Intel is making changes to the PIV faster than they can get them out.
They just announced a new change-that-isn’t-a-stepping.
The change boils down to a small lowering of voltage across the PIV line.
Currently, default voltage on the PIV is 1.5V. The C-1 stepping PIVs will have a default voltage of 1.525V.
These new post C1 processors will come in at either 1.475V or 1.500V. No, you won’t be able to order a 1.475; it will be luck of the draw.
Half Step, Two Step
Like the PIV design itself, it seems that Intel is now making less of a change faster when they modify the PIV design.
For instance, they first optically shrunk the B0 Northwood design, then made a few tweaks and called that a C1 stepping. Two steps rather than one.
I suspect this is the same thing; first change the voltage parameters, then make other changes and call the second set a new stepping.
So I think in a couple months, we’ll see a D0 or D1 stepping being announced.
The 3.06GHz will likely have this change on introduction.
The 2.8 and 2.66GHz are supposed to have this change starting December, though January is probably more like it.
Lower frequencies won’t arrive until late January at earliest, realistically, early March is more like it.
There’s been more than a few words said about 3.06GHz processors possibly needing new mobos. This is almost certainly the reason
why: a bit higher wattage combined with a bit lower voltage means more amperage, which some built-too-closely-to-spec mobos may not be able to handle.
It is very likely that whatever is true for a 3.06GHz and a particular mobo will also be true if you overclock this particular model.
Worth Waiting For?
If you’ve been waiting and waiting for a PIV system, no.
In all likelihood, this will only have a minor effect on overclockability. A little better isn’t worth waiting a few extra months.
Even if it is better than a little better, Prescott is on its way. The summer matchup will be Prescott vs. Clawhammer; no point waiting forever, then three months later, get faced with a new generation.
Unless . . . .
Amazingly, the 1.8A will continue in production and go through this change.
This is good news for two reasons.
Even before this, the C-1 stepping 1.8A (which will be the SL6LA) looked likely to become the best overclocking chip. Ever. We’re expecting a relatively easy 67% overclock to 3GHz with these (SL6LA) chips. Any further improvements (SL6S6) would make it even better.
Even given current pricing, many hobbyist could easily buy a few versions of these chips, and keep themselves amused for the first half of 2003.
Even better for overclockers and consumers (but not so good for certain competitors) is the prospect that current pricing modes may not stay that way next year.
You have to wonder just what Intel has in mind for these chips. Under normal circumstances, the 1.8A would have been discontinued already or be on the verge of it. There’s over a dozen current Intel chips out there now, with more coming.
Yet here we have these Intel documents implying at least another six months for it.
My suspicion is Intel is keeping the 1.8A and other lower speed Northwoods around while it’s deciding (or has decided) whether or not to use these chips to . . . uhhh . . . finish off AMD.
Mike Magee of the Inquirer recently said:
“[W]e have heard that Intel might apply the pricing axe at the lowest clock speeds for this desktop processor at unprecendented prices. $40 for a P4? That kind of price would, of course, totally undermine its current Celeron pricing strategy, but then this is war. And Intel still means to win the war, whatever it takes.”
Keeping the low-end ones around would seem to fit into such a strategy.
No, this doesn’t mean we’ll see $40 PIV prices for sure.
This isn’t a “hurt AMD move.” No, $40 PIV prices mean Mortal Kombat, and it’s a “finish him” move.
If Intel really wants to kill AMD, there will be no better time than the next six months. If Intel makes AMD’s financial situation even worse with an all-out price war, unless they find a savior, AMD would probably go down.
If you’re Intel, you do this to kill AMD before Hammer gets a chance to get off the ground and give AMD a new lease on life.
Intel’s price moves up to now have been meant to cap the amount of money AMD can get at the top end. This move would kneecap them where they make their sales.
It would be an extremely aggressive move on Intel’s part, very risky to its pricing structure, and can easily blow up in their faces for a number of reasons, no matter what the outcome.
If AMD finds a white knight willing to pump a billion or so into the company, the move fails.
Even if it succeeds, such a move is likely to get the attention of antitrust regulators both in the U.S. and Europe.
So one can and should doubt that this will ever happen. What one shouldn’t doubt is that the idea is on the table and being discussed in Santa Clara. Leaving the LSNs around is a way for Intel to leave its options open while they’re deciding, and they just might do it under the right circumstances.
In other words, certainly don’t count on it, but don’t be surprised if it happens. Spring 2003 could get very, very interesting.