The previous installment of our lunch with Gary Smith focused on Globalization, Pace and Product Evaluations. Gary then made a comment that almost seemed like a “throw-away”, but it is a clear indication of where the next era of growth is going to come from - semiconductor vendors will move to solely develop at the Electronic System Level (ESL) and outsource everything from RTL and down.
(This reminded me of a conversation I had with Grant Pierce of Sonics years ago. We were musing about whether at some point in the future fabless semiconductor vendors wouldn’t go the way of HP - providing brand, distribution, sales and support for products that are developed overseas by different companies.)
We can see the motivation - This approach to development would enable shorter development times through greater parallelization. It would also enable greater leverage of competitive advantage and the subsequent reduction of costs.
To a certain extent, this is already happening at the sub-system level. Large semi’s are sub-contracting out blocks and sub-system design and verification already. Is the spec being handed off as executable ESL models or as paper documents though? Either way, it makes for a natural progression. (As Gary noted in 2007, the best example would be the OMAP platform.)
The subsequent discussion moved on to how the electronic design industry really did itself a disservice with it’s (mis-)use of the term “ESL” coined by Gary back in the ’90’s. (The wikipedia definition link provided appears to be another level of obfuscation and could do with a clean-up.)
Intended as a flag round which to rally and understand concurrent software/hardware design, it instead became a market growth trend that every marketing guy in EDA wanted to hang his business projections on.
This resulted in a lose-lose situation. Not only was the real opportunity of ESL delayed by (ten?) years, but there was no way the EDA firms were going to meet their forecasts built upon those ESL projections.
To compound the problem, as we go into the era when ESL is to become a broadly accepted hard requirement and present opportunity, the term is misunderstood and loaded with negative connotations. The industry will need some re-education (or a new term?) to get back to where all those involved are speaking the same language and trust has been re-established.
If this occurs as Gary predicts, we’ll have even greater need to explore solutions to the problems raised in my last post. This also fits nicely into our next topic and the next post: Opportunities for the Next Wave in Electronic Design.
