Gonzhilla
Edit: Meant to quote andressergio.

Is 71-73 that hot really to where it might cause damage? I have the FTW3 Ultra 3090 with the hybrid kit and PWR hits 80c sometimes.
I definitely don't love it considering it's liquid cooled , and I've replaced everything with those super overpriced gelid pads like 4 times trying to get better temps,. Sick of taking it apart and putting it together again. I even bought a second hybrid kit and tried that. I think my card just runs really hot. GPU usually maxes out around 65c and I used a caliper and everything to get the pads right after I wasn't satisfied with original hybrid temps.

Just curious if anyone thinks that that gonna be problematic. It runs fine and can hit 500w, it's one of the OG first two or three passes of cards too from EVGA...

I mean I got a 10 year warranty but who wants to have to put the original cooler back on and wait a month for shipping...

Thank you for your reply this is the first time this happens in a Game that doesn't seem to have super graphics like Metro Exodus EE, for that reason I gave the example. I never seen that temps on the VRM's till now on any game or bench with the Hybrid Kit installed, but well, waiting for the RMA because it sucks that the card goes down to 1825Mhz from 2100 when I put a heavy load to it...
 
I don't like hight temps and the crazy ramps ups of the fan, it disturbs me a lot really. Hope RMA cards are better.
 
All the Best, Sergio!
 

For the hybrid 3090oc rma, I just got this. Anyone else?

what test are yall running to hit 500W? I can only hit about 460W on my new RMA card on timespy extreme. Pin 3 is still way lower than the others. Max TDP is 109% on the image below. Edit: Furmark max at 4k is 470W. I have precision x1 set to 119% power and +125/+750on OC. Did I get another dud?

Hello All, 
 
I have made some test runs comparing the v1 card and the V2 of the card. No I have no idea if my results are typical or just a one off. 
 
And for those that are TLDR:
V2 card didn't overclock as well, but maintained a higher clock speed during the stress test. V2 card also had a much higher GPU Mum junction temperature by over 10 deg C. 
 
See charts below:

vRel. VOp seem 5o be the culprit according to perfCap reason in GPU-z

badass1982
Well I'm not sure why but my card is pulling 470w on the (total power draw) pcie slot is fixed and doesn't really increase over 60 ish now , 8 pins are 146w, 151w and 99w .

Might try the kingpin bios. This is in the OC switch location. Port royal at stock with all sliders up is a measely 13 926 points.

Running at 3440 x 1440 g sync disabled all other programs shut down , running on a ryzen 3960x overclocked to 4.3Ghz on all cores.

Hmmm strange. My new card hit 508 watts just running a stock port royal benchmark without an overclock at all... Now with you saying its vRel and VOp in GPUz and not PWR then maybe its another issue? Not sure what those acronyms stand for so I am just guessing... I'll try to do a run when I get home and see what it states for my perfCap reason and update. My port royal without an overclock was right at 13.9 as well but I hit 508w. Almost hit 15k with an overclock but I have not really had the time to play like I did in September. All kinds of projects going on. But when I can I will take some screen shots and share them so it can hopefully benefit the entire group.

JThomas1215
My new card hit 508 watts just running a stock port royal benchmark without an overclock at all.

What revision card and BIOS are you running? If you've changed the bios, what bios did it ship with and what was the performance?

you have to use the old bios files to get 500 watts.the new bios files won'lt work.

I know but it works. That is we are doing to get 500w.I have made the exe files so you can flash it back and forth easy i will post them later.

dlomond
I know but it works. That is we are doing to get 500w.I have made the exe files so you can flash it back and forth easy i will post them later.

Who is we?
What bios Version is used?


"We" so far this thread is concered are using this 
https://www.techpowerup.com/vgabios/231621/231621

Hello all,
 
I would like to share my experience with the exchange program.
I purchased my 3090 FTW3 card at the end of 2020, but could put it into operation together with the rest of the components only a months ago.
The serial number startet with 2014 und it had red lips.

I tested two OC 500W Bios (94.02.26.48.F8 and 94.02.42.80.27)
No benchmark could exceed a power draw over 450W. Maximum value and not average. Most of he time I got around 400-420W. GPU-Z showed max. 105-107% TDP – although raised to 119% in Precision X1.
Also PCIe power draw sometimes exceeded 80W.

So I reached out to 3090oc@evga.com sending some screenshots and they approved the exchange.
I'm from Germany, so the next steps of the exchange went through EVGA Munich in Germany.

