Keith Richards of the Rolling Stones has been touring the U.S. over the last month promoting his book, “Life,” so it’s somehow fitting to riff on the lyrics of one of his hit songs, “Time Waits for No One” that he co-wrote with Mick Jagger. In the world of electronics, technology waits for no one, especially IC developers or electronic design automation (EDA) suppliers.
Today, most mainstream designs are implemented in 65-nm or larger process technologies, while early adopters are working at the 45/40-nm process node. But cutting-edge designs are already being implemented in 32/28-nm –– and sooner than you think, 32/28-nm will be the mainstream node.
Moving to 32/28-nm process nodes poses new technical challenges as well as hardware and resource management. Technically, managing power, crosstalk effects and process variability and the dramatic increase in the number of operating modes and corners becomes much more difficult.
The increasing size and complexity of designs at this node can stretch engineering resources to the breaking point. Designers must be enabled to build larger, more complex designs without increasing the size of the team. Gone, too, are the days when designers could solve problems by simply throwing more computer power at the problem. Multi-threading is good –– but not good enough. And, last but not least, they are required to meet ever-more-aggressive development schedules. As Keith Richards and Mick Jagger wrote in 1973, hours are like diamonds, don’t let them waste.
Knowing that technology waits for no one and nothing, EDA companies must work to meet the challenges of the 32/28-nm node. Automation is needed to provide higher capacity and deliver faster throughput.
One way to do this is to leverage distributed processing. Intelligently partitioning the design, and distributing the placement and routing across a network of servers, and automatically re-synchronizing the design at major flow stages can offer significantly faster throughput and better results. It allows a developer to tackle a much larger design and achieve the same or better implementation throughput –– measured in cells per day –– than he or she would have previously realized on a much smaller block. This distributed approach returns a 2X to 3X speedup over what is achievable with the best multithreaded flat flow.
Distributed processing has been employed to speed up IC analysis, but so far, it hasn’t been fully applied to IC implementation, which is a much tougher problem. New innovations in distributed processing for solving the implementation problem are on the horizon and will be here soon.
Timing is critical –– the adoption of the 32/28-nm process node is well underway with tapeouts already completed and new design projects already on the drawing board. Design teams can’t afford to be left behind the curve. New innovations in automation are needed to tackle the challenges posed by the rise of the 32/28-nm node. Keith Richards is right: Time waits for no one, no favors has he (or she).
With special thanks to Mick Jagger and Keith Richards. "Time Waits for No One" was released on the album It’s Only Rock n Roll in October 1974.
Smith’s experience as an executive within the EDA industry dates back to 1987. He holds a Master of Science degree in electrical engineering from Stanford University and Bachelor of Science degree in electrical engineering from the University of California at Davis.
so true technology doesnt wait for anyone. once upon a time walk man was a happening technology and now its going to close. People used to be happy getting VCRs and now its obsolete. With blu ray DVDs coming in, regular dvds are considered less advanced. Technology really changes so fast.
Distributed processing is being employed in IC design, analysis and fabrication.Many times i see on the news that a paticular group of engineers were working on a particular IC to bring to the market successfully. As time is available to every one and we need to manage with the time.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.