Cloud computing offers benefits and challenges to EDA tool users, but either way it is here to stay.
EDA in the clouds may be an apt description for the way in which the industry is taking a good long look at cloud computing. While it may seem like a nebulous term, cloud computing means shared resources, software and information provided over a network and thereís nothing nebulous about that. Itís a trend thatís gaining momentum, offering both opportunity and challenge to our industry.
Anyone with graying hair like me remembers batch computing at computing centers, an early generation of cloud computing. Going back even further in EDA history, hardware designers rented computer space from Control Data Corporation for their simulation runs, another example of an early cloud computing application. Ten years ago, it was called Internet EDA or Internet CAD, but neither took off.
Maybe the technology has improved or maybe weíre reset our expectations or perhaps itís because well-defined services are available, but cloud computing is here and itís real. Whatís better, the barriers to entry are low for EDA companies who prefer not to make the investment in expensive server farms or computer equipment. Amazon, IBM, Google, HP and others now offer attractive shared-computing services.
Some classes of EDA application software lend themselves perfectly to the cloud and allow project teams accelerate the design loop significantly. Analog verification and signoff checking, for example, are compute resource intensive, yet do not require much interaction from a designer. In other words, the cloud has a sweet spot for software tools that work in batch mode a la old school computer centers. They can be sped up significantly using the fast on-demand compute resources in the cloud. EDA synthesis and physical synthesis, however, are technically less able to take advantage of large amounts of CPUs. That makes them initially a less suitable candidate for cloud computing.
A cloud computing strategy may offer a small, ambitious EDA company or stealth fabless semiconductor startup a way to get a toe hold into a niche market segment by reducing the barrier to entry. By opting for a cloud strategy, the company has outsourced the information technology (IT) department, saving money and resources, and enabling it to focus on its core competency.
The payoff for a larger EDA firm isnít quite as apparent, given that there are a variety of challenges to overcome. One sticky problem is that EDA doesnít know yet how to charge for a cloud computing license. It is accustomed to yearly or multi-year licenses and does not have experience with the unpredictable revenue from pay per use. That will change as we find a way to make it viable for the industry, which could mean a drastic change in EDAís tried-and-true business model.
Another downside is one of perception that is notoriously hard to overcome: Companiesí teams donít want to place their valuable intellectual property in unknown places. For many, however, security is less of an issue when cost effectiveness is so attractive. Those typically interested in a cloud computing strategy are smaller design houses who find the resources-on-demand business model attractive. These companies buy what they need and donít pay for resources they donít use because theyíre using the tool only one or two weeks a year.
This yearís Design Automation Conference (DAC) hosted a panel on cloud computing that was well attended, with discussions spilling out in the hallway afterward, a sure sign of an emerging trend. A few cloud computing demonstrations were found on the exhibit floor. Expect to see much more next year, in the form of papers, panels and possibly tutorials. Attendees will see cloud computing suppliers on the exhibit floor because itís clear that theyíre part of the design ecosystem, too.
EDA companies must recognize the range of requirements facing todayís project team and evolve existing technologies and innovate new technologies accordingly to help them assemble the necessary differentiated solutions within their time windows and cost requirements. A cloud computing strategy may be the first step. The goal of any cloud computing strategy should be to bring together superior technology, design expertise and a design teamís creativity to deliver profit-driving differentiated silicon.
While itís not a fit for all designer applications or EDA tool, cloud computing is here to stay, which is why I think that EDA is in the clouds.
Patrick Groeneveld is Chief Technologist at Magma Design Automation, and the Chair of the 2012 Design Automation Conference.