I have no idea if this will help but have you tweaked your settings in nvidia control panel?

Mathieas
Memory +1100/Core+160/voltage+25: 14581 maxtemp:58C
 
pretty much back, going to increase voltage to +100 to see if any difference is seen. Then I will hunt for daily stable.

That’s roughly what I was able to get on my FTW3 Ultra at +195 +1100 100% fans. Excited to get the Hybrid Kit


https://www.3dmark.com/pr/567220

QueueCumber

Justikins

Nereus

3090 FTW3 Ultra (NOT HYBRID) ..adding for comparison.
Memory +1000, GPU +150, Port Royal 14,100 - and that seems to be as good as I'm going to get on air on this card.

These cards seem to have a heavy thermal throttle.

Adding as comparison as well.
3090 FTW3 Ultra (Not Hybrid)
Memory +1000, Clock +205 , Voltage +100, Did not update the XOC Bios Beta for more voltage.
Port Royal 14442
Ran on air only inside a Lian Li O11D


Must have been a fluke for that run because I was never able to hit +205 clock again

Had stable runs with mem +1000, clock +195

Good hunting on your benchmarks!

Not a fluke. These cards seem inconsistent in general. But in this case, it is most likely changes in ambient temperatures affecting the entire system. A nice drop in temperature with an open door in the winter can push scores up a bit.

Thanks for the tip!
If that is the case, I'm going to bump my GPU clock a bit higher, remove some side panels and just SEND IT while its still cold and hope for the best haha. 

KingEngineRevUp
This is great but can we get more information on your fan profile and what the fans are running at when you hit these temperatures?
 
This is important because there can be more reduction in temperature with a custom fan profile. What is the fan RPM and percentage, it would help us all out a lot!
 
Lastly, have you thought about doing the Port Royal, Fire Strike and TimeSpy stress test? 
 
https://www.3dmark.com/tsst/1373822
https://www.3dmark.com/prst/47417
https://www.3dmark.com/fsst/1593784
 
Run these three so you're more comfortable going into some gaming. I get some people want to just do max clocks then they find out the hard way crashing during important parts of gaming. 

Thank you King,

I wi get these numbers for you this evening after work!

One this that confuses me is it seems I have lower clocks then most on here but a higher score. Is that due to hybrid cooler? Do I have a bad binned chip that is just consistent due to cooler? I did get stable Port Royal +175 once and consistently at +160... But others mentioned +180 being the norm and even higher.

This is good info, I want to thank everyone who is contributing.

mobhill
This is good info, I want to thank everyone who is contributing.

You are most welcome sir!

arestavo

Mathieas
So my average clock was 2,055 MHz for my +1100mem/+160core 14581 score. Does that jive? How does it compare to your run?

I wonder if I'm getting lower highest set clocks due to the fact that my card can stay at higher clocks longer due to temp being lower? Like air cooled might spend less time at highest clocks then my card thus allowing a higher OC spike frequency but achieving lower average freq? Is that off base?

Really good! My XC3 VBIOS top run was 2101 MHz average: https://www.3dmark.com/pr/542809
 
and my top XOC VBIOS run which was 1995 Mhz (remember, it's essentially the same as the stock OC VBIOS since my card is hard hitting power limits even at 400 to 450W on a 500W VBIOS and this took me days to achieve): https://www.3dmark.com/pr/430091 
 
This difference is why I am betting that the XC3 VBIOS will get you that 15K+.

Ah! So I'm actually doing really well? While I am a scientist by trade, I am a bit of a newbie with over clocking and I am having trouble comparing myself apples to apples to other cards.

How do the XC3 and XOC bios compare to my stock FTW3 bios and each other?

My card is a 2014, Taiwan :(. How would I test if I have power issue? Just run port royal and report max and average power draw from gpuz? Isn't the XC3 a lower version of our cards? Why is that bios doing better? Does it not have this glich? Any sense on if the problem is hardware or firmware issue?

arestavo

Mathieas
My card is a 2014, Taiwan :(. How would I test if I have power issue? Just run port royal and report max and average power draw from gpuz? Isn't the XC3 a lower version of our cards? Why is that bios doing better? Does it not have this glich? Any sense on if the problem is hardware or firmware issue?

Load up GPU-Z and look at the PerfCap / Board Power Draw and run a new game at 4K windowed (use DSR if you have to). EVGA's Jacob used Doom Eternal as his example. It will be obvious with the XOC VBIOS and the card shows power as the perfcap and it's under 450W board power draw and it will be real time. If you try and use average, as soon as you exit the game the average will start dropping. You could output it to a log file, if you wanted to dig into it that far.
 
