Am Sonntag 21 August 2011, 19:26:47 schrieb Nikos Chantziaras: > On 08/21/2011 07:08 PM, Mark Knecht wrote: > > On Sun, Aug 21, 2011 at 8:43 AM, Nikos Chantziaras<rea...@arcor.de> wrote: > >> On 08/21/2011 06:33 PM, Mark Knecht wrote: > >>> On Sun, Aug 21, 2011 at 8:12 AM, Nikos Chantziaras<rea...@arcor.de> > >>> > >>> wrote: > >>>> On 08/21/2011 02:19 PM, Francesco Talamona wrote: > >> [...] > >> > >> The RAM gets hot when there's RAM load (meaning being used heavily), > >> not > >> when there's CPU load :*) > > > > Do you feel heat when your PC is turned on and running hard? Of course > > you do. The whole machine heats up. The CPU under load heats the > > machine so the RAM and drives and everything else heats up also. Not > > as hot as the CPU, but it heats up. So I might agree with you - the > > RAM might not be 'hot', but it would certainly be 'warmer'. > > > > I'm not suggesting that this would cause a normal DRAM stick to go > > bad, but only that if he had a very marginal bit of RAM that it might > > go out of spec... > > On a laptop maybe. On a desktop, the air around the RAM modules get > maybe 1 degree C warmer (I know because I have temp sensor there, > connected to the front panel).
me too - and the direction of air flow from the cpu cooler plus the warmth of the air can change the temperature of the ram sticks by 5°C. So... No, not 'laptop maybe'. > > When it does get warm is when there's GPU and disk load. Those suckers > combined can raise the temp inside the box by 5-6 degrees. those can raise it even more. > > The meaning of all this is that if memtest can't find any errors after a > full run (which can take an hour), the chances of getting an > error that > is really related to RAM under CPU stress are very slim. a full run is more than 24h. Everything shorter is only of worth if you already hit errors after a few minutes. -- #163933
