EDA is still evaluating the hosted application model for chip design. As readers have pointed out, the pay-per-use model and the hosted services model are quite well understood and deployed in corporate IT environments.
The question is how quickly these can be extended to EDA. Pay-per-use already exists in EDA in areas like emulation for example. However, deploying EDA applications using the hosted services model may be a little more challenging because of the nature of design engineering in general. Designs tend to be complex, design data files very large and users are somewhat reluctant to update tools mid-cycle unless the bug is a huge show-stopper or the potential ROI of the new feature more than covers the design changes required.
The implication then is that users need to ask their EDA vendors questions that address not only bandwidth and security issues but also storage and version updates.
While it is understood the applications/tools are hosted in the cloud, where will the design be stored? Will it be stored locally or in the cloud? How often will tools be updated? This is very important in EDA. If my design works with version X of the tool and compensates for known bugs in the tool, there is no certainty that it will work with an updated version that has fixed the bug. Will users be notified ahead of time when tools are updated or do they find out the hard way?
Design files are getting ever larger with each semiconductor process node. Does the hosting provider have enough bandwidth to support very large file sizes? Licensing costs are another important issue to consider. Hosted does not always mean cheaper. More on that topic another time.
Related articles:
Software as a service in EDA: Part 2
EDA must become a partner to customers
EDA is not well; where is it heading?
Watch out for Green EDA
This article appeared in EE Times India, a sister publication to EE Times Europe.
RSS
Digital
RSS| 
Digital| 
Mobile
