Noting that the power supply is often the component that consumes the most energy in current PCs, Barroso highlighted findings from a newly published Google white paper.
The paper, >“High-efficiency power supplies for home computers and servers,” by Google engineers Urs Holzle and Bill Weihl, states that Google has managed to increase the typical efficiency of power supplies from 60-70% to at least 90% efficiency, reducing lost energy by a factor of four.
Barroso and other Google engineers believe that home computers can be made just as efficient as Google servers. Toward that end, Google, Intel, and other partners are proposing a new power supply standard. Assuming the new power supply design gets deployed across 100 million PCs running an average of eight hours a day, Google estimates a savings of 40 billion kilowatt-hours over three years, which translates to $5 billion at current California energy rates.
The problem with today’s power supplies, according to the paper, is that they were designed to provide multiple output voltages. In 1981, chips needed this, but not today. Yet because power supply designs have not changed, power supplies continue to be over-provisioned and inefficient.
Google servers, and the new PC standard Google is proposing, use a simplified 12V power supply that generates a single voltage. When certain motherboard components require something different, the power can be modulated using voltage regulator modules.
Google estimates 85% energy efficiency can be achieved at virtually no cost, while spending about $20 more for higher quality components can lead to over 90% efficiency.
… the ultimate goal, as envisioned by Intel’s terascale research prototype, is to enable a trillion floating-point operations per second–a teraflop–on a single chip. Ten years ago, the ASCI Red supercomputer at Sandia National Laboratories became the first supercomputer to deliver 1 teraflop using 4,510 computing nodes.
Intel’s prototype uses 80 floating-point cores, each running at 3.16GHz, Justin Rattner, Intel’s chief technology officer, said in a speech following Otellini’s address. In order to move data in between individual cores and into memory, the company plans to use an on-chip interconnect fabric and stacked SRAM (static RAM) chips attached directly to the bottom of the chip, he said.
“Performance matters again,” Otellini said, disclosing that the quad-core desktop processor will deliver 70 percent faster integer performance than the Core 2 Duo, and the quad-core server processor will be 50 percent faster than the Xeon 5100 introduced in June. One reason performance didn’t matter to Intel during the last couple of years was because it was getting trounced on benchmarks at the hands of Advanced Micro Devices’ Opteron and Athlon 64 server and desktop processors. That all changed with the introduction of the Core 2 Duo chips this year.