To my understanding, the games are tested "out of box". For manual OC...
https://www.techpowerup.com/reviews/MSI/GeForce_RTX_2080_Ti_Duke/36.html
All data taken from TPU current reviews:
Power (Avg) - Average Power in Gaming
Power (Peak) - Peak Power in Gaming
Temp (Load) - Temperature at Load
Temp (OC) - Temperature at Load w/ OC
Noise (Load) - Fan Noise at Load
OC (fps) - What fps was achieved under manual OC
% Over FE - What % was that over reference Card stock settings
2080 W3 - Witcher 3 Score at 1440p w/ 2080
1080 W3 - Witcher 3 Score at 1440p w/ previous generation 1080
% Increase - Percent difference between generations out of box
Here's my take using data in the review ....
1. Note, I didn't bother listing max core clock, boost clock or memory clocks ... why ? ... cause it has no impact on fps... the highest clock does NOT correspond to fastest (fps)
2. Asus, nice job again on the temps; lower than everybody else w/o more noise. Not a big concern tho was everybody well under throttling point other than FE.
3. MSI again Gaming X takes "top dog" status for best OC'd fps performance tho by not as big a margin as with the 1080.
4. It would appear, again, that how well you do in OC fps, related to how much power the card takes in. Is it worth it ? Thats an individual decision.
5. The cards are delivering 25 - 32% more performance than they did last generation.... far less than the 2080s increase G2G.
6. There's only 4.7% difference between the FE and the top dog... and only 1% difference between the "top dog" and the nearest competitor.
7. The advantage of the best manually OC'd card over the "stock", non overclocked FE card is 16% .... a bit less than 18.5% we saw with the 1080 Ti series, but significantly lower than the 27% we saw with the 980 Ti series.