I am with an EDA company and I can not agree more with Dale's view on "Competing technology standard" that we have wasted too much resource and money competing on something which should not be an issue to begin with. The consequence? As Daniel points out the customer has to spend 4x as much just to make the flow work.
We as an EDA vendor should recognize the number 1 rule that the customer has the right to pick and chose the best-in-class tools for their design projects. This fundamental requirement goes above and beyond any slick unified DB, framework, etc. Anyone who doesn't understand this basic principle is doomed to fail. Don't get me wrong, the unified DB, framework, platform, whatever you call it, is all great but only if it allows the user to pick and choose best-in-class tools, which more often than not won't be from one vendor.
Knowing this fundamental user's requirement, it's no surprise that Mentor's Falcon and Cadence's OPUS failed miserably and the user still go back to the rudimentary standard file format such as SPICE, LEF/DEF, and GDSII, etc. albeit not as efficient as if they are all on the same DB, but at least these standard file formats allows the user to be able to choose and pick the best tool in class.
The solution? With the claimed openness of Milkyway from Synopsys and OpenAccess from Si2, we may finally see some relief. However, I would like to point out a few serious issues that could make OA ill-fated if these issues were not addressed and fixed soon. OA has a great mission that any OA compliant tools should be able to "plug-and-play". However, what could turn this good will into evil spirit is that some vendor is trying to hijack and cripple the OpenAccess system with yet another proprietary Skill language and constraint format, and what make Si2 double evil is that Si2 committees even allow it to happen.
So My plea here for the benefit of the whole design community is
1. Let us EDA vendors compete on the technology and not on the languages or spec formats.
2. The Si2 committee should ban any vendor who ride on the OA bandwagon and yet unwilling to be OPEN in the first place.
3. The Si2 should stop supporting any claimed proprietary stuff such as Skill language and Constraint management until they become OPEN to the public. Supporting a single vendor's proprietary stuff is a clear contradict and violation to the OA's mission statement.
OA is too good a system to be abused and spoiled by a single vendor.
Silicon Canvas, Inc.
Edited by hau_chen on 10/16/06 02:59 PM.