Suckers: the Second Half of the Story

Continued from ‘SLI and CrossFire: There’s a Sucker Born Every Minute‘….

How’s this all relate to SLI and CrossFire? Well, it’s important to first realize that with SLI, you must use graphics boards containing identical model Nvidia GPUs, with identical frame buffer sizes. ATI’s CrossFire requirements are a bit looser; you can mix-and-match a limited set of familial-related GPUs, but the CrossFire combination will down-throttle to the feature set of the lowest-end GPU in the mix (i.e. if you combine 16-pipeline and 12-pipeline GPU, four of the higher-end GPU’s pipelines will be disabled). Hold that thought; we’ll get back to it in a minute.

Here’s the other key wrinkle; SLI (at least currently) not only requires Nvidia GPUs, it also requires that the system be based on a Nvidia core logic chipset. ATI’s CrossFire requirements are not yet known; at minimum, it’s reasonable to assume the technology will work with ATI core logic ;-). We’ll have to see if ATI also supports Intel or other vendors’ chipsets. So, using SLI as an example, you need to buy a more expensive, dual-graphics slot motherboard, based on Nvidia core logic, and containing Nvidia graphics technology. Is this scenario beginning to sound familiar?

If not, let’s continue. While it’s true that crazed power users (like Maury’s son, although I don’t think he ended up getting his wish) might bundle two high-end graphics boards from the get-go, the more common sales pitch goes something like: "Can’t afford a high-end $600 graphics card? Well, spend $400 on a mid-range card now….and a few months or years down the road, when you’ve saved up your allowance, buy another one for a resultant dual-card configuration….and you’ll have equivalent performance to a single $600 card". Now does the scenario sound familiar?

Same scenario, same Achilles Heel. A year from now, when you’re ready to buy that second graphics card, the pace of PC technology has inevitably advanced. 1.5 Gbps SATA HDDs have been replaced by 3 Gbps SATA HDDs. DDR2-667 SDRAM has superceded DDR-400 SDRAM. That 100 Mbps Ethernet transceiver is looking long in the tooth; motherboard-resident Gigabit Ethernet transceivers are now du jour. Etc….

Even if you decide to keep your year-old motherboard, the graphics vendors have another ‘card’ (pun intended) to play in luring you to upgrade. Remember, you can’t mix-and-match different GPUs, or if you can, they degrade to the lowest common feature denominator. That $400 mid-range card of a year ago? Well, it’s now $250, which is a good thing. But it’s now a low-end card, which is a bad thing. The former $600 high-end card is now the $400 mid-range card. And a new $600 high-end king has replaced it on the throne.

So you could spend $250 and fill that second graphics card slot. Or, with the $400 you had budgeted a year ago to spend at this point in time, you could instead buy the former high-end card, yanking out the existing, now low-end card in the process. Because "a few months or years down the road, when you’ve saved up your allowance, you can buy another one for a resultant dual-card configuration….and you’ll have equivalent performance to a single $600 card". Heard that one before?

I have to commend ATI and Nvidia’s marketing folks; they’re very clever. A few years from now, I’d love to see the stats on how many SLI- or CrossFire-capable motherboards contain two graphics cards. I bet they’ll look eerily familiar to anyone privy to the OverDrive upgrade data.

Hopes and dreams, the most potent of sales motivators. Your thoughts?

Happy weekend, all!