ESL: Systems vs. Semiconductor Game 1

So, I’m watching hockey playoffs and I’m astounded by the urgency of play.  The winning team always shows more will to win.  That got me thinking about ESL and the two camps I see trying to define ESL - Systems companies and Semiconductor companies.  Whoever urgently drives the ESL definition is going to wind-up defining this sector and driving the infrastructure.

I see Systems houses showing great eagerness to adopt system-level tools out of the necessity to control requirements and deal with complexity. They have lower expectations of automation and validation as they have relied on manual processes. They seem to understand that there are many definitions of a “system model” and frequently have multiple views optimized for different uses; whether thats for rapid prototyping or algorithm development. I see great willingness to invest in any tool that can help trace requirements, add verification/validation or help optimize the architecture. There is also solid understanding that system-models will require refinement as development progresses to be used in either HW or SW flows.

Preparing Recommendations

On the other side are the Semiconductor companies who tend towards incremental designs and maintaining their strong RTL portfolios. Though most ESL tools today would be easier to adopt by the semi’s, there appears to be less investment in ESL as a whole - certainly not considered a must-have in an era of tight budgets and slim headcount. There seems to be conflict on whether ESL should abstractly model the functionality of both HW and SW or contain detailed timing to model bus transactions (it can, with refinement) - and until it does both at the same time, we’re willing to wait.

The benefits of ESL (higher-performance, lower-power, and lower-cost systems) are clear. Those who invest today have the opportunity to drive the development of ESL tools and methods, but who will take the lead?

More Blog Posts

Preparing Recommendations