Engineering Dean, Richard Newton: EDA icon Richard Newton dies.

I was fortunate to hear Dean Newton speak at a "Berkeley in Silicon Valley" lecture last April. My notes from that lecture note that "synthetic biology", the application of engineering to living organisms (on a nano scale), was one of Dr. Newton's current passions, and his family requested donations in his name to the Berkeley Center for Synthetic Biology.

Update: John Cooley's Memories of Professor Richard Newton.

Tuesday, January 02, 2007

Happy New Year in ASIC & EDA

I thought about writing a forward-looking article, making some predictions about our industry. But, so far, I am too lazy. ;-) Besides, that's what Gabe & Gary get paid the big bucks to do. So, here are links to their New Year columns with a bit of color from me.
  • Gabe's column, Last minute deals open opportunities for 2007 and beyond, is focused and sounds plausible. His DFM story sounds good -- it needs to get more into the design phase, as a proactive guide to help people laying out the circuits. And I like his call for "Locality Sensitive Cells" that would be DFM-correct by construction. His other point, about Mentor's moves in the ESL and C-synthesis space, discusses alternatives to SystemC, and where C may be lacking as the perfect system-level language.
  • Gary's "column" isn't really written by Gary. He has Richard Goering as his ghost writer! I have more problems with Gary's predictions that Gabe's. Gary is still predicting the ascendancy of ESL, FPGA, and embedded software. I still see plenty of challenge and opportunity in plain old RTL-based standard cell design! Gary does have 10 predictions, and it's worth looking over the rest of them for thought-provoking variety.