On Thursday 17 February 2005 16:38, Stephan Grupp wrote: > That's interesting about the Dells. What I have noted (especially in the > smaller case Optiplex systems), is that the fan noise is intolerably > loud when P95 is running.
Not uncommon ... > Crack the clamshell case a bit, and the fan > speed and noise drops considerably. I'm not sure that makes things > WORSE, as the air path is still through the duct and over the heatsink, > whether the case is closed or cracked open. Fan speed is controlled by a temperature sensor, I don't know where that is mounted. In any case the best solution is probably to replace the manufacturer's heat sink & fan(s) with better quality units. Hint, large fans running at a relatively low speed are generally much quieter than a small fan running faster to shift the same air volume. That's why laptops with tiny case fans are often distressingly loud. It's also why I like cases with lots of case fan positions; if you fit them all with low-noise fans, you can generally reduce the fan voltage and still get plenty of airflow through the case without making too much fan noise. Obviously you need to fit a good CPU cooler as well. Also, fit a cheap low-performance graphics card (with a passive cooler) rather than a high-performance graphics card unless you really need the performance for gaming etc. But in any case don't ignore the placement of the system unit - if it's in a well-ventilated area so that the input air to the case is cool and the hot exhaust air is not recyled back into the case, the system will run cooler and maybe a bit quieter as well. > > These high end P4 systems generate a LOT of heat. I wouldnt be > surprised if the CPU's go down a little sooner running P95 than they > would just grinding idle. I don't know what the relative aging rate caused by simply running hot is compared with the aging rate caused by continuous large variations in temperature. But is it certainly the case that hot components (not just CPUs) are much more prone to "random" glitches which can cause undetected errors making the final residue of a LL test run incorrect. Regards Brian Beesley _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
