Having just attended Synopsys User Group (SNUG) in San Jose, there was a nice little surprise for those of us interested in cloud computing related to EDA. In his keynote opening on Monday morning, Synopsys CEO Aart de Geus announced that Synopsys has a cloud offering available for customers to use – TODAY.
While most of us were aware that Synopsys has been looking at cloud, I am not sure that anyone was expecting his statement “We’re open for business”.
The specifics of the offering are still sparse, but the announcement by de Geus and subsequent presentations by David Hsu (see his blog on this here) indicate that the offering is for burst capacity related to VCS workloads.
While this is targeted at a very specific offering to begin with, it seems like a very appropriate target. VCS jobs include verification regressions, and can make up 30%-40% of the overall workload executed on an EDA compute cluster. The data sets involved (input and output) are finite and reasonable, so this would be a prime candidate for cloud use.
This has the potential to significantly offload the internal cluster resources, giving companies time to figure out how cloud works and what their strategy is for leveraging cloud beyond this. In addition, it gives them some breathing room on their ever expanding datacenter crunch, which is causing a great deal of pain today.
In Hsu’s presentation, he detailed out the process Synopsys went through for testing and evaluating cloud and the work they had did to make the extension into the cloud (using Amazon presently) as seamless to the engineers as possible. Also during David’s presentation, Synopsys invited Qualcomm’s Mike Broxterman to discuss his evaluation of the capability and the results seen by the Qualcomm team using the solution how a customer would use it. Both were well presented, and seemed to be well received.
The solution looks to be well researched by both Synopsys and the customers who participated in the POC phase, and this is probably the first of many steps to move EDA into the Cloud.
Blog originally posted at HPC in the Cloud.