A Collateral Shade of Green

Cloud-based computing is hardly a new concept. It’s been around since the 1960s, when it was known as time-sharing. The repackaged version, along with commercial outsourcing, has been talked about in the EDA industry for the better part of a decade.

What’s different, though, is that the original concept was for software as a service, or SAAS, whereby full tools would be available online and data would reside offsite. The new version is much more focused on supplemental but temporary compute capacity, particularly in areas such as debugging, simulation and library characterization. In these cases, being able to rent resources by the hour, by the day or by the week is a far cry more efficient than having to own all of this equipment and leave it sitting idle most of the time.

Most corporations have re-discovered this trend over the past few years. In IT, the cost of buying commodity servers is relatively insignificant. But keeping them powered up and ready to process costs a lot of money for energy—millions of dollars a year for some companies. And for the most part, that energy is wasted. Servers are underutilized, and with racks upon racks of them in densely packed cabinets they have to be cooled with expensive air conditioning. That’s a double whammy on energy consumption.

Cloud-based computing’s big advantage is flexibility. As a capital asset it can be averaged between peak and trough demand, which is the most efficient way to use servers. As long as there are enough compute cycles available—and usually some are held in reserve but switched off until needed—then efficiency can be maximized. Many of the largest companies run their most critical operations internally, but they do use a mix of internal and external cloud-based approaches.

EDA is an unusual case because the parts that will need to be run on external clouds will involve mission-critical data. But what’s also different in the EDA world is that the software licenses are so expensive that this kind of arrangement makes sense. It’s not just the energy efficiency that customers are worried about. It’s the cost of energy plus the user license fees.

Still, the energy savings will become an important factor as companies begin to understand just how much they’re saving by offloading this cycle-intensive computing. Inside of many corporate IT departments there is a recognition that it’s good to be green, but it’s much better to be green if you can do it while saving money. Cloud computing is a big step in that direction, and it’s one that will ultimately make the design of more energy-efficient chips…well…more energy efficient.