famich
Hello guys, AFAIK these cards have a bad PCB design, this has been posted and even proven many times.
if you try Strix or Suprim , you are going to have power draw 450W without any problem.
Do not expect anyone from the company to accept or confirm that. Period.

Seems that way. What's frustrating is some 3080s manage to go up to 450w, others don't. I'd probably say most don't.

Given EVGA officially released a 450w BIOS for the card it's really on them to explain why most cards can't power balance beyond 400w.

Did you ever get to the bottom of this I have just fitted a LHR versions and flashed the xoc bios and its doing exactly what yours is doing not going over 395-400w, to be honest I don't really need 450w but it would just be nice to know that it actually works and can pull that if it needs to.

va02stephen
Did you ever get to the bottom of this I have just fitted a LHR versions and flashed the xoc bios and its doing exactly what yours is doing not going over 395-400w, to be honest I don't really need 450w but it would just be nice to know that it actually works and can pull that if it needs to.

EVGA customer service won't acknowledge it, Jacob won't acknowledge it and it's pretty much don't talk about it at a company level.

It's a hardware fault with the power delivery system on these cards. Either EVGA cheapened out or something is flawed in the design of the FTW3 cards. They did not fix it when transitioning from non-LHR cards to LHR cards.

It's fixed on other models like the Ti and I think the 12GB model. Good luck getting EVGA to RMA you though. Even then you're most likely to just get a refurb/RMA with exactly the same power delivery issues.

Yeah that's what I had thought, nevermind it is what it is I suppose nothing we can do ☹️ I suppose it dosent effect the use of the card as apart from that the card works perfectly thanks for your time to reply 

Hopefully Jacob has a solution for us soon.

Problem is I don't think Jacob or evega acknowledge it's a fault so changes are slim unfortunately 

Apaul82
Hopefully Jacob has a solution for us soon.

Nah Jacob doesn't respond to questions about it.
 
Design flaw that EVGA would rather just sweep under the rug, please upgrade to a 12Gb or Ti model, thanks.
 
I'm surprised they even bothered releasing a 450w bios, that has just caused more pressure for them.

So, here is where things get interesting. I've previously tried a Gigabtye 3080 LHR 10GB bios and it just totally broke power delivery, even although it was a 450w BIOS and 3 pin card, it seemed the EVGA card got limited to 2 pin. The sub system ID also changed, but this is quite normal.
 
So I dug up another BIOS I never paid attention to in the past because it's not 450w. It is instead, 430w. Also, it's sub system ID, 1462 3897, is the same as the EVGA card VGA Bios Collection: MSI RTX 3080 10 GB | TechPowerUp Again, sub system ID doesn't really matter too much as long as you know the BIOS you are flashing is from a like-for-like card. But it's interesting to note anyway the MSI card has the same ID as the EVGA card whereas the Gigabyte does not.
 

 
Power delivery reporting is off again, pin 2, lmao, but more importantly pin 3 is responding this time unlike on the Gigabyte BIOS where it reads as 0w. We are still pwr cap limited, but ignoring the 366w max which I believe is wrong, the card reported use is at 111% in the screenshot and was kissing 113~114% during the run. A 430w BIOS has a max of 116% on the slider.
 
On the EVGA 450w BIOS for this card it pretty much stops responding around 99%, maybe if I am lucky I would see 101~104% spikes.
 

 
So I increased the resolution in Kombustor to 2560x1440, again, pwr reporting cap (this is fine, at my temps I'll always be power limited). More importantly the power usage is again around 111~114%, and importantly at 2560x1440, just with a +100 on the curve (no finesse of manually handling voltages), I'm maintaining a 2055mhz core. With a power cap of 400w, maintaining such a core in Kombustor at this resolution was previously a challenge for me. Running at 1995mhz was more likely, and that was with a manual voltage curve to eek out efficiency. A quick +100 on core, +1500 on memory and still maintaining 2055mhz on core at 2560x1440 in Kombustor at 1.042v suggests 430w in action.
 
Now, I don't have proper equipment to try and find out the actual power draw, but using my HX1000i digital reporting info, with the CPU "only" at 128.8w and the power supply reporting 620w in (much higher than I've ever seen before), I actually believe this BIOS is letting the card draw up to 430w.
 
