ITRS 2.0 Tackles System Level Integration
The International Technology Roadmap for Semiconductor (ITRS) has changed. It is no longer primarily focused on manufacturing challenges that need to be overcome to enjoy the benefits of continued scaling. Instead, ITRS 2.0 as it’s called, is starting with a system level approach and working down to the chip and transistor level. “ITRS 2.0 is alive and well,” says Paolo Gargini, longtime ITRS chairman.
This new approach is described in detail in this month’s story on pg. ?? “ITRS 2.0: Top-Down System Integration.” System Integration prescribes ways of assembling heterogeneous building blocks into coherent systems.
“The world has changed so much that if we didn’t look at system requirements and then project that down onto devices, we were going to be out of sync,” Gargini said. “We spent last year essentially doing this.”
Interestingly, Gargini said the semiconductor industry has evolved to the point today where it is similar to what he experienced in the 1970s when he was working at Fairchild.. “The design was 100% in the hands of the system companies. It was very difficult to handle the business because typically they would not reveal what was right or wrong.” One such project was how Polaroid handled the development of a new chip for its SX70 camera. Fairchild and TI got the assignment
By the end of the ‘70s, Gargini moved to Intel. “We got lucky in that IBM selected Intel and Microsoft. They thought it was going to be a very small business. They just wanted to make sure they could cover everything from the large super computer to the miniscule PC,” Gargini said.
By the end of the 80s and throughout the 90s that technology had become so efficient that there was a switch from a three year cycle to a “real” two year cycle. “In reality, for processors, it was a three or four year schedule. The real change was to make this two year cycle real,” Gargini said. “In essence, Intel and Microsoft dominated the business and what was left for the PC company was to assemble the memory, screen and other elements. The business really got turned around.”
Another big change came around 2007, with the first iphones and mobile devices. “The main difference was that instead of buying a processor, a dynamic RAM, a static RAM, a video chip, the interface and so forth, the system designer could take just a little bit of everything and put it on a single die or maybe two dies at the most that would completely satisfy the system that they were building,” Gargini said. “Like in the 70s, the system designer had absolute control over what the die was supposed to do.”
By the second introduction in 2010 of the tablet, this really new business model began taking traction. “It was clear that the world had changed. By 2012, I outlined to the ITRS groups that we were going to change,” Gargini said.
As a first step, the ITRS groups looked at all of today’s applications and dissected them into their basic building blocks. “First of all, we needed system integration that would look at all these different applications and identify what the basic building blocks were,” Gargini said. “We looked at iphone and tablets and the reverse engineering analysis and the different chips. Out of this we said ‘okay’ we understand the overall system integration. The main element is that you have system-on-chip and in addition you can stack different chips. You’re really in the world of heterogenous integration.”