Is Intel Turning Into Apple?
I don’t mean that seriously, but it does seem like Intel’s been helping itself to Apple’s playbook this week.
Their new mantra “performance per watt” seems quite reminiscent of the claims Macsters used to make about the PowerPC using much less wattage than x86 designs. It was a bogus argument then, and it’s a bogus argument now. The only place where “performance” and “cool” play well together as metrics is air conditioners.
The V//V logo and brand, well, of course this is Centrino For Desktops, but you have to wonder if Intel has the notion of IntelPCs in the back of its mind somewhere down the road.
More ominously, here’s a “leaked” Applesque benchmark which shows an Intel system with two dual cores “beating” a single dual core Athlon 64 in a single application.
Perhaps the simplest explanation for this is that anybody is prone to do a big song and dance when they really don’t have anything to offer anytime soon.
In the short-term (i.e., the next twelve months), what Intel will actually put on the table seems to a bunch of stunts and tweaks to hold the fort until they really have something. Sort of like Willamette Reloaded, except that Willamette really was a new design, and the early 65nm processors won’t be.
For instance, to get some people hot and bothered, Intel popped a few extra current chips onto a current CPU. This is a stunt.
It would mean something if some of the future chips they had running included these items, but they don’t. All this concoction tells me is that Intel won’t have a chip like that until at least 2007.
The biggest impression I got from IDF is that, for the short term, the BS quotient is likely to skyrocket. A lot of their talk, again, for the short-term, set off my bullfecal detector
However, since AMD doesn’t look like it will really do much in 2006 that will significantly affect performance, Intel will get away with this in the overall market just like they did with Willamettes, but that doesn’t mean you have to help them do it.
If you’re one of those people who need to pick his spots in upgrading, it’s hard to see how machines are going to get appreciably better (at a reasonable price) than they are now until 2007, so if you’re hurting, you might as well take the plunge soon so you’ll get some serious use before the Longhorn or even 45nm era dawns.
Right now, 65nm looks like it’s worth skipping.