The recipe of smaller, faster and cheaper is long gone. But what will take the place of this decades-old market trend? Bernard S. Meyerson, IBM fellow and vice president of innovation and global university relations at IBM, shared his predictions at yesterday’s keynote address titled “From Gigahertz to Systems to Solutions: Our Industry in Transition.”
Bernard S. Meyerson, IBM fellow and vice president of innovation and global university relations, IBM“You can’t scale down silicon forever,” Meyerson said to an audience that spilled out to the hallway, acknowledging that Moore’s Law was right but silicon manufacturing has its limits. “Scaling died five years ago, but new technologies like carbon nanotube transistors aren’t ready for prime time either. Plus, silicon can’t scale to these dimensions. There’s a difference between building a few high-performing chips and building 5 billion. This will be a recurring theme.”
The industry has transitioned to innovative drive performance with the introduction of multi cores, he said, but innovation will need to take additional forms. Meyerson used the example of 3D stacking as a creative way to increase performance, especially since the “speed of light is dreadfully and woefully slow.” By bringing memory and processing in extremely close proximity to each other, he said, you’re able to improve upon the limitations caused by the speed of light.
“Radical collaboration is another type of innovation,” he said, referring to companies that are combining intellectual and financial capital. Meyerson noted how unit process has gone out of favor as a testing method. Now, he said, companies band together to create facilities that cover everything from R&D to manufacturing, like the IBM Alliance near Albany, N.Y., where IBM plays a role in the advanced research consortium but shares efforts with other companies. He referred to the region as Silicon Valley East.
“People who think at the system level will dominate this industry going forward,” Meyerson said. He told attendees they need to consider what’s happening to IT as a whole. Whereas there were about 1 billion RFID tags in circulation in 2005, there are over 30 billion now. Meyerson predicted that devices producing data will continue to grow, and how these units connect and share information will increasingly affect the industry.
Right now, automated toll payment on U.S. thoroughfares provides data that documents where, when and by whom payment was made. This information, however, could also be used to predict traffic patterns. Or, on a grander scale, the data could optimize regional transportation systems. Systemic thinking like this can make a dramatic change in the quality of life, he said.
Stream computing is also becoming increasingly practiced, and Meyerson used the example of what affects citrus fruit prices. In the past, analysts could retroactively sift through weather reports, SEC filings and news releases to come up with reasons why orange prices spiked or plummeted. Now, however, that information is instantaneously available to aid in decision-making. Stream computing has other applications, like healthcare, where disease or illness can be detected far earlier just by noticing a small decline in blood pressure or a slight increase in body temperature, the beginnings of what could lead to something far more serious.
As for cloud computing, Meyerson wonders if there will be a trend where small companies use the technology, as those systems consistently run above 90% utilization, opposed to often underused local desktop and server systems.
At the beginning of his address, Meyerson introduced problems related to water shortages, transportation congestion, food shortages and overages, and healthcare issues. The common thread to these problems was that many result from either too little information on the subject or so much that you don’t know how to use it. He called for those who create and deploy technology to step up to solve these worldwide problems.
“Our industry has the ability to render solutions that are both economically and technically feasible,” Meyerson said.
— Arthur Patterson, SEMICON West Daily News