This is what I got in return
- 3090 FTW3 in original box (sealed)
- serial number starts with 2112 (label on box says made in Taiwan - I thought 12 means China)
- red lips
- BIOS 94.02.42.80.14 (non OC) and 94.02.42.80.15 (OC)
- PCIe retention tab show version 1.0

I'm irritated by the old BIOS versions (most of you seem to have gotten the version 94.02.42.C0.02) and by the red lips cooler.

Setting Precision X1 to 119%, +150 cpu clock, +800 ram clock and voltage slider to the right position.

Kombustor benchmark including FurMark and Port Royale showed 450-460W max. - but now continuously. 500W were not reachable.
PCIe power stays between 65-70W - so this issue is fixed.

I tried to update with BIOS with 94.02.42.80.27 from the start post like I did it with my old card.
This was not possible because of a mismatch of the PCI subsystem (so this card IS different from my orginal one?).

WARNING: Firmware image PCI Subsystem ID (3842.3987)
  does not match adapter PCI Subsystem ID (3842.3982)

Using nvflash with --protectoff and -6 I could flash BIOS version 94.02.42.80.27.
Thanks to orez-bus for the notice, that only with this bios he could reach the 500W.

Port Royal benchmark result: 14.611 (119%, +150, +800, max. voltage, stock cooler)
(AMD 5950X no overclock, Asus Crosshair VIII Dark Hero).

So everything is fine? I think so...
But I'm not so sure that I got exchange within the special OC campaign (BIOS versions, red lips - but new 1.0 at the retention tab).

Best,
Thirty-Thirty
 

koroem
Bought a new power supply for my rtx3090 with three 8 pin connectors. It is a Be quiet dark power pro 12 and I'm a bit confused by the pcie plugs. It comes with 5 pcie cables, each one has two 8 pin connectors. I know the wisdom being spread around here is to use separate cables for each plug.

On my old seasonic psu, both sides of the cable were 8 pins. So obviously one cable per plug. With this new power supply, the psu end of the cable is 12 pin instead of 8. Is this a true dual 8 pin connector? Is it safe to use both 8 pin connectors on the one cable or should I still be using three separate cables with the extra 8 pin tied back/off?

This is a difficult question.
The best answer is to email bequiet and ask them if each 12 pin block, has SIX *ACTIVE* pins of 12v connectors, so in other words, three live +12v connectors per group of 6 pins.
if it does, then it's perfectly safe to use the daisy chained 8 pin, because it means two of the connectors on each 8 pin are not used.
If you have proper measurement hardware, you could try taking a reading from the pins yourself, but you would need a pinout chart I think.
 
On some much older PSU's that had this 12 pin layout on the PSU box, where each 8 pin was actually linked to a 6 pin, only two +12v connectors were used (instead of 3) per each 6 pin block, but I don't know which PSU's were wired up like that.  That came from an old reply on jonnyguru forums, again without anyone telling which PSU had this "missing" 12v pin.
 
I do know that the old Seasonic SS-1000 XP platform and related lines around that time (circa 2011) had that layout on the back with 2x8 pins going to a 12 pin but again no idea if the third 12 pin per 6 pin block was connected on one, both or none of the 6 pin * 2 blocks.  Even if it was, I remember people saying that you should still use a separate cable for each 8 pin on something like a Vega 64 or R9 290X...

Thirty-Thirty
Hello all,
 
I would like to share my experience with the exchange program.
I purchased my 3090 FTW3 card at the end of 2020, but could put it into operation together with the rest of the components only a months ago.
The serial number startet with 2014 und it had red lips.

I tested two OC 500W Bios (94.02.26.48.F8 and 94.02.42.80.27)
No benchmark could exceed a power draw over 450W. Maximum value and not average. Most of he time I got around 400-420W. GPU-Z showed max. 105-107% TDP – although raised to 119% in Precision X1.
Also PCIe power draw sometimes exceeded 80W.

So I reached out to 3090oc@evga.com sending some screenshots and they approved the exchange.
I'm from Germany, so the next steps of the exchange went through EVGA Munich in Germany.

This is what I got in return
- 3090 FTW3 in original box (sealed)
- serial number starts with 2112 (label on box says made in Taiwan - I thought 12 means China)
- red lips
- BIOS 94.02.42.80.14 (non OC) and 94.02.42.80.15 (OC)
- PCIe retention tab show version 1.0

I'm irritated by the old BIOS versions (most of you seem to have gotten the version 94.02.42.C0.02) and by the red lips cooler.

