Friday, February 29, 2008

HD-DVD is Dead. How about Dueling EDA Standards?

As anyone following consumer electronics knows, the nascent high-definitition DVD market was held back by dueling standards: HD-DVD, backed by Toshiba/Microsoft, and BlueRay, backed by Sony and many movie studies. Microsoft XBox 360 had an HD-DVD option, and Sony PlayStation 3 had built-in BlueRay capability.

This unfortunate situation harkens back to the old "Beta vs. VHS" VCR format fiasco. Sony lost that battle, but in the case of BlueRay, they've come out on top. Nearly all studios and retailers have announced support for BlueRay instead of HD-DVD, and it looks like the industry will finally be able to focus on who can produce the best products at the best prices.

What does this have to do with EDA? Unfortunately, our industry is also plagued by dueling standards, and by companies jealously protecting de facto standards. In the beginning, there was Verilog, which Cadence refused to open up, spawning many man-years of duplicated work in VHDL. It happened again with Synopsys' .lib format. When Synopsys refused to open that up that standard, the industry spent a bunch of time on a competing OLA/ALF standard. Finally, Synopsys opened up "Liberty" and we can focus on who has the best tools, with library access for all.

In the present day, we still haven't learned! I hope for a quick and orderly consolidation of

  • Advanced timing modeling: CCS (Synopsys) vs. ECSM (Cadence)
  • Low Power Constraints: UPF (Synopsys) vs. CPF (Cadence)
  • Though I'm not a Verification guy, it sounds like the same thing is happening with VMM (Synopsys) vs. OVM (Cadence/Mentor)

Come on, big vendors! For the sake of the your customers and to have a dynamic, innovative industry, tear down those walls!


KarenB said...

Hello, John -

From one the "big vendors", we want one standard, too, and are willing to contribute in a formal setting to make it happen:

The Standards Game

I hope my counterparts will be willing to do the same. If not, I hope Accellera will try anyway for the sake of EDA customers.

Anonymous said...

Just to be clear - UPF was developed and supported by Magama, Mentor and Synopsys.

John said...

Hi John:

Something similar exists in the DFT/Test world - proprietary test compression schemes. And what makes this worse is that once you've implemented one vendors compression IP in your chip, you're stuck using their back-end tools (data collection, yield/failure analysis) to interpret results.

There was a standard drafted through Accellera (Aug 2006), and although all the big EDA firms were represented, I've yet to hear of anyone implementing to it...