Look at all my Kombustor pictures previously in this topic capped at 400w. They weren't getting the same core clock. 552w power out was my prior limit on my HX1000i screenshots, it's now 579w power out. Pretty much 30w more.
 
I need some more help testing this BIOS, but this looks like it COULD partly be a crappy BIOS from EVGA hurting these cards. These cards do seem to have PCB/power management issues, but maybe there is hope to brute force out a bit more performance, even if it's "just" going to 430w!
 
If only EVGA and Jacob would actually communicate with us, because this seems like I may have stumbled upon a competitor BIOS from a 3 pin LHR 3080 10GB that shows hope for us getting more than 400w max.
 
I might create a new topic to try and catch the attention of EVGA and others to test this BIOS, but for now, I'll try and do some more testing of my own. Initial results look promising.

Card is definitely drawing more power  (again ignore the 341.9w below from rivatuner, power reporting figure is just broken using an MSI bios on an EVGA card)
 
Metro Exodus Enhanced Edition, in Ultra, with Raytracing at 3440x1440 is able to hold 2100mhz on core, +1500 on memory at 1.031v and shows spiked usage on the GPU of around and above 109% of the 113% slider max. So, pretty much getting near power limit and likely will in some scenes.
 
Metro Exodus EE (with raytracing on) is great for GPU stability testing, chances are 2100mhz will crash 1.031v pretty quickly, but the important testing here is the GPU pulling enough power to maintain such a core clock. I've never been able to maintain 2100 in Metro before. Even aiming for 2070/2085 under the EVGA BIOS conditions could be difficult and that was me trying to run 1v or under due to power limit. 2055 was pretty much my best.
 

 
Sorry about the phone screenshot but this game behaves weird for me when trying to do print screen or even change the system volume sliders. Seems to boot in some sort of 'exclusive' fullscreen mode or something.
 
edit - Yup, quick to crash, but that is OK. 1.031v for 2100mhz would be some galaxy-tier silicon  Most important thing here was discovering the card is indeed drawing more power than the EVGA 450w BIOS.

Some more testing to show the BIOS works as it should up to 430w
 

 
1.05v is at least allowing me to play at 2100mhz (doesn't mean it's fully stable, but it's not crashing within a minute or two). So, at 111% GPU power usage out of 118%, still holds the frequency.
 

 
On this in-game scene with an NPC and a fair bit going on, you see a jump up to 118%, this is the max for this BIOS, so power level hit. Voltage drops and core frequency drops. As expected. Because yeah, Metro Exodus with RTing on at Ultra is just a monster. Can eat 430w for breakfast.
 
It returned to 2100mhz when the scene calmed down.
 
Now, how on earth do I contact EVGA about this? The last conversation I had with EVGA CS when they were going to speak to the BIOS team resulted in me being ignored, never got a follow up. I've quoted Jacob on this forum, and was ignored.
 
This BIOS works to get this card drawing up to 430w, but I'd very much like EVGA to fix their BIOS and make it so we don't need to use an MSI BIOS on our EVGA card. Given the wattage monster that is Metro Exodus, that 20w extra up to 450w would be even better, but I would take a small win with EVGA releasing a 430w BIOS.
 
edit - Power supply power usage has shown a spike to 700w in/652w out whilst playing Metro
 

 
I'm beginning to think this BIOS is allowing the card to draw more than 430w, as I cannot "find" what else would be causing me to see a spike to 700w. Around 600w power in is what I used to see on the EVGA bios. Maybe a little more. Nothing else in my build has changed since I made this topic. Even giving my old pictures in this topic some breathing space past 600w to 610~615w, that's still 80w extra now.
 
I guess I'm questioning whether or not the MSI BIOS sets a 430w limit or if, by chance, it just happens to interact in a way with this EVGA 3080 that it essentially sets no power limit and lets the card draw whatever it can. Safe, in theory, unless there is no balancing protection on any pin and it's a free for all lol. My card isn't getting any hotter than 42~44 degrees playing Metro on this BIOS, but I guess I should open my case side panel and check how hot the pins are.
 
Another reason why EVGA need to look at this and figure out why a 430w MSI BIOS is "unlocking" power draw on this card past 400w.