ty_ger07
It's funny that without evidence, you resort to personal attacks. What is your function in this discussion?

It's funny, you don't even own one of these cards, you're not on the EVGA Engineering team that designed the cards, yet you seem to know how they work.

So nobody wants to Furmark one of these cards and check average power?  Proving they can use 450w?

kram36

ty_ger07
It's funny that without evidence, you resort to personal attacks. What is your function in this discussion?

It's funny, you don't even own one of these cards, you're not on the EVGA Engineering team that designed the cards, yet you seem to know how they work.

Is it not obvious?

FunktasticLucky

ty_ger07
It's funny that without evidence, you resort to personal attacks. What is your function in this discussion?

You still didn't answer my question. How is this hurting performance?

I have, numerous times.

First of all, as before, SOME will have the problem more than others. For some cards, it will not hurt performance at all. For others, it will hurt performance a lot more.

How does it hurt performance? When the card hits power limit, the clocks drop. You can't overcome the drop via overclocking. It is a hard power limit. The card, in the screenshot provided, limited itself to 405 watts instead of 450 watts because PCI-E #2 hit power limit WAY before PCI-E #3 was even close. This dropped its power limit by 10%. That can have a substantial impact on performance.

It's the same problem that the 3080 and 3090 FTW3 cards have/had. EVGA acknowledged the problem affected some 3090 cards, but refuses to acknowledge the 3080 cards with the same symptoms.

ty_ger07

kram36

ty_ger07
It's funny that without evidence, you resort to personal attacks. What is your function in this discussion?

It's funny, you don't even own one of these cards, you're not on the EVGA Engineering team that designed the cards, yet you seem to know how they work.

Is it not obvious?

FunktasticLucky

ty_ger07
It's funny that without evidence, you resort to personal attacks. What is your function in this discussion?

You still didn't answer my question. How is this hurting performance?

I have, numerous times.

First of all, as before, SOME will have the problem more than others. For some cards, it will not hurt performance at all. For others, it will hurt performance a lot more.

How does it hurt performance? When the card hits power limit, the clocks drop. You can't overcome the drop via overclocking. It is a hard power limit. The card, in the screenshot provided, limited itself to 405 watts instead of 450 watts because PCI-E #2 hit power limit WAY before PCI-E #3 was even close. This dropped its power limit by 10%. That can have a substantial impact on performance.

It's the same problem that the 3080 and 3090 FTW3 cards have/had. EVGA acknowledged the problem affected some 3090 cards, but refuses to acknowledge the 3080 cards with the same symptoms.

Define a substantial amount of performance. The most you might get from 45w is 30 to 45 mhz. That's only a 2 percent increase in clock. And that's assuming the card doesn't run enough hotter that it drops a bin (15 mhz).

I'm not particularly happy that my card limits early on some games/benchmarks, but the difference is small.

kevinc313
So nobody wants to Furmark one of these cards and check average power?  Proving they can use 450w?

I ran Kombustor and posted that back before all the bickering started.

FunktasticLucky

talon951

kevinc313
So nobody wants to Furmark one of these cards and check average power?  Proving they can use 450w?

I ran Kombustor and posted that back before all the bickering started.

I don't understand the need to potentially burn up a gpu.

For gaming I settled on .850V @ 2050Mhz clock and 10500Mhz Mem. 24/7 stable all day every day.

You cant "burn up" the GPU....some are simply questioning the power balancing in relation to there GPU's NOT being able to reach the power draw the Cards are advertised as achieving....in other words if a BIOS update (XOC Bios) of 450W is flashed, why is the Card still "stuck" at 400W (take thermal throttling out of the equation if under full custom loop).
 
Many of us arent just about what OC's we can achieve "for gaming", but seeking those high benchmark runs, this is inhibited by "power limitations".....

Question: Is the GPU able to keep it's core boost speed 100% constant during stress test??? or is it unstable? If the core boost clock is 100% constant during stress test, there is no issue. To me at least

tyranus7
Question: Is the GPU able to keep it's core boost speed 100% constant during stress test??? or is it unstable? If the core boost clock is 100% constant during stress test, there is no issue. To me at least

I have never seen any video card keep it's core clock stable during a stress test (even under H20 cooling), but I have never done LN2, so.....

FunktasticLucky
Bud did you not read the whole thread? I set the record with a FTW3 Hybrid with no mods or flashes bios. It's a none issue. People are basing this assumption off a 3rd pin power draw but they are forgetting that the card doesn't receive all it's power from the 8 pin connectors. It can receive up to 75W from the pcie slot. 150-75 is you guessed it... 75W. So 150(8 pin)+150(8 pin)+75(PCIe Slot)=375W. Leaving another 75W for the 3rd 8pin. The 3rd pin is being supplemented by the slot which usually hovers around 50-60W.

1) This isn't about you.
2) Not all cards have this problem. Same story for the previous 8 months.



3) The PCI-E slot doesn't provide 75 watts of 12 volt power.
4) Your math only works if the card power balances properly. If your card is one of the cards which doesn't power balance properly, you get the result above.


