This article is more than 1 year old
Benchmarks haunt AMD's Turion
Apples vs. Bananas
Chip and server makers have an awful habit of unintentionally highlighting their weaknesses by making a big deal of dubious benchmarks. Intel has done it. IBM has done it. HP has done it. Sun Microsystems has done it. And, most recently, AMD has done it when it launched the mobile Turion 64 processors last week.
The Turion chips were designed to make AMD more competitive against Intel's Pentium-M processors. Specifically, AMD is looking for Turions to find their way into the thin'n'light notebooks that account for an ever larger chunk of laptop sales. Customers seem fed up with lugging around massive lap warmers.
A key factor in the thin'n'light category is the balance a notebook strikes between performance and battery life. So when a vendor - in this case AMD - puts all of its attention on performance and doesn't say one word about battery life, you know the product in question might have some balance issues.
AMD rolled out plenty of performance benchmarks in front of the press, stacking a Turion 64 notebook against a Pentium-M notebook in office productivity, digital media and gaming tests. But, while the 2.0GHz clock on both companies' chips would seem to indicate an apples to apples comparison, AMD really had a rather special system on its side.
Since no Turion laptops are actually on the market, AMD created a "reference" laptop of its own. The AMD system ran on a 35 watt Turion 64 and had a graphics processor from ATI. That's a pretty handy pairing when you decide to compare it against a 27 watt Pentium-M with Intel's integrated graphics processor. See the AMD system specs here (PDF) and the Intel system specs here.
AMD could well have picked its own 25 watt part to stack up against Intel and used a Pentium-M laptop equipped with an ATI or Nvidia graphics controller. But what would that have done to the benchmarks?
"If they had compared a 25 watt system and a lower power graphics controller, the numbers would be down a lot," said Martin Reynolds, an analyst at Gartner and an especially close PC industry watcher.
An AMD spokeswoman insisted the company picked "the most comparable offering from the competitor" that it could find, even though it didn't actually do that. AMD declined to provide any of its own battery life measurements at this time because there are no production laptops on the market to measure, she said. Somehow it's okay to use a "reference" system for performance results but not for battery life results.
The only third-party battery life indicator AMD could come up with was from a MSI Megabook S270 review written in German. Google's translator tells us, "The Subnotebook weighs approximately 1.8 kg and is with its lithium ion Akku (4400 mAh) approximately 4.5 hours to hold out." So, if holding out is important to you . . .
Why is AMD being so coy?
"The answer is that the battery life isn't so good," Reynolds said, adding that Turion-powered systems could have up to one-third less battery life than laptops running on Intel's ultra low voltage products.
As Reynolds pointed out, AMD didn't do anything terribly unique with the Turion benchmarks. It saw that performance was the chip's strongest aspect and then tweaked the comparisons to make sure it outperformed Intel. Chipzilla has done this many times as well - one incident in particular comes to mind.
Potential customers, however, should be aware that Turion barely beat Pentium-M on numerous benchmarks even with the deck stacked in its favor. And it killed Pentium-M in gaming and digital media tests because AMD had ATI and a higher-powered chip on its side. An Itanium laptop would surely crush either of these chips on any benchmark - in those few minutes available before it melted to the desk. AMD should have picked processors with similar power envelopes and graphics.
By being so proud of its performance, AMD might have alerted all of the reporters covering the Turion launch to take a closer look at the benchmarks. Few reporters really did though. (We can think of just one that kind of hinted at the lack of battery life.)
This kind of trickery in the benchmark game does little for the vendor. It makes you question any future performance claims and draws extra scrutiny to the product. While battery life is a function of the chip, hard disk and screen, you can't help but wonder if something really quite odd is going on with the processor in this case - something that has a painful effect on your Turion laptop's life.
A number of readers have argued that the wattage on the Pentium-M would go higher once you include the memory controller. In addition, AMD measures TDP (thermal design power) differently from Intel. Those factors, however, still don't account for the graphics edge or the fact that AMD is right off the bat pushing much harder against the power envelope you want in a thin and light notebook. No matter how you slice it, AMD drilled right in on performance for a reason - because it would prefer not to talk about battery life.
Intel 'plans' Pentium M price cuts
AMD tweaks mobile chip roadmap
Intel's Q1 'a little better' than expected
AMD details its Turion mobile processor
VIA unveils ultra-compact media chip
AMD unveils Centrino spoiler