Theres no worry about performance, i mean...its a 3080, as above, all 3080's pulling the same wattage will be virtually the same.
And largely, an 50w power target in games wouldn't translate to much at all anyway. 


But then theres those of us that like to tweak everything out of the card that we can get, we watercool for more performance, etc, etc and end of the day, it should hit 450w and it isn't because of a design flaw eVGA is just happy to stay silent on (no doubt they'd have stayed silent on the other issues if they hadn't hit the headlines and prompted a response, some forum posts are never going to be enough to prompt it, you'd need a few highly regarded youtubers to take it up and they won't unless they can use it to their advantage somehow).

So yeah thats all there really is too it, we want to push the card GPU and the MEM till it won't go anymore and as it stands, theres heaps of headroom left but its hamstrung.
and also thanks to how GPU Boost works, a watercooled card will for example keep smacking the power limit and downclocking over and over (requiring a custom curve to be set in) thanks to low temperatures but not enough power to go further

Really if they came out and said "Sorry guys, we F'ed up and theres no way to fix this, the card is capable of achieving 400-420w and thats all we can do", i think many more people would be "somewhat" happy as opposed to radio silence (that would never happen because you know people, instant reason to RMA, ask for refunds, etc). Really they should just "accidently" leak a custom test bios that plays outside of spec and this issue would probably disappear lol

Darkiv

JulienA1
I have an EVGA 3080 FTW3 Ultra Gaming, what should I do to see if my card is faulty ?

My EVGA 3080 FTW3 Ultra died, the last week, playing The new world...
I had the revision 0.1.
 
I RMA'd this gpu dead, I hope they send me a 1.0 and not a 0.1 :/

I don't think there is a 1.0 for the 3080 FTW3.
I got a brand new one from a big computer shop in France and it's rev 0.1. I had to wait for a week until they received it

I hope the RMA process will go smoothly, keep us updated :)

JulienA1
I don't think there is a 1.0 for the 3080 FTW3.

Check Martin's post earlier in the discussion.  Apparently there are some component changes on the 3080 but the PCB itself didn't get a new rev number.  Perhaps it was a component level swap only and the 3090 actually got a revised PCB.

MartinUK
Yea, its not really a problem. Just people (myself included) are a bit disappointed they cannot hit the 450w they expected. When its pulling 400w it performs every bit as well as any other 3080 that is also pulling 400w. Its just that its not quite as fast as a card that is pulling 450w.
 
Since the early batches of 3080 FTWs could hit 450w then people with more recent ones feel like they are missing something. Originally there was some speculation that this behaviour was indicative of a faulty card (which would be bad), but it appears that its just related to some recent design/component changes with the newer batches, so it least its not anything to worry about.

Can you guys confirm that you and another user pulling more power are actually getting different benchmarks?

Check average clocks and voltage. Do it at same offset and memory clocks.

Just because someone draws 450W and your 415 doesn't mean that extra 35W is doing anything or sensors can also be wrong.

MartinUK
It appears this may well be a feature, as opposed to a bug. Seems that at some point, even for the 3080 FTW3, EVGA switched from the UP9511 voltage controller to the same Onsemi NCP81610 used in the 3090 FTW3 1.0 cards and the 3080Ti. (The power stages also changed from AOZ power stages to Onsemi, same as the 3080Ti)
 
The controller is mentioned in this 3080 FTW3 review from March:
EVGA RTX 3080 XC3 en FTW3 Review - Tweakers
with a picture of it on the board:
2004196198.jpeg (5292×3814) (tweakers.net)
 
It seems that along with this switch they tightened up the power balancing to reduce the PCI-E slot draw, and to limit the 8-pin #3 power draw, which stops the cards from being able to pull more than 400w, but also presumably stops them dying randomly (since that was the fix for the 3090's to stop them dying).
 
It makes sense they switched to use the same components on all the boards, for manufacturing/sourcing reasons. 50w is a pretty heavy price to pay, but I suppose its better than your card blowing up.

I thought they "fixed" the 3090 by actually having more balance across the 3 pcie plugs to achieve 500 watt power draw. You make it sound like they capped the power draw on the 3090 to stop them from dying.

A bit old but I stumbled upon the thread searching for which PCIe power is which one. So I decided to make an image to make it clear in case anyone else sees that :)