25 May 2010
A new vision for eda
No matter what industry sector you look at, the participating companies always want to grow the available market – it's a natural function of the business world. One example is the programmable logic market, which has always wanted to capture business from other sectors – such as dsps and asics – as its devices become more capable.
Now, Cadence's recently appointed chief marketing officer John Bruggeman (pictured) has released a discussion paper called EDA360 – The Way Forward for Electronic Design (www.eda360.com), which outlines ways in which he thinks the eda sector can be developed from what he calls a 'business as usual' $5billion a year industry into a 'crazy growth' sector generating $25bn a year.
"How?," Bruggeman asks. "By expanding into some new market segments and tapping emerging trends. We know this shift can't just happen by Cadence alone and the participation of the eda ecosystem – including our traditional competitors as well as the eda user base – is critical to reaching the end game the vision talks about."
Bruggeman says there has been a 'profound shift' in consumer buying habits which will change the industry 'as we know it'.
"This shift has been underway for years," he claimed, "but has been accelerated by the iPhone. Now, it's all about apps and if you can create an apps community, you can sustain differentiated products."
"In order to support this," he said, "manufacturers will have to move from providing entire solutions to becoming application enablers." And this is having a chain effect. "System companies are asking semiconductor companies to provide application ready platforms which not only include hardware, but also software. Semiconductor companies are challenging themselves on how to deliver this and are asking eda companies to expand the set of enabling capabilities they provide." It is this expanded environment which Bruggeman calls EDA 360.
Looking more closely at the consumer world, EDA 360 points out that a 22nm SoC is likely to cost $100million to develop and recouping that investment might need 80m units to be sold. Concluding this is unrealistic, the vision paper suggests the appearance of what it calls integrators – companies working at the silicon, SoC and system levels – who will supply application focused platforms to product developers.
"The eda industry, to date, has only served the needs of creators," says the document. "It has almost completely ignored integrators, who need a different set of tools and capabilities."
According to EDA 360, three steps are said to be needed to create this new environment:
• Create. Whenever or wherever hardware or software IP is developed … it must be created with integration in mind. It must come with the right documentation, be fully verified and be configurable for the end application.
• Integrate. Design teams must be able to integrate IP into platforms, then verify the complete SoC or platform, including interfaces between analogue and digital blocks, as well as hardware and software.
• Optimise. Design teams must be able to reduce die and package costs while ensuring quality, lowering power consumption and trimming test costs through automation.
The document also believes another stage is needed – the creation of a 'collaborative ecosystem' of semiconductor design companies, IP providers, service providers, foundries, eda vendors and assembly/packaging houses.
"Without such an ecosystem," says the document, "the electronics industry will collapse back into a handful of large, vertically integrated companies."
While making these suggestions, the document also leaves plenty of room for discussing the 'hows' and 'whens': where, just to take one example, does fpga design fit in this? Even though they may not appear to be so, the likes of Altera and Xilinx are big eda companies in their own right, having developed over a long time the software that enables their devices to be targeted at a given application; often through the use of third party IP.
There is also the question of how the embedded software world might fit into EDA360's new future; where, for example, do the programming tools for mcus fit? ARM has already created in mbed an environment in which mcu IP can be shared; seemingly an issue of interest to EDA 360.
Growing revenues organically by 500% will be more than a tall order for an industry that has traditionally made the vast majority of its sales to a handful of leading edge companies – the so called creators. Developing software for an entirely new set of users – the integrators – isn't something that will happen overnight.
EDA players could, of course, redefine their boundaries and then acquire companies which have the necessary technology to fill in the blanks. But that will not grow the industry per se, just the revenues of eda companies.