IBM races for clock speed |
 |
By Matt Stansberry, News Editor
23 Feb 2006 | SearchDataCenter.com |
|
|
IBM is making the most out of the surprising performance gains of its new chip architecture, the Power6 processor, but more than a few experts are scratching their heads over the brouhaha.
Brad McCredie, Power6's chief architect said the primary focus of this launch centers on clock speed. IBM claims the chips are hitting between four and five gigahertz in the labs, much faster than IBM had been able to push Power before.
But clock speed, the frequency at which a processor can negotiate commands, isn't the ultimate measure of processor performance anymore. The performance gains that came from cranking clock speeds often came at a cost of hotter processors and power inefficiency.
In fact, many competitors and experts have been confused by IBM's move to bring clock speed back to the forefront. Charles King, principal analyst with Hayward, Calif.-based Pund-IT Research said clock speed is like talking about how high you can rev your engine while your car is parked.
"The industry, Intel, was going to get away from talking about clock speeds and start talking about heavy lifting, like multi-threading," King said. "But at the same time, with Power6 I think it will be a differentiating point. The competition is saying they've hit a ceiling."
Clay Ryder, president of the Union City, Calif.-based Sageza group seemed puzzled by the move as well. "Is a four-gig Power6 processor possible? I'm sure they could do it," Ryder said. "But I'm not sure the market is leaning toward that."
Jeff O'Neal, director of product marketing for Sun's Sparc chips said clock speed is how the industry has made performance gains on the cheap in the past. According to O'Neal, the better option for the future will be workload specific processors that run at a more relaxed clock speed, beneficial to both heat and power concerns.
Even IBM agreed that clock speed is not the only or even best way to measure performance. But McCredie said the pendulum has swung the other way and people have shifted too much emphasis away from clock speed.
According to McCredie, the industry has gained clock speeds by using deeper pipelines, the design that lends itself to power and cooling inefficiencies. But the new Power6 processors are hitting these higher clock speeds with the same length pipelines..
One reason the chip is able to do this is its advanced manufacturing process. This will be the first Power chip to be manufactured with 65-nanometer features (a nanometer is a billionth of a meter). Many chip manufacturers currently use 90 nanometer manufacturing. McCredie said the smaller size will allow for more density and performance improvement in the chips.
Many articles have reported that Power6 will become part of the next generation of IBM's zSeries mainframes, but that is false. McCredie said it would be theoretically possible to use Power6 on the zSeries, but no such plans are in place at this time. Though the chips will be available on IBM's iSeries, pSeries and xSeries servers next year.
Let us know what you think about the story; e-mail: Matt Stansberry, News Editor
|
|
|
|
