03whitegsr
Been doing some initial testing on the 3080Ti FTW3 and so far, I haven't been able to beat my XC3 hybrid scores. Pretty disappointing.

It is drawing up to 450W, but it can't get anywhere near the F/V performance. Fans at 100%, temps are right around 70C. Stopped at +150 capped at 2040MHz in timespy, think it was around 0.98V and 450W for a Timespy graphics score of 20,700. Managed the same score on the XC3 using like +225 capped at 2040 and drawing 365W.

Put a Hybrid cooler on the FTW3 and it should clock and perform better.

If only they were available.

Ok, I know I previously stated that I wasn't going to muck around much with OCing my new 3080ti FTW3 until the Optimus water block came in and I had it installed under water in my machine, but I couldn't help myself.
 
I set the three fans manually to 100% and looked at what the temps were like. I'm not used to numbers in the 70s and 80s on my GPU since I've been water cooling them for years, but they were low 70s with the fans on full blast, so I drove on.
 
I nudged both the VRAM boost and the GPU core boost up a bit at a time, and as I crept closer and closer to breaking 20k in Timespy (from a score running the 3080ti FTW3 stock in the ~18900 range) I got "break 20k-itis" and kept nudging them up higher, a little bump at a time.
 
At +1000 on the VRAM and +150 on the GPU core I finally broke 20k in Time Spy with a score of 20,034. It looks like the temps topped out at around 77C on the core (wasn't watching the VRAM temps) and the GPU core fluctuated in the low to upper-mid 2000MHz range. 3DMark reports the average GPU clock was 2008MHz with a max of 2070MHz and a memory clock of 1313MHz. I'll say this: at 100% on all three fans it was loud AF, but damn this thing was dissipating over 400W of heat and kept the GPU core at <= 77 C. That's impressive. And impressively loud.
 
Here's a comparison of this >20k run with the very first Time Spy run right after I'd installed the 3080ti and it was all stock, plus an example of the scores I was getting in Time Spy with my water-cooled 1080ti. Note: I'd gotten higher Timer Spy scores on the 1080ti before, but the OC that generated those higher scores was subsequently demonstrated to be unstable in certain other tests and programs and so I'd dialed things back a bit. I don't care about the highest benchmark scores I can ever get: I care about the highest benchmark scores I can get with an OC that's rock stable under all conditions, indefinitely.
 
I really am going to call it quits on the OC experimentation with this 3080ti FTW3 now. I'll revisit it again when I've got it under water once the Optimus block arrives. I know that even with the same settings as now I'll get higher scores under water if only because the GPU temps will be well under the threshold where the card starts thermal throttling, and as it is now it's over that temperature, so I should some steps above the current clocks just for temperature reasons. In the meantime I'll run a bunch of other benchmarks and tests, and games, with this current OC just to make sure that it's not just "Time Spy stable" but rather just stable, period. 
 
Edited to add: and here's a comparison of the 13940 score I just got in Port Royal with the 12950 I got in PR with the 3080ti FTW3 running stock. Now I'm getting "break 14k-itis," but I'm going to force myself to wait for the water block, and just trust that I'll already break 14k with the same settings just due to temperature. Lol, just broke 14k in PR just by not having HWinfo64 running at the same time. I'd wanted to get a look at the memory and hotspot temps, which seemed fine to me. Here's the 14020 in PR compared with the 12950 I saw on stock settings. The case of "break 14k-itis" has gone into spontaneous remission, praise Jebus!

03whitegsr

russsl8
Put a Hybrid cooler on the FTW3 and it should clock and perform better.

If only they were available.

Found a local guy with a hybrid that I was able to make a deal on. Straight out of the box with default settings I was able to almost match the max I was able to get out of the air cooled card.

I put in a bit of effort and ended up 3rd in the world for Timespy on the 5600X and 3080Ti. Might have to revisit my CPU overclock and toss the PC outside once it gets cold.

http://www.3dmark.com/spy/23867115

sethleigh
Ok, I know I previously stated that I wasn't going to muck around much with OCing my new 3080ti FTW3 until the Optimus water block came in and I had it installed under water in my machine, but I couldn't help myself.

 
The Optimus block is a monster.  Literally, the thing is heavier than you'd expect.
I just redid my loop in copper, and turned on nicehash overnight to stress it.  The thing stayed nailed at 52C.  I still have the regular BIOS loaded, and just came looking for some OC ideas.

set fan speed to 100% (graphic card and case fans), room temp 73.5F
+165 core, +1000 Mem (Timespy graphic card score 21429； PortRoyal 14430)
 
all default (Timespy graphic card score 20379)
 
the score looks good to me, but 100% fan is way too noisy.
I undervolt my card to 0.85V 1890Mhz or 0.9V 1965Mhz to play most of games. For Cyberpunk 2077, I have to lower frequency by another 30mhz, otherwise, the game will crash within 30 mins.
 
Now I'm waiting for the Hybrid kit to come in stock
 
 

jmazoso1
The Optimus block is a monster.  Literally, the thing is heavier than you'd expect.
I just redid my loop in copper, and turned on nicehash overnight to stress it.  The thing stayed nailed at 52C.  I still have the regular BIOS loaded, and just came looking for some OC ideas.

