The tricores will begin inauspiciously with TLB-challenged B2 stepping tricores, running at 2.3 and 2.1GHz. If you’ve been holding off on buying a Phenom because they’re just not broken enough for you, you’ll soon be able to buy broken broken CPUs.
To add insult to injury, these won’t be available until March. Geez, it’s one thing to be delayed getting fully-functional pieces out, but delays in getting broken ones out?
Yes, I’m probably being more than a bit unfair, tricores won’t just be broken Phenoms, but putting out B2 tricores in March makes no sense. If you have a bunch of broken ones now, you might as well put them out now. If you don’t have enough of those and have to make unbroken tricores, why would you want to make a bunch of B2 tricores? Surely it would be better to put your limited resources into making unbroken B3 tricores.
B3 tricores, broken or not, are supposed to show up sometime in the second quarter, at 2.1, 2.3 and 2.4GHz.
These tricores are supposed to have a “high price/performance ratio.” Well, at 2.4GHz max, it sure isn’t going to be a performance barnburner, so what’s the price likely to be?
Dual-core K10 designs are supposed to show up towards the end of the second quarter. They haven’t announced frequencies yet, but I’m going to go with my gut, and climb out on a limb to predict them.
How fast is a processor that never sees the light of day?
The whole scheduling of products seems to be upside down at AMD; it seems more like a stunt than a true generational production ramp. Intel is going to make four-five millions Penryn quads a quarter, less than 10% of total production, but the Penryn production ramp is going to be for Wolfdales. Dual-cores will continue to be the main breadwinner for Intel.
In contrast, AMD is putting K10 duallies last, and do you know what? At this point, I would bet we’re never going to see any at 65nm. As indicated in the second link above, AMD has already decided it will end production of the lowest speed 65nm X2s. A few weeks ago, they announced a bunch of new 65nm X2 models.
Why would they do all that if they were just going to replace them a few months later?
A quad or tri-core at least offers more cores than an X2, but what does a dual-K10 give you over a dual-K8? More precisely, what does a 2.3GHz K10 give you over, say, a 2.6 or 2.7GHz K8.
I’d say little if anything. So if you’re AMD and you’re really tight on money, and you’re not going to get any more money for them, why bother spending lots of money to convert to 65nm K10 duallies? Spend the money on 45nm and make dual-core K10s if you feel like it then.
Then you don’t have to worry about the two AMD lines competing against each other, and the OEMs can just keep cranking out the bargain boxes they are now cranking out, just with slightly faster processors.
It seems to be a sensible cut to me under the dire circumstances.