Shraf2k

talon951

Shraf2k
 
 
I'm thinking the same thing. one of those fastboot related issues that got way overblown. 

Except that the screenshots show 105% PL vs 118% PL.  Not the same bios.

I know it's odd, but this happened to me the first time I flashed a bios but didn't have FB off. He info etc registered the data but it wouldn't ACT like it had the XOC bios on it. I had to go back in and turn off FB and actually flashed it twice using their file. TBH, I have less issues using nvflash but this file is made to be easy for anyone to use.

 
Fast Boot needs to be off for any hardware or firmware change .... otherwise the MB Boots without poling for hardware changes .. sometimes a Ture cold boot is still Best ... unplug anything connected to the outlet, wait for system to fully discharge capacitors, holding start button 15 seconds after unplugging PC can speed dischage of capacitors
 
Windows Fast boot also needs to be disabled & it is preferable to have previously - shut down Windows - "holding the Shift Key Down, until the PC is fully Off" ..... otherwise Windows restarts with the "saved profile file" - "of the last previous startup" & any updates to driver or hardware are usually missed.   If you did a normal shutdown, without holding shift down ... just use the "restart" of Windows twice ... then it usually sees the changes

ty_ger07

HeavyHemi
My GPU has no problem pulling slightly over 450 watts on Furmark.

If you had one which never pulled over 400 watts no matter what you did, would you say that it is performing as expected? Would you say that the two cards are equal?

HeavyHemi
Seems perfectly logical considering all the phases on the GPU are not exactly the same.

That's what power balancing is normally designed to overcome. You see that there are more current shunts than inputs. Why do you suppose that is? Some of those shunts are for the circuitry which transfers load across from one input to another. Why aren't they able to balance? Why do some cards balance better than others? Why do power limited cards tend to balance worse than other non-power-limited cards? Is it a coincidence? Why does EVGA say that a 3090 with the same symptoms has a defect, but if it is a 3080, there is nothing to care about?

If I had one, which I don't. However, I noticed someone HERE, the OP HERE, had set an ABSURD +1600 on the memory. Lets stick with specifics here, instead of your rambling what ifs about OTHER GPUs. That would be swell. 
Your attempt to explain how this all works to me, is kinda hilarious, and you know why.  Yes indeed, IF AN INDIVIDUAL GPU has an issue boosting correctly, of course that is an issue.
So HERE, WITH THIS ISSUE, I'm simply requesting the OP reduce the ABSURD +1600 memory OC and then see if it works normally. I don't understand why you're so intent on arguing this.  Most here know that overclocking your memory too high, REDUCES performance. But no, you'd like to RAGE about how they are apparently are ALL defective in design. One note wonders, get boring after awhile.

HeavyHemi

ty_ger07

HeavyHemi
My GPU has no problem pulling slightly over 450 watts on Furmark.

If you had one which never pulled over 400 watts no matter what you did, would you say that it is performing as expected? Would you say that the two cards are equal?

HeavyHemi
Seems perfectly logical considering all the phases on the GPU are not exactly the same.

That's what power balancing is normally designed to overcome. You see that there are more current shunts than inputs. Why do you suppose that is? Some of those shunts are for the circuitry which transfers load across from one input to another. Why aren't they able to balance? Why do some cards balance better than others? Why do power limited cards tend to balance worse than other non-power-limited cards? Is it a coincidence? Why does EVGA say that a 3090 with the same symptoms has a defect, but if it is a 3080, there is nothing to care about?

If I had one, which I don't. However, I noticed someone HERE, the OP HERE, had set an ABSURD +1600 on the memory. Lets stick with specifics here, instead of your rambling what ifs about OTHER GPUs. That would be swell. 

They already replied to you that +0 on the mem made no difference. So, we are back to hypotheticals about when you would finally be satisfied with saying that something is not quite right with their card.

I never said that they are all defective. I have always maintained that the problem is random where some are defective and some are not. I maintain that there is something wrong with the design to cause this condition to so easily randomly exist, for so many people, but that doesn't mean that I believe that they are all bad. The thing I rage about is all of the hours of discussion over the past year where no one wants to accept that a clearly defective card is defective, half a dozen people say stuff like "mine is working fine, so it must be something you are doing wrong", and EVGA says "don't worry about it". The gaslighting is off the charts. Have you EVER seen me talk about bad cards in someone's good card thread? No. When I talk about bad cards, I talk about specific bad cards in that owner's specific thread.

I said that I would never buy one, but if you buy one and it treats you well, good for you. I have always maintained that some are good and some are bad, and I hope that you will start giving people the space to come to terms with their card being one of the bad ones, instead of giving them a 4 week assignment to try to find what they are doing wrong.

So, is it a Bad thing ... to reach your GPU's Max MHz before your card Hits it's Max Wattage ?
 
--> Worded differently:  If OP hit Max wattage & only 2000 MHz max. would they be happier ?
 
 
Does the OP's Voltage cure or Temps contribute any to this limit?
 
 
Memory Errors/ Correction does hurt performance - well before crashes happen ... I have experienced it first hand on my own EVGA 3080 Ti FTW3 Ultra cards.
 
 
After all we are talking about Gaming & not world record Sub-zero benchmark runs .... so, 15 MHz = how many FPS ...  1 or 2 maybe, if that?
 