Is 5% or 10% performance impact substantial in your opinion? It can't be defined. It depends on each person. If I paid $1400 for a video card, I wouldn't be able to accept it. It's especially embarrassing that EVGA didn't fix this same problem when it released this new product line.

kram36

tyranus7
Question: Is the GPU able to keep it's core boost speed 100% constant during stress test??? or is it unstable? If the core boost clock is 100% constant during stress test, there is no issue. To me at least

I have never seen any video card keep it's core clock stable during a stress test (even under H20 cooling), but I have never done LN2, so.....

Mine is. Rock stable. Water cooled. The clocks don't budge an inch.
But, I also power-modded my card, and that's why. That's why I especially think that a power limited card is unfortunate; especially one which didn't need to be born that way.

kram36

tyranus7
Question: Is the GPU able to keep it's core boost speed 100% constant during stress test??? or is it unstable? If the core boost clock is 100% constant during stress test, there is no issue. To me at least

I have never seen any video card keep it's core clock stable during a stress test (even under H20 cooling), but I have never done LN2, so.....

Check this, on air cooling 100% stable clock boost speed: https://youtu.be/aqPBk1K5p3A?t=943
 

ty_ger07

rjbarker

kevinc313

talon951

kevinc313
So nobody wants to Furmark one of these cards and check average power?  Proving they can use 450w?

I ran Kombustor and posted that back before all the bickering started.

Thanks man, looking for the HWiNFO average reading.  From what I've seen so far these cards pull 40-50w less than my 3080.  Prove me wrong.

kevin...what does your Card draw total ? I'm assuming close to 450W with 450W XOC Bios ?

Correct. He has said that numerous times in numerous threads. He admits that his card has no problem, but fails to believe that other people do have a problem or should feel bad when they have the problem. Rose colored glasses.

All I want to know is if the latest batch of cards is partly bad, mostly bad, or all bad on the power balancing, if they ACTUALLY hit power limit and if there is potential for a 500w bios.

kram36

tyranus7

kram36

tyranus7
Question: Is the GPU able to keep it's core boost speed 100% constant during stress test??? or is it unstable? If the core boost clock is 100% constant during stress test, there is no issue. To me at least

I have never seen any video card keep it's core clock stable during a stress test (even under H20 cooling), but I have never done LN2, so.....

Check this, on air cooling 100% stable clock boost speed: https://youtu.be/aqPBk1K5p3A?t=943
 

That's a shunt power modded card. I'm talking a factory non modded card.

yeah. In fact if a video card is not reaching thermal limit the only reason for the core boost speed to fluctuate is because the video card is constantly hitting power limit. Which IMO shouldn't be. It should be more like an Intel CPU which holds the boost clock if enough power and cool are being supplied

kram36

tyranus7

kram36

tyranus7

kram36

tyranus7
Question: Is the GPU able to keep it's core boost speed 100% constant during stress test??? or is it unstable? If the core boost clock is 100% constant during stress test, there is no issue. To me at least

I have never seen any video card keep it's core clock stable during a stress test (even under H20 cooling), but I have never done LN2, so.....

Check this, on air cooling 100% stable clock boost speed: https://youtu.be/aqPBk1K5p3A?t=943
 

That's a shunt power modded card. I'm talking a factory non modded card.

yeah. In fact if a video card is not reaching thermal limit the only reason for the core boost speed to fluctuate is because the video card is constantly hitting power limit. Which IMO shouldn't be. It should be more like an Intel CPU which holds the boost clock if enough power and cool are being supplied

He also locked his GPU core clock to 2,000MHz, so yeah the card isn't going to bounce around when you lock it that low.
 
https://youtu.be/aqPBk1K5p3A?t=955
 
Actually if you look at his GPU-Z, the GPU clock is bouncing around.

Yea the trick is to lock it at a voltage point that never exceeds the power limit (whatever that is). I can run my 3090 on the KP XOC bios at the 1093mv limit and it doesn't bounce, but that can exceed 500w easily. But I could just as easily limit the VF curve to 950mv or a little more and my 3080ti wouldn't bounce either.

ty_ger07

FunktasticLucky
Bud did you not read the whole thread? I set the record with a FTW3 Hybrid with no mods or flashes bios. It's a none issue. People are basing this assumption off a 3rd pin power draw but they are forgetting that the card doesn't receive all it's power from the 8 pin connectors. It can receive up to 75W from the pcie slot. 150-75 is you guessed it... 75W. So 150(8 pin)+150(8 pin)+75(PCIe Slot)=375W. Leaving another 75W for the 3rd 8pin. The 3rd pin is being supplemented by the slot which usually hovers around 50-60W.

1) This isn't about you.
2) Not all cards have this problem. Same story for the previous 8 months.



3) The PCI-E slot doesn't provide 75 watts of 12 volt power.
4) Your math only works if the card power balances properly. If your card is one of the cards which doesn't power balance properly, you get the result above.


Is 5% or 10% performance impact substantial in your opinion? It can't be defined. It depends on each person. If I paid $1400 for a video card, I wouldn't be able to accept it. It's especially embarrassing that EVGA didn't fix this same problem when it released this new product line.

Let's see core clocks, and voltages from these settings? What are your afterburner/precision settings? If you're slamming all sliders to the right you're doing it WAY wrong.