Hello. I'm yet to overclock a graphics card, but I think I have a sound theoretical understanding of it. It seems a good idea to check with you guys before I kill one of my cards. [Cards 9600gso, 8800gt, 280 gtx]
There are three main things to overclock, core frequency, memory frequency, shader frequency. I'm using nVidia cards, so I believe I can change these independently. Am I right in thinking that putting core frequency up will make the processing core hotter, memory frequency will make the ram hotter, and shader frequency will leave temperature basically unaffected?
Assuming that to begin with I leave voltages untouched. I intend to overclock the three separetely using evga's tool, testing with some form of stabilty program, and with folding at home, to see where the upper limits are. I assume I have to leave it running for long enough for temperatures to equilibriate for the test to be valid. The quickest the card will run is then a compromise between these three values.
Core frequency important for gaming fps, shader frequency important for folding, memory frequency similarly important for gaming? The relative importance of these seems difficult to find information on.
Having found and tested a set of values I am comfortable with, I shall back up the card bios, edit it with nibitor, and reflash, hoping for the best. Ill then retest, and if all is stable, move onto the next card.
Is there anything obvious I'm missing here?
The motivation for flashing the bios of the cards is that I use linux. The intention is to install xp, spend time overclocking the q9550 which really shouldn't be sat at stock, and overclock the graphics cards on the same installation. Obviously I wont stability test the graphics cards on an unstable processor.
Cheers guys
Jon
There are three main things to overclock, core frequency, memory frequency, shader frequency. I'm using nVidia cards, so I believe I can change these independently. Am I right in thinking that putting core frequency up will make the processing core hotter, memory frequency will make the ram hotter, and shader frequency will leave temperature basically unaffected?
Assuming that to begin with I leave voltages untouched. I intend to overclock the three separetely using evga's tool, testing with some form of stabilty program, and with folding at home, to see where the upper limits are. I assume I have to leave it running for long enough for temperatures to equilibriate for the test to be valid. The quickest the card will run is then a compromise between these three values.
Core frequency important for gaming fps, shader frequency important for folding, memory frequency similarly important for gaming? The relative importance of these seems difficult to find information on.
Having found and tested a set of values I am comfortable with, I shall back up the card bios, edit it with nibitor, and reflash, hoping for the best. Ill then retest, and if all is stable, move onto the next card.
Is there anything obvious I'm missing here?
The motivation for flashing the bios of the cards is that I use linux. The intention is to install xp, spend time overclocking the q9550 which really shouldn't be sat at stock, and overclock the graphics cards on the same installation. Obviously I wont stability test the graphics cards on an unstable processor.
Cheers guys
Jon