Well, I took a look at some of those dirty pictures today. 🙂 (If you don’t get it, go here.
The Hammer mobo pictures, of course.
There’s an easy way to tell the difference between a Clawhammer and a Sledgehammer mobo.
The Clawhammer CPU socket (officially socket 754) has a little square hole in the center.
A Sledgehammer CPU socket (officially socket 930) won’t have a whole because that’s where the extra 176 pins will go.
Only Gigabyte flat out called its mobo a Clawhammer mobo, the others were more coy.
Of all the mobos I saw with that little square hole, though, none of them mentioned dual-channel DDR. Indeed, most of them only had two memory slots, another rarely mentioned limitation of Clawhammer. (The couple that had 3 memory slots followed the “three slots but you can only fill them with two slots’ worth of memory” we see rather frequently lately on the Intel side.
There was one mobo that did promise dual channel, the Asus K8M. However, there’s no indication which socket it is, and the socket itself is covered up.
Why Are You So Obsessed By This?
The games and applications the members of the audience like prefer more bandwidth to less bandwidth. Intel is going to dual DDR to get more bandwidth (and get the RDRAM monkey off its back). So is Via, so is SiS.
This looks to be a potentially significant advantage to Intel.
Nor is this a matter of a strictly technical AMD design decision. It’s not a matter of lower memory latency from an integrated memory controller or dual DDR. You can have both, and that’s just what AMD will give you with Sledgehammer. Not the processor for whom you see these mobos.
So AMD obviously thinks dual DDR is a good idea, too. It just looks like they’re not going to give it to you with Clawhammer processors.
Obviously, we don’t know a lot of important facts yet.
We don’t know how much dual DDR will help the PIV, yet.
We don’t know how much dual DDR will help the Hammers, yet.
We don’t know how much more a Sledgehammer system with the goodies will cost over a Clawhammer system without them.
But it’s certainly more productive to identify what to look for down the road than just ogle pictures.
Top Secret: 1.6GHz?
As mentioned in yesterday’s article, the speed of the Hammer is “top secret.”
Well, in that case, a certain mobo manufacturer committed treason, since they listed the speed of the Clawhammer at 1.6GHz. See for yourself here and here.
Hmmmm, they got that number from someplace, and since this was just a mockup, it wasn’t like this was a working demo running at that speed.
If that number is semi-solid, unless the AMD engineers have been doing absolute wonders with reduced memory latency lately and not telling those giving technical briefings, it’s hard to see how a processor running slower than current Athlon XPs is going to blow the doors off a PIV that will be running faster than current ones, almost twice as fast as 1.6GHz.
Not saying it can’t be done, but show me.
I remember the last Computex. A year ago, Computex was filled with starry-eyed people saying that nForce was going to conquer the world. I was rather more skeptical.
Well, did it?
It was delayed, it was not so wonderful, it was too expensive for too long (though it did prove to be less buggy than I had suspected).
So when I see yet another PR blitz with much sizzle and little if any steak, it pays to beware.