Intel talked a bit about its upcoming 2007/2008 processors yesterday.
Two observations need to be made:
1) This is a die shrink with a few tweaks, folks Outside of SSE4 (which won’t have any real impact in real apps for sometime to come), the core is functionally little better than a C2D. Yes, Intel has increased the cache and really ramped up the FSB, but if a 3.2 Penryn can only do 20% better in some games than a 3.0 Conroe under those conditions, real improvement, clock-for-clock, is relatively minimal.
In the megamarket, the real issue is: Will Penryn be better enough than AMD’s new offerings to keep Blue on top punching Green in the face?
Intel apparently thinks so (and apparently is less than awed by what they expect AMD can do with its next generation if they think this is good enough), but they can bump up the speeds somewhat in case AMD does better than expected.
Of course, overclockers don’t have to wait for AMD to do good in order to bump up the speeds, but that brings up point two:
2) A CPU with a 400/1600MHz FSB is going to be a PITA to overclock, at least initially Those of you (and that includes me) who were thinking it might be a good idea to buy an Intel Bearlake-based mobo with a C2D sooner, then buy a Penryn later, might want to think twice about that. We don’t even know yet if the initial Bearlakes will even officially support such a speed.
That in-and-of-itself will probably not prove to be that big a deal, but C2D style overclocking with a Penryn could be a real problem.
Today, you can take a 200MHz or 266MHz FSB C2D and push the FSB to 400 or 500 MHz without too much problem. Taking a 400MHz processor and pushing it to 800MHz is a different matter.
True, it’s probably unlikely that every Penryn will be a 400MHz (the OEMs will see to that), but 333 is no bargain, either. High FSB is effectively an anti-overclocking device.
Yes, eventually motherboards will be built that can run at the higher speeds, but that could be later rather than sooner.
Of course, we don’t know how much faster Penryns will be overclocked (remember, this is a new formulation, past performance is not necessarily indicative of the future), and that’s the real question for us.
However, from a non-overclocker perspective, it doesn’t seem like Penryn brings a whole lot more to the table, unless they can run a whole lot faster than 3.2.
Put another way, if AMD’s new offerings can break even with current high-end C2Ds (even when giving away a lot of clock speed); Penryns aren’t going to be a whole lot faster than that.
My suspicion is that AMD’s new processors may well match or somewhat exceed C2Ds clock-for-clock, and might match Penryn, but they won’t be able to overcome the expected clock handicap.
Getting the clock speeds up is Job #1 for AMD.
We haven’t mentioned Nehalem, which Intel talked about a bit. At least some models will incorporate an integrated memory controller, which certainly boosted Hammer’s performance quite a bit.
However, given all the tinkering and extra cache Intel has been including in its processors in lieu of IMC, I suspect that IMC isn’t going to give the performance boost to Nehalem that it gave to Hammer.
That being said, at least Intel is talking about a new generation of processors. AMD isn’t. All they talk about is Fusion and Torrenza, and they’re essentially just new tricks and stunts for the Hammer core. Yes, they’re likely to be very useful stunts for the very low and very high end markets, but neither is a mainstream innovation, nor a substitute for a new architecture.
In the article linked above, there’s an interesting statement towards the end:
“This is the exact approach we would like AMD to embrace as well, keep the public informed, especially if you have exciting things to talk about. We understand the desire to try and protect trade secrets, but with the complexity of modern processors and the amount of information that gets shared between partners, not to mention “corporate espionage”, it seems likely that Intel and AMD already more or less know what their competitors are planning. Changing course late in product development is nearly impossible to accomplish, and especially when current products begin to lag behind the competition we would expect companies to try to garner interest by talking about the future.”
This is a very diplomatic way of saying, “You’d better start telling people what you’ve got brewing, AMD.”
The irony is, for all of AMD’s secrecy, Intel probably knows better than anyone outside of Green’s employ what they have, and if you read between the lines, Intel seems to be saying, “Not much.”
And if they don’t hear otherwise, directly, that’s what people are going to believe.
The problem, and maybe tragedy for AMD probably is, the honest answer probably is, “They’re right.”