2080ti not boosting Like it should

[hlennie]

Well the new card was a little bit of a let down at first as it was only overclocking to about 1750 memory will do 1200 thought to my self I had come last in silicon lottery, tried numerous drivers uninstalling drivers totally in safe mode. still crap. flashed to the FTW3 bios no joy but I could set power to 130% now. was wondering should I send it back as it seem such a dud every where I looked people boosting to 2000+ easily.

Then I noticed in afterburner card would only use about 67% power and no matter what settings I tried it would not go above this search internet forums like crazy these last two weeks could not find a thing.

so finally I removed the AIO I have on it and reapplied the liquid metal put back in computer and no boot for a second I thought I had killed me card had to boot up with built in gpu and could not find the 2080ti in device manager thought it was well strange. like an idiot I had not pushed one of the power cables in fully. pushed that sucker back in and booted.

started up a game to see if any thing had changed and now the card is boosting like it should its hitting around 2085 @100 % power 120+ core memory +1200. 0V. temps idle 26/27C Gaming its hitting 46C

Have any 2080Ti owners had similar problems?

 

[Grim5]

Yes it's common - the Turing cards seem to be very sensitive to the cooler mounting, if it's even slightly wrong the card will throttle itself back very hard. There was another person on this forum a week ago with the exact same issue and re-seating the cooler fixed the problem for them too.

https://forums.overclockers.co.uk/t...2080ti-extreme-poor-time-spy-scores.18854291/

Gigabyte even shipped Auros Extreme 2080ti's that wouldn't boost properly and/or overheated because their QC failed to find that the cooler's weren't quite mounted correctly which led to RMA's

On a side note: I used to always play games with MSI stats on so I could see the GPU temp and clock speed - and then eventually I realized the clock speed was pointless unless the card was at max load because Turing likes to load Cuda cores before it ramps up the clocks, it's a good thing because the cards are far, far more efficient using cores and keeping clocks low. So now I've enabled the power draw which makes much more sense to use as a guide to see what kind of load the card is actually under and does it hit the load that it should be doing

 

[hlennie]

thing was the temperatures where ok loaded it was hitting 60C but with redoing the liquid metal 45C. Even at 60C I had read peoples 2080ti boosting over 2000. so I really don't think it was temp problem unless mine is really sensitive lowest clock ive seen now is 2055 highest 2150 but not stable at high clock.

side note. dropped my core to +100 and now boosting to 2100 and more stable.

 

[Grim5]

As per the other thread I linked to, that person also had about 60c and thought "these temps are fine" - they are not fine for power draw you were seeing, that's why when you fix the mounting/paste you get that massive temp drop.

The bios is able to detect that the card at 60c is too hot for the load and it bottlenecks itself not to overheat - if it didn't cap it and let you just pull the full 130% power draw then who knows where the temps would end up, maybe 90c+

As a quick example for you, you say the card capped itself to 67% power draw - that is 165 watts on a FE card - if I cap my air cooled asus strix 2080ti to 165w it runs at 45c under max load and if I had an AIO on it it should be even cooler than that - so yes at 60c it was overheating