Setting Precision X1 to 119%, +150 cpu clock, +800 ram clock and voltage slider to the right position.

Kombustor benchmark including FurMark and Port Royale showed 450-460W max. - but now continuously. 500W were not reachable.
PCIe power stays between 65-70W - so this issue is fixed.

I tried to update with BIOS with 94.02.42.80.27 from the start post like I did it with my old card.
This was not possible because of a mismatch of the PCI subsystem (so this card IS different from my orginal one?).

WARNING: Firmware image PCI Subsystem ID (3842.3987)
does not match adapter PCI Subsystem ID (3842.3982)

Using nvflash with --protectoff and -6 I could flash BIOS version 94.02.42.80.27.
Thanks to orez-bus for the notice, that only with this bios he could reach the 500W.

Port Royal benchmark result: 14.611 (119%, +150, +800, max. voltage, stock cooler)
(AMD 5950X no overclock, Asus Crosshair VIII Dark Hero).

So everything is fine? I think so...
But I'm not so sure that I got exchange within the special OC campaign (BIOS versions, red lips - but new 1.0 at the retention tab).

Best,
Thirty-Thirty
 

Have  you gained performance with the apparent boost in power?  There's got to be a reason they segmented the PCI ID on these cards. Please note this is 'seems to work for me' status. No warranties expressed or implied.

I am missing something when trying to use NVflash.  Do I have to do something to the rom file?  It always gives me the I\O Error.  I am trying to learn so please don't give me the if you don't know what you are doing you shouldn't do it.  
 

lucasfrance

Thirty-Thirty
Hello all,
 
I would like to share my experience with the exchange program.
I purchased my 3090 FTW3 card at the end of 2020, but could put it into operation together with the rest of the components only a months ago.
The serial number startet with 2014 und it had red lips.

I tested two OC 500W Bios (94.02.26.48.F8 and 94.02.42.80.27)
No benchmark could exceed a power draw over 450W. Maximum value and not average. Most of he time I got around 400-420W. GPU-Z showed max. 105-107% TDP – although raised to 119% in Precision X1.
Also PCIe power draw sometimes exceeded 80W.

So I reached out to 3090oc@evga.com sending some screenshots and they approved the exchange.
I'm from Germany, so the next steps of the exchange went through EVGA Munich in Germany.

This is what I got in return
- 3090 FTW3 in original box (sealed)
- serial number starts with 2112 (label on box says made in Taiwan - I thought 12 means China)
- red lips
- BIOS 94.02.42.80.14 (non OC) and 94.02.42.80.15 (OC)
- PCIe retention tab show version 1.0

I'm irritated by the old BIOS versions (most of you seem to have gotten the version 94.02.42.C0.02) and by the red lips cooler.

Setting Precision X1 to 119%, +150 cpu clock, +800 ram clock and voltage slider to the right position.

Kombustor benchmark including FurMark and Port Royale showed 450-460W max. - but now continuously. 500W were not reachable.
PCIe power stays between 65-70W - so this issue is fixed.

I tried to update with BIOS with 94.02.42.80.27 from the start post like I did it with my old card.
This was not possible because of a mismatch of the PCI subsystem (so this card IS different from my orginal one?).

WARNING: Firmware image PCI Subsystem ID (3842.3987)
does not match adapter PCI Subsystem ID (3842.3982)

Using nvflash with --protectoff and -6 I could flash BIOS version 94.02.42.80.27.
Thanks to orez-bus for the notice, that only with this bios he could reach the 500W.

Port Royal benchmark result: 14.611 (119%, +150, +800, max. voltage, stock cooler)
(AMD 5950X no overclock, Asus Crosshair VIII Dark Hero).

So everything is fine? I think so...
But I'm not so sure that I got exchange within the special OC campaign (BIOS versions, red lips - but new 1.0 at the retention tab).

Best,
Thirty-Thirty
 

What is your new board S/N? I should get one 2112 board on Tuesday from EVGA Europe too... Hope it will come with announced new bios version and black lips or I will return it...
WHAT A MESS!

Well I should get my RMA replacement too at Monday/Tuesday (at least UPS claims during 24th lets see)
I too hope that my replacements will be the proper one... (hope dies last, right? :-))
 
Will keep you posted
 
kind regards
Tomas