I have a XP2500 and an A7N8X rev 1.04. I know that this revision is suppose to be unstable at 400fsb speeds, even though its suppose to go that high. But Asus just came out with new bios updates that are suppose to make it more stable at 400fsb speeds. I was wondering if it would be stable at 400fsb, and if people with revision 1.xx boards were sucessful with overclocking to 400fsb. Also is it true that the XP2500 is an underclocked XP3200 and is more than stable enough to overclock to 400fsb. And its it better overclocking with 2 sticks of 256 in dual channel mode, or 1 stick of 512 in single channel mode. Thanks

I meant to say is it true that some XP2500s, like the green ones, are just underclocked XP3200s.

dual channel versus single channel will hold back your motherboard fsb a little - maybe 5mhz almost all rev 1 nforce2 boards can get over 400fsb with a newer bios. A 2500+ and a 3200+ are the same chip with a different label and different testing. My understanding is that if the amd plant is told to produce 1000 3200+s and 1000 3000+s and 4000 2500+s in a given week, they'll make maybe 9000 cpus, figuring 3000 will be duds. They then sort them according to max speed likelihood, which means putting those from the center of wafers first and those from the outer of wafers last. They then start testing the ones from the wafer centers to see if they will run at 3200+ speeds and set aside those that pass as 3200+s and I don't know what they do with the duds. Once they have 1000 that pass as 3200+s, they test the next ones to see if they will run at 3000+ speeds, again sorting them into a dud pile and a 3000+ pile. Then they do the 2500+ batch. If they run out of chips to test before they have enough 2500+s do they go back to the duds from the 3200+ tests and see if they will run at 2500+ speeds? I've never heard whether they do or not but the evidence is that no they don't. Overclockers who believe this will buy only 2600+s, protecting themselves against getting a 3200+ dud, but 2500+s appear to overclock as consistently as 2600+s which is why I say no they probably do not recycle the duds. The manufacturing process is refined enough that the inner chips end up being about equal to the outer chips. If you look at entries in overclocking databases around the web there is no correlation to model number and max mhz.

And after all the sorting, the cache circuit is added. It will be wasting money to add the cache size before sorting. Done that twenty years ago and the same process is still valid today in chip manufacturing.