The technology looks the same, but to get to the next level will require some cooperation.
Every now and then you get the feeling you’ve been here before, and with technology this is a persistent theme.
Virtualization looks remarkably similar to time sharing, which is what most engineers in their 40s and 50s used when they were in college. And 3D stacking, particularly the 2.5D version, looks eerily like the old MCM, aka multi-chip module.
There’s nothing wrong with resurrecting old concepts and putting a new spin on them. Shakespeare did it. So did Apple with the iPod and the iPad. And current system-level design approaches have been talked about for the better part of a decade.
What’s different is how all the pieces go together to make something successful. Timing is everything, and in technology it’s more than just end-market acceptance. It’s a tightly integrated ecosystem with interplay between different tools, different manufacturing processes, design flows and various levels of abstraction. In the future, it also will be about being able to bridge multiple process generations on stacked die into a single package or even a single chip.
All of this will require a combination of new technologies, such as TSVs and TSV interconnect models, tools to bridge several generations of analog and digital, a deeper understanding of how software will utilize the hardware, and a far better understanding of how IP needs to be characterized and verified. But on a more mundane level, it also will require integration of existing generations of tools with each other so that high-level synthesis and software prototyping can work effectively in existing hardware, software and manufacturing flows.
These are tall challenges, and no one company can do it alone. Done in unison, the market can expand at an accelerated rate—even drawing in new customers, which is something the EDA industry has been desperately searching for. But so far the efforts in this respect have been rather paltry, driven more by standards organizations than by tools vendors themselves.
Customers have been asking for more integration, and vendors have been looking for measurable ROI to make that happen. But the real ROI may be much bigger than the cost of the integration itself and raise top-line revenue for years to come. In an interconnected and interdependent world, what’s good for one company has to be good for multiple companies.
–Ed Sperling
Leave a Reply