Gabe's EDA Update
In June 2012 Gabe Moretti will celebrate 44 years in EDA. Gabe has contributed to the industry first as a developer, then as a senior manager and now as an editor and industry observer. He is a Senior member of the IEEE and the recipient of the IEEE RonWaxman Meritorious Award. Gabe has worked … More »
Software Costs Up: This Is Not News
June 13th, 2012 by Gabe Moretti
Steve Leibson posted an article in the EDA360 Insider that reports on a panel at DAC chaired by Wally Rhines. The panel on ESL touched on the cost of IC development and Wally pointed out that the cost of software development is much higher than the one for hardware. In fact, software development costs are rising.
This should not come as a surprise to careful observers of the industry.
What is happening is that IC manufacturing costs are increasing significantly but hardware development costs have not been rising very much. Wally stated that the increase is around 10%. Re-use is the principal reason for the stability of hardware development cost. The use of standard cores, like those from ARM is so widespread that when combined with standard busses and functional blocks, most IC’s look more like standard computers than ASIC.
This reminds me more of the IBM 360 computer than of a special proprietary system. Thus in order to simplify the hardware development the weight of developing a unique product has been shifted to software content. After all software is stored in memory, a subsystem that is relatively easy to produce at leading edge processes, while new complex hardware has a high probability to reduce yields, at least for the first few months when it is important to have the largest possible unit count available to establish a credible presence in the market.
The question that comes to mind is: should we let foundry based companies develop and market standard operating platforms and just develop applications that differentiate products from one another? May be we should take advantage of 2.5D technology and build a die containing memory and a “simpler” hardware module that is connected through a standard interposer to a commercial computer. Yield issues would be limited to the application specific die thus decreasing costs. And hardware development would be even less expensive, since most of the complexity would have been handled by the standard computing system which will be used in very large volume.
All this is and will even more significantly impact the EDA business model. We must prepare for it and not “just” be a victim of it.