Running at 1080P native (that's your resolution right?) will likely not push the card hard enough - except in benchmarks that run a higher resolution like Port Royal (1440P) or Time Spy Extreme (4K).

My resolution is 1440p. I'll try port royal, time spy and tomb raider tonight and report back. I have a second monitor so I can put gpuz on the 1080p screen.

Did Jacobs test prove there is an issue or was his run showing over 500W?

Mathieas
My resolution is 1440p. I'll try port royal, time spy and tomb raider tonight and report back. I have a second monitor so I can put gpuz on the 1080p screen.

Did Jacobs test prove there is an issue or was his run showing over 500W?

Jacob didn't say whether he has a Chinese or Taiwan manufactured card (and not all cards follow the same results, just the vast majority). He showed his results because there was a lot of stink made about it before end users made the connection themselves. His card did pull around 500W with the XOC VBIOS in Doom Eternal (he used the Geforce overlay from Geforce experience, the beta mode within it had to be turned on - and I refuse to install that garbage).
 
Jacob has stated, at least on Twitter, that they are still working this issue. His last acknowledgment of this problem on the XOC VBIOS thread was over a month ago now that they are aware of this and looking at it. Many folks have despaired of getting their Taiwan manufactured 3090 FTW3 to pull close to 500W - until one user discovered that the XC3 VBIOS bypasses whatever issue those cards have and allows greatly increased power draw. I only use it for benchmarking because the fans don't work correctly and drop below 100% while benchmarking and runs my air cooled card a lot hotter because of the increased power draw.

Mathieas
Ah! That's makes sense. Thank you for the detailed explanation. When checking power draw would peaks above 400 means I'm OK or am I looking for sustained numbers? I will take a screenshot of my gpuz when I run the test regardless.


Re XC3 bios: I wonder if my Hyrbid would have the same fan issue? If that bios gets around the issue that at least suggests there is hope that it is a bug rather then defect.

It's the average current board power draw that really matters, or average if you can trim off the low power draw before and after the test (log output, excel spreadsheet it and average it?). Peaks can be for only milliseconds and drop down to 400W, or 420W and hover around there (like my first card would (RMA'd for out of spec PCIE power draw when stock)). Or with the refurbished RMA, around 450W (with the 500W XOC VBIOS).
 
And according to Jacob, there is no difference in components between the chinese and taiwan boards. At this point, that seems to point to a manufacturing process difference if that is true.

raiden2a
Dont be afraid of the Taiwan vs China bs .  Vince said it in the live stream, dont be afraid and push your card to its limit. This generation is temperature dependent.  My PR on a Taiwan, day one card on air X0C bios: https://www.3dmark.com/pr/504619.   Cant wait for water.

Looks like a Chinese board to me, am I right? Serial number starts with 2012?
 
I've had two 2014's from Taiwan, and they are power limited (except with the XC3 VBIOS).
 
I also agree to not be afraid when overclocking. You won't kill the card unless going extreme with LN2.

arestavo

raiden2a
Dont be afraid of the Taiwan vs China bs .  Vince said it in the live stream, dont be afraid and push your card to its limit. This generation is temperature dependent.  My PR on a Taiwan, day one card on air X0C bios: https://www.3dmark.com/pr/504619.   Cant wait for water.

Looks like a Chinese board to me, am I right? Serial number starts with 2012?
 
I've had two 2014's from Taiwan, and they are power limited (except with the XC3 VBIOS).
 
I also agree to not be afraid when overclocking. You won't kill the card unless going extreme with LN2.

My card is a 2014, Taiwan card.

raiden2a
 
My card is a 2014, Taiwan card.

Congrats to you! You should probably go buy a lottery ticket as you are the very small minority, and really lucked out. I can only get that score with the XC3 VBIOS. Otherwise I top out at 14233.

Mathieas

arestavo

Mathieas
I ran Port Royal on 1000+mem and 130+ core. 
 
Average draw was: 429.7W
It started lower at 420W then ended with mostly high 430s to low 440s.

Sounds about right for the stock 450W VBIOS. If that's the XOC VBIOS, the power draw is low and would really benefit from the XC3 VBIOs.

This is the stock FTW3 bios at max power slider(107%). Says it tops out at 450W. Does that change your opinion of my chip quality? If I'm getting correct power then does that make my OC result low or average or still good? Thanks again!

Nope, looks good for the stock 450W VBIOS. If it does the same thing with the XOC VBIOS, then the XC3 VBIOS would help. But you may be lucky and be able to pull around 480W+ average with the XOC and not need to use the XC3 VBIOS at all.