Gadi Singer, vice president of the Intel Architecture Group and General Manager, SOC Enabling Group for Intel, provided an excellent keynote on Wednesday at DAC (June 8th) that didn’t so much set out his company’s shopping list as its strategy for the context of tool and methodology development.
We’ll take a closer look at many of the points Singer raised soon in the technical section but there were some headline takeaways of note. And again, the need for concurrent hardware-software development (within informed iterative cycles) was a consistent theme.
First, expanding on some of the themes from Tuesday’s ESL Symposium, Singer said that while ESL is seen as a progression in design abstraction, it does not encompass all that needs to make up system-design.
Rather, ESL reflects hardware first and foremost and a small proportion of the firmware component in one of today’s designs. Above that are a series of issues such as—rising from the silicon—drivers, operating systems, middleware and applications.
If one takes ‘silicon modeling and development’ as being what constitutes ESL today, full system flows need to also incorporate and integrate ‘software development and debug’ (OSs and drivers) and ‘system optimizers’ (applications and middleware) in broad terms.
Singer’s ‘winning solutions’ for this process, will be based around five principles.
The speech also set out that this is an immediate requirement. “This is not something we’ll be doing in five or 10 years time; it’s how we’re doing design now,” said Singer, who also noted that Intel has created its own software to run on top of EDA tools to start creating this integrated approach.
Singer concluded that not only does ESL need extending but that the design tools industry might also consider renaming itself, from EDA to ESDA for—as you’ve probably guessed—Electronic Systems Design Automation.