Lies, damned lies, and marketing
Originally published 2001 in Atomic: Maximum Power Computing Last modified 03-Dec-2011.
Marketing people. Don't you love them?
Truly, they are the grease that allows the mighty wheels of commerce to continue to turn.
And they're so helpful. Why, if I had one handy, he'd probably have given me a better metaphor to use just then. One that didn't refer to him as "grease".
Oh, all right. I'll turn off the Sarcasmatron for the rest of this page.
I've been feeling disturbingly mellow about computer gear marketing people lately, because I haven't noticed them committing any major crimes against truth in my immediate vicinity.
OK, new video chipsets keep coming out, surrounded by the usual pink foam of specification gibberish. Hard drive manufacturers keep making drives with an alleged capacity of X gigabytes that only format to 0.92X real gigabytes. Monitor manufacturers don't go out of their way to tell people that there's not enough phosphor on the screen to clearly display the highest resolutions. Intel have expunged from their Web site all evidence that they once alleged that the P4 would somehow be useful if you wanted to "talk to your fellow gamer in Russia", and that it would greatly increase sound quality.
(But Google remembers.)
And Microsoft... well, at least Microsoft have added some worthwhile components to pop culture gumbo.
Who cares about all that stuff, though? Advertisements lie. Big news. If the marketing people aren't really perpetrating a good, old-fashioned scam, then their pinguid venality isn't particularly entertaining.
It entertains me from time to time, because it gives me a reason to say things like "pinguid venality".
(If you don't know what "pinguid" means, let Chris Patten tell you.)
But I can appreciate that something more is needed if I'm to maintain an audience.
I actually feel for the people trying to market Via Technologies' current C3 CPUs.
An 800MHz C3, like the one I reviewed a while ago here, loses by a substantial margin to a 600MHz Celeron in pretty much any benchmark you care to name, despite costing more. But the C3 only draws an average of five watts in normal use; it's very cool-running. Presto, a marketing line. This CPU is "Cool!"
Via first point out that the C3 runs cool, which is absolutely true and important for battery-powered devices, and to anyone trying to build a silent no-fans PC. They then go on to rave about the C3's Cool Technology, Cool Performance, Cool Innovation and Cool Value. I think they then create a distraction and run for their lives.
Fair enough, I say. I doubt I could have come up with a better strategy.
AMD are dealing with the opposite problem. They've been making Athlons that substantially outperform Intel P4s running at faster clock speeds, while offering rather better value for money, for quite a while now. And yet Joe Average keeps buying the more expensive Intel chips, 'cos they've got a higher clock speed, and must therefore be faster.
Joe probably understands that a 2.2 litre Toyota Camry is not 10% faster than a two litre Honda S2000, but he hasn't quite managed to come to a similar realisation about PC processors.
If I were AMD, my ad motto would be something like, "Athlon. It's faster. It's cheaper. What are you, some kind of idiot?"
It's probably just as well that I'm not AMD, all things considered.
Instead, AMD decided to cry havoc and loose the dogs (or possibly the grease) of marketing, resulting in "QuantiSpeed Architecture".
QuantiSpeed is a new take on the old "P-Rating" idea, in which CPUs are specified not by clock speed, but with an abstract number which indicates the clock speed of the competing processor that a given chip's meant to equal. AMD took pains to point out that their QuantiSpeed numbers had nothing to do with the P4, no sir, nuh-uh, but everybody knew what they meant.
So you create your QuantiSpeed ratings, you print them on the chips, you tell your sales organisms to talk about clock speed exactly as often as Heinrich Himmler talked about rather liking bagels, you turn a blind eye to the fact that many people are going to incorrectly assume that an "Athlon XP Processor 2000+ with QuantiSpeed Architecture" runs at 2000MHz instead of 1667, and then you buy yourself a monocle and a white fluffy cat and start talking with a zoologist about the care and feeding of the alligators with which you'll be stocking the Death Pool in the underground fortress which you'll be buying with your profits.
I think that's the plan, anyway. There might have been more full stops in the original document.
The P-Rating concept got a bad reputation when AMD and Cyrix used it last. An IBM/Cyrix "6x86-300" CPU, for instance, actually ran at 233MHz. It was, for 3D games, not much better than a 200MHz Pentium. The P-Rating was based solely on business application performance, for which floating point power was pretty much irrelevant.
Fortunately, the Athlon XP genuinely does perform, for both integer and floating point tasks, like a faster-clocked P4. Yes, there are exceptions, but for most desktop computer tasks the current "XP somethingorother+" designations are quite fair. If anything, they understate the Athlon's performance compared with P4s, especially the cheaper P4s with DDR memory instead of RDRAM.
So in the Big Book of Marketing Cheats, the QuantiSpeed Wossname idea doesn't really rate.
And yes, before a zillion P4 enthusiasts e-mail me, I am aware of the fact that Intel currently holds the PC performance crown with their 2.4B and 2.53GHz P4s, running from a quadrupled 133MHz bus rather than the quadrupled 100MHz of previous P4s. For some tasks, these two top-end P4s beat a 1.73GHz Athlon XP by a quite wide margin.
Plough through the benchmarks in a big fat review of the new CPUs, though, and it's clear that the Athlon actually doesn't lose by much for a lot of tasks, including the ones that most home and office computer users are going to do.
Who's got the absolute fastest money-is-no-object processor doesn't matter much; most people don't care about a few per cent either way in the benchmarks, if they can get a PC that's, to use a technical term, a buttload cheaper.
Athlon XP 2100+? Maybe $US225 at the moment, from a reasonably reputable dealer. You can pay less than $US160, for a marginally slower XP 2000+. Here in Australia, Aus PC Market are currently shifting Athlon XP 2000+s for $AU451, delivered, including a nice copper Bitspower CPU cooler.
These top-end Athlons will drop straight into most reasonably recent Socket A motherboards, though you'll need a board with DDR memory to get full performance for more demanding tasks, and there's a good chance you'll need to update the BIOS to recognise the newest chips.
The P4 2.4B and 2.53, in comparison, are currently $US575 and $US675, respectively, if you're a good shopper. Plus the price of the motherboard to go with them, because old 4X100MHz P4 boards can't run the new chips at full speed. Yow.
QuantiSpeed Wossname, mind you, doesn't address the OTHER reason why people buy Intel systems.
If you want something with one minute of downtime per year, a cheap grunty Athlon box is more likely to have unacceptable personality defects than will an expensive, somewhat slower P4 system, provided that system has a good power supply and quality components besides the CPU. There are many very stable Athlon motherboards out there, now, but the target market for almost all of them still isn't the server community. P4 and, heck, even P3 server boards are likely to be less quirky.
Given the market success of Windows 98 and ME, though, I can only conclude that ironclad reliability isn't a high priority with most shoppers.
So QuantiSpeed doesn't bother me at all, despite the fact that it sounds like an unpopular Transformer who only gets to turn into a particle accelerator.
If a marketing initiative persuades people to buy better computers for less money, then maybe the marketroids have done something good, for a change.
