The principal research scientist at Hewlett Packard Labs is on a quest to find ways to use power and multicore processors more effectively across a broad range of devices. "My job is to think about systems architectures," said Partha.

Partha spent the last seven years at HP Labs after gaining his doctorate in electrical and computer engineering at Rice University. In recognition of his promise and accomplishments to date, MIT Technology Review magazine named him to its annual list released Wednesday (Aug 15) of technology innovators under 35.

Among his many projects to date, he helped define a way to light only the pixels a user needs to see on a handheld device using an OLED or other emissive display. The approach could extend battery life on the system a whopping ten fold.

Most of Partha's efforts have been in the area of data center computing closer to HP's mainstream businesses in servers and storage arrays. He has helped define two benchmarks that could become useful in measuring a system's energy efficiency.

"One of the big problems we have is everyone knows we all want to go green but no one knows when we can declare victory," said Partha.

Joulesort provides a guideline to designers trying to optimize the energy efficiency of a system in development. It is now going through academic review to determine its usefulness beyond HP's walls. "That's a first step," he said.

Another metric, the Energy Scale-Down Efficiency measurement, aims to check power use of systems operating at the low utilization levels typical of many of today's servers in the field. "This will be adopted faster because it is a big problem in our industry," Partha said.

An even more ambitious project in the works, called Power Struggle, aims to define an overarching scheme for coordinating power management across a broad range of systems and components.

"Today we have power management technologies at the chip, operating system, blade, system and data center levels. One of the big challenges of the next 3-5 years is how do we put these all together. Will they all work together and be globally optimal?" he asked.