The Inquirer reports some of the nitty-gritty details about shifting over to the BTX standard, most particularly in the cooling area.
It says that Intel wants to use passive heatsinks to cool its spawns of Satan. BIG, HEAVY heatsinks.
How big and heavy? Prototypes are weighing in at 1kg (2.2 pounds for the metrically challenged).
Just for comparison’s sake, that’s about 50% heavier than today’s heaviest-duty all-copper heatsink of today.
That’s for a nice little default system. God knows what an overclocker might require and what it will cost. Anybody got a copper mine?
Quiet people might like the idea of no buzzing fan cooling the CPU, but the BTX brick is only technically a passive heatsink. It’s supposed to be cooled directly by the system fan which is supposed to funnel outside air directly to the CPU area first.
That ought to be an interesting design, and again, an overclocker will probably want a heftier case fan.
For the average person, this is going to raise the cost of a BTX system by some tens of dollars. For an overclocker, it’s going to reduce the cost benefit of air cooling vs. water.
There’s a more fundamental issue here, though. BTX is beginning to look more and more like it really means Beelzebub’s Technology Extended (a definitionfor the Biblically challenged). Just about all the changes seem to be driven by Intel’s (and to a lesser degree video card maker’s) inability to control heat and power.
BTX isn’t the specification from Hell, it’s the specification for Hell.
As the Inquirer article implies, Intel’s answer to those problems is to basically pass the buck to the Taiwanese mobo makers. It will cost them more to implement these changes, and when costs increase, they’ll be the fall guy.
But even that is a relatively minor problem compared to the potential impact on the average customer. If BTX ends up meaning a more expensive, noisier, hotter box, a lot of Intel customers, especially Intel business customers, aren’t going to like that at all.
While current Hammers aren’t exactly cold, odds are that at least the lower-rated 90nm Hammers will use not much more than half the power of equivalent PressHots. AMD doesn’t seem to want to go BTX, and with that kind of power consumption, they won’t have to. SOI may well prove its weight in gold provided AMD can get 90nm out fairly timely.
There’s more than a few indications that Intel has realized it has been going down the wrong path, and are now trying to fix things. Unfortunately for
Satan Santa Clara, switching tracks will probably take a few years.
That’s a long time in Hell, well-ventilated as it might be.