Ed Moltzen

The Chart

December 23, 2005

The reaction I received to CRN's head-to-head competition between the Fujitsu Lifebook S2110 (based on the AMD processing platform) and the Fujitsu Lifebook S6240 (based on Intel's Centrino) got right to the point. To recap: In a head-to-head test between those notebooks, Intel won on performance and battery life, but AMD won on price.

Some folks were unhappy that the test found the Intel-based system performed better, and suggested it was simply an unfair test. Several wrote in and complained, variously, that the Passmark test was unfair, that the comparison between Lifebooks was unfair, and that Fujitsu's decision to build an AMD system with the specific type of graphics chip was unfair.

Wrote one emailer:

I thought the test was some what transparent, but as an engineer first thing I look at is memory type and memory speed, video card, and harddrive capacity. First is the test is not a true test since the memory type and speed are different and memory does make a major difference in speed and stability of a system. Intel system has DDR2 at 533Mhz speed while the AMD system has a DDR at 333Mhz speed. Also Intel using dual channel while AMD it does not say and . Therefore this is a major flaw in the test.

Next is the L2 cache, It shows a weakness in Intel with you have 2MB of L2 cache and AMD has 1MB of L2 Cache and the score is only 50 points difference. Everytime I deal with Intel there marketing always tell me to buy Intel because it has more L2 cache gives them 10x performance over their competitors.

Wrote another:

OK, I'm an "AMD fanboy", I'll admit it, but I'd also be the first to admit, that the Pentium M Dothan notebooks still consume somewhat less energy than even the best Turion64 MT based systems. (Though, as even the quickest Turion64, the MT40 consumes only 25W max - on die memory controller included - I guess this is probably the fault of all the other 3rd vendor chips of the notebook).

However, I do think that performance wise it would have been more of a tie than a hands down victory for Intel, if only you paid attention to the following…

...The MT28 is clocked at 1.6Ghz but, has only 512kb cache versus the MT30 clocked at the seem speed, however with 1mb of cache. Since the Pentium M has 2mb of cache anyway, the MT30 would have been a more appropriate counterpart, especially given that according to AMD's official price list, the MT28 and the MT30 both cost EXACTLY THE SAME, USD 150, that is. (Not to mention, that the Turion MT32 is only 9 (!) dollars extra and, although has only 512kb cache, is clocked at 1.8Ghz).

And, wrote another:

What (the test) does show is Fujitsu vs. itself. While battery life is an advantage held by Intel in the mobile segment, the performance differences in this test can be attributed to Fujitsu's design choices. The comparison is not meaningful to those seeking an answer as to how Intel and AMD compare. Why? A few reasons. First of all, the integrated ATI graphics chip included with the S2110 is known to be a poor performer (and it certainly isn't made by AMD). There are plenty of much-higher performing alternatives from both ATI and NVidia that Fujitsu could have chosen. The result would have been a notebook that completed the entire benchmark, and scores that were actually comparable.

The responses were thoughtful, measured and, notably, passionate. While past tests and head-to-head matchups between AMD systems and Intel systems always have seemed to have gotten responses, my guess is that seeing the test conducted, on video, before your very eyes provokes a deeper response.

A few thoughts, though:

The Passmark Performance Test 6.0 was used to compare the Fujitsu systems. It's a test that anybody in the world can use, free, for thirty days. Other benchmark tests, like the Sysmark 2004, can cost hundreds of dollars to download and use. We wanted our test to be so transparent anybody could do it.

Also, there were criticisms that the test was unfair because of Fujitsu's use of the ATI Radeon XPress 200M for the AMD-based Lifebook, while using Intel's 915 chipset in the Intel-based Lifebook. Some readers suggested a higher-performing graphics chip, integrated with the AMD Turion 64 MT28, would have given AMD's results a boost. No other Tier 1 vendor, though, even uses the AMD Turion 64 MT28. Hewlett-Packard, in a couple of thin and light systems, uses the ML 28 with ATI graphics. Alienware makes AMD-based notebooks, but uses the Athlon 64 and Opteron chips.

And AMD CEO Hector Ruiz has said he wants PC makers to have the choice of GPUs and chipsets in building processing platforms. In fact, Ruiz has criticized Intel for its aggressive marketing of the Centrino platform, which combines Intel chip, Intel chipset and Intel graphics into one box.

Here's what Ruiz told us earlier this year:

Systems makers want people to buy their brand because it has good value in it, not because it has Intel inside it. I think this is giving us an opportunity to go to customers that are becoming more interested in getting state-of-the-art CPUs from AMD and state-of-the-art motherboards from Foxconn to create a best-in-class system with their name on it. Our strategy is to continue to offer customers the option of not being marginalized. Any system builder that thinks warming up to the Intel platform brand is anything other than giving their soul away is nuts. They're naive.

(You can criticize Fujitsu's choice of graphics chip. But you can't say it gave away its soul.)

But how does this relate to the CRN test between the AMD-based notebook and the Intel-based notebook?

Here's now: No matter what dream "test bed" you may want to surround processors in comparing AMD technology to Intel technology, you'll need good luck and more to find a manufacturer who actually uses that test bed in a real product. AMD decided, on its own, not to build chipsets. Intel does.

A higher-performing graphics chip in the AMD-based system would probably have cost it its price advantage over the Intel-based system. And as Frank Ohlhorst, director of CRN Labs, noted in the segment on CRN TV, solution providers may opt for AMD-based notebooks for customers who are sensitive about price.

The Intel-based notebook simply outperformed a comparable AMD-based notebook. But the AMD-based notebook is still good and is cheaper.

That's the choice that AMD wants the market – and vendors like Fujitsu - to have.