so.. been reading up.
I had a 3-way GTX 680, but one card was under volt. and my RMA failed cause this is Japan. (happens often here.)
In despair, I learned Volt mod, in hopes to fix my bad card.
and I think I found something.
there are 3 5ohm resistors that tells your GPU how much voltage is passing, and if I short them out, the read remains, but actual voltage is much higher, resulting is higher GPU base clock possible.
mind you, this card is on waterblock.. so temps probably not for those on air.
but as of now, I can get baseclock +450Mhz on first try.. and surprisely.. no crash. temp at 58'C on load on water. (and my card is an OC version, so base clock was at 1150)
my conclusion so far:
- It is probably possible, that nVidia purposedly underclock their Kepler chips for further upgrades and new cards. the Kepler chip to me, is CLEARLY capable of MUCH higher clocks. Though this is just day1, I am uncertain if the lifespan will be much shorter. ... nonetheless, thought I post my results.
3Dmark11 score, single card there was 13272.. (at clock 1600)
I had a 3-way GTX 680, but one card was under volt. and my RMA failed cause this is Japan. (happens often here.)
In despair, I learned Volt mod, in hopes to fix my bad card.
and I think I found something.
there are 3 5ohm resistors that tells your GPU how much voltage is passing, and if I short them out, the read remains, but actual voltage is much higher, resulting is higher GPU base clock possible.
mind you, this card is on waterblock.. so temps probably not for those on air.
but as of now, I can get baseclock +450Mhz on first try.. and surprisely.. no crash. temp at 58'C on load on water. (and my card is an OC version, so base clock was at 1150)
my conclusion so far:
- It is probably possible, that nVidia purposedly underclock their Kepler chips for further upgrades and new cards. the Kepler chip to me, is CLEARLY capable of MUCH higher clocks. Though this is just day1, I am uncertain if the lifespan will be much shorter. ... nonetheless, thought I post my results.
3Dmark11 score, single card there was 13272.. (at clock 1600)