What’s so Special about Your SoC Design Data?

Last week on The Breker Trekker, we discussed the resurgence of interest in EDA tools in the cloud. Like our first post on the topic two year’s ago, last week’s entry was very popular. Clearly this is a topic of interest to both our regular and occasional readers. Two more announcements regarding EDA in the cloud also surfaced during the recent Design Automation Conference (DAC), so it does seem as if there is more effort going toward finding a technically and financially successful industry solution.

Last week we summarized five barriers that have helped prevent cloud-based EDA from achieving mainstream adoption:

• The EDA vendor’s effort to port to a cloud-based platform
• Worries about GUI and interactive responsiveness
• Ability to support users of cloud-based tools
• Lack of an established, proven business model
• Concerns over security of the design and verification data in the cloud

This week we’ll focus on the last of these barriers. When we ask customers and prospects what they think about EDA in the cloud, security is almost always their first concern. Consider what would be uploaded in order to complete simulations for an SoC. One could imagine uploading a completely encrypted model and a simulator executable that has every custom and licensed add-on already included. This could be quite a big set of files, but it would address at least some of the security concerns.

But the whole goal of using the cloud is to avoid having big servers and lots of tools in-house. If the users have to compile the hardware design, compile the software, build executables, and consume lots of power just to get the SoC ready for the cloud, maybe they should just run everything locally. So usually if teams want to run simulations in the cloud, they’re thinking about doing all the compiling and building there as well.

In that case, the SoC designers are uploading their RTL code, the embedded programmers are  uploading their C/C++ source, and the verification team is uploading models that capture a lot about what the chip is supposed to do. That’s a lot of proprietary data that presumably could be stolen by a competitor to clone the entire product, or used as a starting point for creating an even better solution. This can and should be a cause for concern.

What’s interesting to us is the special place that this SoC data holds. It’s very likely that most if not all of the engineers on the project place a great deal of personal trust in the cloud. Many people shop, pay bills, manage investments, maintain their resumes, access medical records, connect with friends and family, and all but live online. The amount of precious personal data ripe for exploitation is staggering. Yet most of us implicitly trust the cloud with much of our lives.

At the corporate level, things are not so different. Electronics companies must pay their taxes and file a wide variety of reports online. Cloud-based accounting packages are common; enterprise employee payroll and benefits systems are almost universal. Many companies place in the cloud every bit of information about their employees and finances. This would be of immense value to competitors (and others); is it really any less critical or sensitive than SoC design data?

With the massive popularity of customer resource management (CRM) tools, many companies manage their entire sales pipeline online. Every current customer, every lead, every contact, every quote, every meeting summary, installed base records, and more are often in the cloud. Imagine how much damage a competitor could do if CRM information is not secure. But, again, it is really any less valuable than RTL or simulation models?

So you see why the status quo is interesting to us. People say that they’re nervous about EDA in the cloud at the same time that the engineers on the project and the companies themselves have virtually every other precious corporate data asset in the cloud already. It doesn’t seem to be a lack of trust just in clouds supplied by EDA vendors, since there are almost no EDA tools running on the same clouds that already contain so much other data.

Are we missing something obvious? Have you tried any cloud-based EDA tools? If not, why not? Is security your biggest concern? How could we help address this? We strongly encourage you to comment below and would love to get a dialog going on this topic. Thanks in advance.

Tom A.

The truth is out there … sometimes it’s in a blog.

P. S. Don’t forget: this is your last chance to a $50 Amazon.com gift card just by filling out a survey of which proposed TrekApps are of most interest to you. The deadline is midnight Pacific time tonight, June 30. Thanks!

Tags: Accellera, Breker, cache coherency, cloud, cloud computing, EDA, functional verification, graph, graph-based, IP, low power, portable stimulus, scenario model, simulation, SoC verification, TrekApp, use cases, uvm, VIP

This entry was posted on Tuesday, June 30th, 2015 at 4:44 pm. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.