That's what I'd like to know tooOriginally Posted by STEvil
One thing that I'm certain about is that high frequency causes memory degration (or degration of the MC) on these cards. Another thing that I'm certain about is that current draw in an integrated cicuit like the memory or the MC (part of he GPU IC) increases +- linearly with frequency. I'm certain that an over current is damaging either the MC or memory ICs on these cards.
It's unlikely that the MC get's damaged in this situation as the temperature of the GPU has a great impact on current draw of this part and this temperature is very variable. Nvidia would have come across this issue and they would probably have fixed it before the release of these cards.
Seeing that it's most likely the RAM ICs failing on these cards we can conclude that the higher current going trough these devices is causing the issue. There are two other important factors that determine current draw in an IC: temperature and voltage. Voltage stays the same in this situation so temperature is the only other factor that's of importance here.
From what I've heard just a small increase in memory frequency can cause damage on these cards (DDR1800->DDR2000 is only +-11%. With sufficient cooling this will up the current draw by only +-11% as, pointed out earlier, cucrent going trough the device increases linearly with frequency. I don't believe an increase of just 11% can cause damage in a short periode of time. My conclusion is that the cooling device (thermal pads with a 'heatsink made of air' in the fan area
) isn't sufficient at a higher frequency resultion in a more than linearly increase of working current in the device. I think that with PROPER cooling the ICs are unlikely to suffer permanent damage at any realistic frequency.
Bookmarks