Cooling Processors quietly
15 posts • Page 1 of 1
I've run my passively cooled NB Asus P4P800 at 250FSB for extended periods of time with absolutely no instability. I can't see why 260FSB would be much harder on it. The NBs are rated to some fairly high temps (I've seem 80-100°C tossed around for 865/875 chips) and your Zalman cooler is probably more efficient than the little stock Asus cooler on my NB.
Sigh....Maxamus wrote:Reading here in another forum i see people mention a "sudden NB failure" that OC'ers observe and it has got me worried.
People seem to be getting all mixed up here.
The "SNDS" that you're probably referring to is an acronym for "Sudden Northwood (note: NOT Northbridge) Death Syndrome. This has absolutely nothing whatsoever to do with overheating the NB itself. This refers to OCer's that are killing their P4 Northowod CPUs by a process called "electromigration" which comes from running too much voltage therough them for too long. It's not a heat issue at all and has nothing to do with the NB chipset.
You'll need to look into some of the big pdfs that are chipset specific. Start by looking, eh nevermind, I'll find it for you.Maxamus wrote:ive been busy trying to find anything on intel's site that has a recomendation for NB temps but i cant find jack!
..Ralf comes back 20 minutes later with the link:
ftp://download.intel.com/design/chipset ... 252803.pdf
Look on page 17, table 3.
See it says "99°C".
Show this link to all the lost souls that are just spreading FUD and we'll all be on the same page.
Don't worry about your NB temp. You will be fine.
I've found it's easier to work from the top down. For example, here's the log from overclock testing on my Athlon XP using prime95 for stability verification:Maxamus wrote:ahh well OCed it to 3.3GHz at default VCore.
Ran prime95 for 2hrs
Ran 3dMark03 3x
All ran successfully and no problems as of yet.
gonner try to see if i can get it upto 3.4
You can see by starting high, I spent a total of only about 15 minutes (including reboots) to reach the point where it was stable for 2.5 hours; after that test the next one ran stable for over 15 hours. In my book, anything over 12 is good, and if you look at the time incrases for each 1mhz drop in FSB, it would seem that 166 was the magic number to get total stability.Prime overclock testing:
2158 (13x166): Fail in 0.03 mins
2148 (12.5x172): Fail in 0.1 mins
2137 (12.5x171): Fail in 0.6 mins
2122 (12.5x170): Fail in 2.7 mins
2112 (12.5x169): Fail in 3.2 mins
2098 (12.5x168): Fail in 8.4 mins
2084 (12.5x167): Fail in 2.5 hours
2075 (12.5x166): Pass 15+ hours
Just as an FYI, prime95 says "failed after X minutes" which is not very precise. What I did was bring up task manager and base my time on the actual minutes and seconds of CPU time used by the prime95 application. This is a much more accurate way to go about it when we're talking about speeds where it's failing in 0 to 3 minutes.