C1s Are Erratic
We have a very unusual picture shaping up when it comes to C1s. I’ve been heavily perusing databases and forums, and this is the picture that’s emerging:
1) Low-speed C1s are beginning to show up: People are beginning to find 1.8A and 2A retail boxes with C1 stepping, though they are few and far between.
2) They do or they don’t. There seems to be three kinds of C1s right now: those that hit 3.0GHz pretty easily, others that can be coaxed, and some that won’t hit 3 for love or money.
It looks like most fit categories one and two, and I suspect some in category three are having other problems, but I can’t honestly say that all of them are having other problems.
The distribution of results on these chips is very odd. Normally, you’ll see close to 100% success up to a certain point, then a quick dropoff. That’s not happening here. You have a sizable proportion, perhaps a majority of C1s able to hit 3GHz with little voltage help, but a significant proportion that does not.
3) The more expensive ones are now doing better than the cheaper ones. While some of the early high-speed C1s did poorly, they do seem to be doing better overall than the lower-speed ones. Not enough to justify the higher price, but enough to make one suspect that Intel is binning more than usual.
4) The 3.06 looks to be something else. While there are few reports of these around, those that are seem to indicate that the 3.06 does significantly better than the other PIVs. I am not at all suggesting you lay out $700 for one, it’s not worth it, but there seems to be something different about these chips.
Perhaps it’s binning, perhaps something else, but there’s a noticeable difference which we’re likely to see in affordable chips down the road.
5) Voltage is not too enticing Something quite a few people are reporting is that these C1s jump up the scale very nicely at default or little more, then they hit a brick wall, and get little more no matter what voltage you throw at it.
6) Prime isn’t fine A number of people are reporting that Prime95 is the bugaboo in their stability tests; they’re running other benchmarks at higher speeds, but Prime collapses at much lower speeds.
7) Getting ahead of the curve Expectations were high on these chips, but people’s expectations appear even higher than that. Realistic expectations on these chips were around 3, but people seem to be setting their expectations around the occasional wonder chip. It doesn’t work that way, that’s why people call them wonder chips.
I see threads like “I want 4GHz.” Well, that’s nice, but odds are you’re not going to get that no matter what you do. A few will, but that’s more luck of the draw than anything else.
8) There seems to be a bit of improvement as time goes on The more recent the chip, the better it seems to do, but there is no dramatic improvement at a certain point like we saw with the 1.6A. Nor is there much data, which makes the statement even iffier.
It’s Still Pioneer Days
While the C1s are hardly bad CPUs, they aren’t yet quite meeting the “3GHz with your eyes closed” expectations for them.
If that’s the situation you need to buy, especially if you need to buy a new mobo to go with a new CPU, don’t. If you have a socket A system, just go with a cheap TBredB when they come out.
Given the not-quite-there-or-even-there C1s combined with the not-quite-there-or-even-there Granite Bays, along with Intel’s promised spring offerings (which will absolutely, positively support Prescott and 200MHz FSB and other goodies), probably best to forget about this combo and revisit this in the spring.
A spring PIV/Canterwood system will probably not be significantly topped at an affordable price for about a year on the Intel side, probably with a Prescott/DDR II mobo. That seems a better long-term bet than C1/Granite Bay now.
It’s disappointing that the C1/Granite Bay didn’t pan out, but the silver lining is that Intel thinks Northwoods have more life to them than initially thought, and the reason why Granite Bay doesn’t look so good now is primarily because it’s near-term successor looks better than expected.