
So i have an 1080 ti and the way the GPU market looks atm i figured i have better luck trying to overclock my current and wait for next. Downloaded MSI afterburner and tried +100mhz core clock but the benchmark just crashes with that, won't run at all. Same with +75 and +50. Only core clock i can get running is +25 but is it even worth it at that point,….?
For memory overclock +200 works, anything lower doesn't work and higher i havnt tried yet….this is all very confusing :X
2 Comments
You can’t increase the clock speed too much without increasing the voltage too.
Alright mate, for reference, I use a 1080ti on water and keep it very cool (think sub 30c loaded). That said it is just less boost bin dropping etc, you can still oc it on air.
You want to do whatever you can to lower core temps as boost bins are affected by this.
Repaste the card too will help.
This card is hamstrung by power limits before voltage limits. So you don’t need more than 1.050 core voltage, I do 2100mhz with that. Any higher voltage and it smacks power limits and I have to downclock. – so more voltage isn’t necessarily more core clock here.
Basically every 1080ti can do 2000mhz core all day easily, and most would do 2050. Some 2100, very few over that. This is all unmolested of course.
*Monitor your perfcap with gpuz as you do all this*
I would use the curve to overclock the core up to somewhere between 2000-2100 on 1.050 limit. You can just up the power slider but don’t touch the voltage slider.
1080ti will be easier to oc if you don’t try do it on the 1.093v limit.
I generally run a comfy daily 2060mhz core 1.050v using curve and +500 mem
Ymmv with temps. But yeah try mess about with the curve optimiser and don’t Max the voltage, you’ll run out of power fast. And do memory after core.
You can back the core off a bit for some memory oc when hitting power limits, but results changes there won’t change a whole lot. I just do this to beat hardware combo specific 3dmark records.
My next step is shunt modding, so I can use the last of the vcore.