sailer :
That's the fly in the ointment. Intel has delayed the Penryns, the X48 motherboard, and more because of no competition, also called no reason for them to release newer tech and only have themselves to compete with.
There's little reason for Intel to hold back on Penryn. They have to have the 45 nm tooling and process set up and at least somewhat ready to make chips in some volume already if they are shipping much of anything made on the process. This means that Intel has already sunk all of the costs into 45 nm already. The equipment is already bought and the R&D necessary to get the process up and going has also been spent already. Penryn, Wolfdale, and Yorkfield, and Harpertown are smaller dies and thus will statistically yield better and be less expensive to produce than the 65 nm units, meaning that it's cheaper for Intel to make a 45 nm chip than a 65 nm one. Also, Intel can migrate chipsets and such to 65 nm and move them off the old 90 nm tooling, resulting in smaller chipset dies that run cooler and cost less to make. Plus, Intel can sell the old 90 nm tooling or use it to make other ICs. Any way you slice it, it's nothing but a winning situation for Intel to stop making 65 nm CPUs and start making 45 nm CPUs as soon as they can.
Now with Nehalem/Bloomfield/whatever the other code names are, it might make sense to hold back on full-scale production and release if the competition is not there. The tooling is already being utilized by the Penryn/Wolfdale/Yorkfield/Harpertown generation and these chips are smaller and less expensive to make than the much bigger Nehalems will be. Intel would just be competing with itself and rendering their current lines of CPUs obsolete and unnecessarily killing profits. Also, the more 45 nm Core 2s Intel can run through their 45 nm tooling, the easier it will be to run Nehalems through there as the process will be much more mature. They are NOT rendering their 65 nm CPUs obsolete with the introduction of the 45 nm units as the 45 nm CPUs are by and large identical to the 65 nm units and more or less interchangeable in the market. (I'm talking about OEMs and such here, not enthusiasts. The 45 nm CPUs *do* render the 65 nm ones obsolete as they overclock better.) Performance clock-for-clock is within 5%, power consumption isn't all that much different, and the range of clock speeds is similar as well.
Some rumors have floated around that they are having some chip problems as well, similar in idea to AMD's Phenoms, and they are holding back until the problems get fixed. And since there is no competition, there is no need to hurry. So when does Nehalem come out; Q4, or maybe next year some time? Who knows.
This sounds more likely. Putting four cores on a single die is a complex proposition and it's no surprise that Intel has run into problems as well. They also don't have nearly as much experience in dealing with integrated memory controllers and point-to-point interconnects as AMD does, which probably makes it harder as well. Yes, the prototype PIII Celeron "Timna" had an IMC and the 386SL had a memory controller on-package (but not in-die). But that does not hold a candle to the five years and millions of K8s and K10s that AMD has had experience with. Intel luckily has the luxury of being able to wait and get everything sorted out before launch due to low competitive pressure, something AMD didn't have.
Of course, if it is the lack of competition that's causing the delay, then we are seeing some of the prophecies in the AMD side coming true; that without competition, Intel has no compelling reason to release new products and can end the cycle of cheap CPUs, leaving only more expensive ones in the future.
Intel will have a hard time ending the cycle of inexpensive CPUs. They drove the price of CPUs waaay down with the Core 2 introduction in the middle of 2006 to kick AMD in the face when they were already hamstrung with the ATi purchase. To reverse it, Intel will have to offer a product that is *much* better than the current units to be able to significantly raise prices, else people will just buy the cheap Core 2s. And if Intel suddenly jacks up Core 2 prices, AMD isn't anywhere near out of the game. A $230 Phenom 9600 may be an okay but not great performance deal compared to a $280 Q6600, but if the Q6600 shoots up to $500, even a $350 Phenom 9600 looks awfully damn attractive. AMD will certainly undersell Intel if they need to- that has been proven time and time again. I'd actually say that AMD regaining the top performance spot would do more to cause CPU prices to rise across the board than anything else. Intel has a history of selling slower parts at non-competitive prices when AMD is on top- look at the later PIII days and the entire P4 era except for the brief period when the Northwoods were the top dog.
I want the X48 to come out, along with a new quad to stick into it! And I don't want to wait for six more months or something.
The X48 isn't much more than a cherry-picked X38 anyway and there are a bunch of X38 boards out there today. And the Yorkfields are not all that different than the Kentsfields, unless you are a serious overclocker. And if you are, you can buy the QX9650 today if you want. I wouldn't, it's way too damned expensive, but it's out there. And even if you are an overclocker, you will probably get at best 20% more speed out of a Yorkfield than you will a Kentsfield from the reviews I've seen. Q6600s typically get into the mid 3s if you intend on replacing them in a year and the QX9650 will run in the low 4s with the same expected lifespan. You just have to weigh the options.