known strong 478 & 939 overclockers?

Hi Folks

Going a bit older school than I thought. Found a guy with boatloads of 478 and 939 cpus. Anyone know of a listing of part numbers, steppings, etc for known, generally solid cpus for overclocking (like q660 g0 slacr for 775)?

tia!

Pentium 4 SL7PN and SL7B8 were good...Prescott 478.

There are no known good clockers, each CPU is only as good as the silicon it's made from, two same CPUs in the same board will give different results, its just luck as to how good of a CPU you get. The site linked is an excellent resource for referencing part numbers to figure out what CPU is what, I use it all the time.

http://www.cpu-world.com/Sockets/Socket 939.html

As far as socket 939, the newer steppings generally perform better and overclock further than the older ones. For instance E6 stepping is nearly always better than E3 or E4, at least in my testing of them with air cooling.

Clawhammers, Winchesters, and Newcastles don't typically overclock that well, on air at least. They usually net me another 300-400MHz, maybe 500MHz with a really good one.

Venice CPU's with a decent board, good cooling, and some good ram you're pretty much guaranteed a 600-800MHz overclock (up to 900-1000MHz with the low clocked chips with high multipliers (Athlon 64 3200+, and some others).

San Diego's aren't bad, but they're stock speed is set not far away from their max speed no air. I usually get them up to 2.8-3.1GHz.

Manchester dual cores aren't bad (E4 stepping), I usually got a 500-600MHz overclock from my Athlon 64 4200+ X2.

Toledo-based dual cores (E6 stepping) are king as far as the dual cores are concerned, I typically get 800-900MHz overclocks out of them. I ran one daily at 3GHz (stock on that one was 2.2GHz). It ran a little warm at that overclock, but it was still stable.

There's an overclock database here, I'm not sure how often it gets updated these days though.

Thanks folks! Appreciate all the feedback and guidance.

keny- I 100% agree with you (and also dig around cpu-world from time to time). For example, even if I got a Toledo-based E6 dual-core it may not overclock at all. That said...I'd like to imagine that it'd have a higher probability of being a better overclocker than cpus that, in general, seem poorer overclockers (the more submissions per cpu the better, of course...sample size=1=0)