Brian, as a regular reader I think that I understand why you comment 'SW engineers can be put on that task' in terms of writing sw for increasingly complex/larger chips. However, Gary Smith's observation is well informed and deeper than it might at first appear. Throwing bodies at the problem won't work both practically and economically. We are going to need tools, abstractions and programming models that give small teams leverage.

The emergence of ESL already provides a promise of things to come. It has taken time longer than we all thought for the industry to accept that ESL is inevitable. Similarly, it will take time before ESL goes up another gear and acts as a bridge to the software world with virtual platforms enabling fully integrated instrumentation, compilation, analysis and mapping/partitioning of applications.

The trouble for everyone in the industry is that it is already more than enough of a challenge just to keep up with the requirements of the next manufacturing node (scalability, new problems, verification) let alone trying to change engineering paradigms. Broadening the scope of their engineering envelope within organisations is a tough one for the EDA companies. The RoI is there but someone needs to grasp the nettle.

The EDA guys know precisely how to engage with and sell to hardware people. They have proven processes and organisations that can do this. Adding software into the mix is as much a cultural challenge as it is a technical one. Software economics, programmer productivity and what we term 'certified legacy code' all have to be taken into account. I think that this is what Gary was referring to.

As you are at DAC drop by the Compaan/ACE booth and have a talk to the guys as they have been in the trenches quietly doing some of this stuff.