I'm totally expecting that. I bought this Lian Li mounting bracket that screws onto two of the ATX motherboard hold-down risers and has a little adjustable shelf to support the end of that card. I really don't want all that weight hanging off the PCI-E slot, even if my mobo has a thin steel reinforcement around that slot. No shipping confirmation from Optimus yet, but I'm still patient. Others have waited far longer than I probably will. The card's running fine on air for now.

I just finished my build with 3080 ti done first test and have one conclusion... I need waterblock!
 
here are the results:  /spy/23947747
 
 
Because I mount card vertically I made a little cheat on temps - I removed the side glass to get below 70C

ziemniak666
I just finished my build with 3080 ti done first test and have one conclusion... I need waterblock!
 
here are the results:  /spy/23947747
 
 
Because I mount card vertically I made a little cheat on temps - I removed the side glass to get below 70C

You and me both brother! I'm running the 3080ti FTW3 on air for the last week while waiting for the Optimus water block to show up. I did do some overclocking with the 3080ti on air with the fans manually pegged to 100%, but stopped before I ever saw any errors or performance regression.
 
Here's a comparison between your score and the best Time Spy score I got with my machine in its current configuration. Your side panel off thing combined with whatever difference in our ambient room temps may exist got you 3 C lower than mine, but I averaged about one step higher on the gpu core and also a higher memory speed. Since I stopped the OCing before hitting any errors I know there's undoubtedly still headroom to be explored once I've got this under water, not to mention avoiding whatever GPU core throttling I'm seeing in the 70s C that I won't see with it below the thermal throttling threshold. I had the side panels on my case, and the GPU is mounted just a few inches above, and facing a 280mm radiator with fans blowing straight up into the GPU, so it's getting plenty of airflow, albeit with air that's above room temperature from passing through a radiator first. I'm just not going to push this to the limit on air because there's no point since it's going to be on water soon.
 
Not to mention I'm hitting my monitor's 120hz refresh rate already at stock settings in pretty much everything I've tried so I haven't even been using my OC profile while gaming.

I've gotten my hands on a couple different 3080Ti's lately and figured I'd put up some of the results I've managed to get.
These are all in the same case and the highest settings I could get out of them with 100% fan speed, max PL, etc. I had the case side closed in some and open in others so have noted that as well.
 
Card Variant / Changes to card / case open
TS Graphics / core temp (hotspot) / Memory Temp
 
XC3 Hybrid (kit) / Artic P12 fans, Kryonaunt paste, Gelid Extreme on VRAM and VRM, Gelid Ultimates on backplate / Closed Case
20,744 / 45.9 (55.9) / 64
 
FTW3 Air Cooled / stock / Open Case
20,796 / 73.4 (83.3 / 72
 
FTW3 Hybrid / stock / Open Case
21,853 / 52.3 (59.7) / 74
 
FTW3 Hybrid / Arctic P12 Radiator Fans / Closed Case
21,740 / 57.8 (65) / 78
 
FTW3 Hybrid / Arctic Bionix P120 Radiator Fans / Closed Case
21,727 / 56.3 (63.5) / 76
 
FTW3 Hybrid / Arctic Bionix P120 Radiator Fans, third front case fan at bottom / Closed Case
21,855 / 57.8 (65.1) / 78
 
FTW3 Hybrid / Arctic Bionix P120 Radiator Fans, third front case fan at bottom, Artic P12 fan blowing into shroud fan / Closed Case
21,672 / 57.2 (64.4) / 78
This did have the lowest ICX PWR temps, but only 2-3 degrees cooler then the Artic P12 baseline
 
FTW3 Hybrid / Arctic Bionix P120 Radiator Fans, third front case fan at bottom, Artic P12 fan blowing into the backplate above the core / Closed Case
21,756 / 56.4 (63.7) / 76
 
FTW3 Hybrid / Arctic Bionix P120 Radiator Fans, third front case fan at bottom, Artic P12 fan blowing into the backplate above the core / Open Case
21,809 / 56.8 (64.3) / 76
 
FTW3 Hybrid / Arctic Bionix P120 Radiator Fans, third front case fan at bottom, Artic P12 fan blowing into the backplate above the core, Fans powered by motherboard / Open Case
21,687 / 56.5 (64) / 76
This test drew about 8W less total board and GPU power.
 
In the end, whole lot of time and a few $$ to find that the stock setup actually did the best anyway...for overclocking. The Bionix P120s did exactly about what the fan speed said it would do, with the P12s running at 1900 RPM, the Bionix running at 2100RPM and the stock fans at 2500RPM, the Bionix ran cooler then the P12s but warmer then the stock fans.
 
Now, noise level, that's a whole different story and the additional fans running at ~70% are considerably more quiet then the stock fans ever get and they do it while running cooler. The bionix are only marginally louder then the P12s where the stock fans a loud AF. Currently, the shroud fan is the lower limit on the noise levels. I would like to remove the stock fan and put a P12 on the shroud by it's self to see how that helps noise. Just need to find a connector so I can run it off the GPU.