EVGA clearly states OC is not guaranteed --- therefore --- Any MHz over advertised = a Bonus, not a guarantee & Watts are not mentioned as guaranteed

 
Silicon lottery effect ?  maybe that is why this is not a "golden card" ?
 
Nvidia Boost Technology:  Overrides all Functions of OC.  Measures MHz, Load, temps & Volts then Nvidia Boost does what it does.
 
From a purely theoretical & deep dive into how & why OCs work best, sure it is an interesting phenomenon. 
 
Maybe EVGA can tweak the BIOS & squeeze more out of these cards ? 
 
I believe the performance gain in gaming - will be irrelevant in FPS, just in the margin of error range
 
 

Cool GTX
So, is it a Bad thing ... to reach your GPU's Max MHz before your card Hits it's Max Wattage ?

Definitely not. The only time it is a bad thing is when it's "max" wattage is way below what it should be AND the card is power limited by that wattage. If its "Max MHz" is only the max because it is capped by an erroneous power limit, then that is a problem. Max MHz due to heat, silicon quality, voltage, or safe long-term operating range is what you want. Max MHz due to the card power-limiting way below its supposed actual limit is what you don't want.
In this instance, it seems that the OP has s pretty sweet card, even though it is funky.

Memory Errors/ Correction does hurt performance - well before crashes happen ... I have experienced it first hand on my own EVGA 3080 Ti FTW3 Ultra cards.

They already addressed that.

EVGA clearly states OC is not guaranteed --- therefore --- Any MHz over advertised = a Bonus, not a guarantee & Watts are not mentioned as guaranteed

Anything that is erroneous due to the hardware is a defect covered by warranty. EVGA clearly stated that the symptoms of unbalanced input and power limit below expected power limit, for the 3090, is a defect covered by the warranty. Other models with the same symptoms should also be considered affected.

For what its worth I ran some tests....hadnt done this in quite some time, pretty much exactly as I had expected and pretty mcu the same as what Im found on previous 3080 (Non LHR) and previous 3080Ti.
 
3080Ti Stock vbios OC +120 / +675
12900K 5.1P / 4.1E
 
PR 14.6K
429w
#1 148w / #2 152w / #3 75w / Slot 53w
 
TS 21.6K
440w
#1 149w / #2 154w / #3 82w / Slot 55w
 
Superposition 4K Optimized - 18.5K
416w
 
Kombustor (no surprise here)
452w
#1 159w / #2 156w / #3 78w / Slot 58w
 
I would post screenshots but haven't figured out how to do it with Imgr..... ?

So, after telling me to flash a non-LHR BIOS, then me bringing that up, then stating they would go check with the BIOS team, again, customer service just ghosted me.
 
Seems this is indeed a hardware fault. Just surprised EVGA haven't fixed it properly for newly manufactured cards...
 
Given I'm using a EK Vector waterblock (I believe it's revised model) and all the issues plaguing the 3080 FTW3 and red lights with this block, I really don't think I want to risk the RMA/refurb route. If the card died or something, sure, but right now I'd be at the mercy of not receiving a dodgy refurb... And it's like £30 a pop for me to send a card back to EVGA. I can just imagine me sending this new card back, getting a refurb and then it too has power balancing issues...
 
What's annoying me most is the total lack of communication from EVGA. Like they decided the best thing to do was a communication lockdown. Even Jacob last year saying he'd get a response and going quiet. That's not good customer service, just a brand trying to sweep something under the rug because they know a large number of buyers will just use stock BIOS.
 
I'm grateful for EVGAs step up program, as I am for their usual great warranty, but this leaves a bitter taste. Especially when other brands 3080s work fine.

I just got a 3080 12gb ftw3 ultra and running stock settings with a 112% power limit it easily drew 443 watts running the uniengine superpositional benchmark.
 

 
Also any benefit to using the evga precision app for gpu tweaking instead of afterburner?
 
edit: Answered my own question since its the only way to control the rgb. Welp off to install precision x1. 

nonnynono
I just got a 3080 12gb ftw3 ultra and running stock settings with a 112% power limit it easily drew 443 watts running the uniengine superpositional benchmark.
 

 
Also any benefit to using the evga precision app for gpu tweaking instead of afterburner?
 
edit: Answered my own question since its the only way to control the rgb. Welp off to install precision x1. 

12GB models seem to be using the same power balancing controller as the 3080Ti model. IIRC. This means the issue is fixed. Or, it should be more reliable that your 12GB 3080 model is fine.
 
It's the 10GB 3080's that seem to be a total crapshoot. EVGA ghosting customers just about sums it up. Chances are it would require a sizeable recall to fix it, so they'd rather only those who go OCing get upset and maybe if you're lucky you'll get an RMA. But then I wouldn't be surprised if half the refurbs are all the same.
 
Even Jacob going quiet on it, who is normally happy to help the community, makes it seem likes its a PR decision/decision from up top. I mean, it's great it's not exploding cards or something, but this is still poor from EVGA given they know the FTW3 cards are what people who OC buy in order to, you know, not be limited running power draw to stock... Could save a lot of money buying another 2 pin brand of 3080 if you're happy with stock.

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.