talon951
They could definitely fix it by increasing the limits in the bios but it would also drive the 8pin power further out of spec. Maybe that's why they have not done anything too. But that's just speculation on my part.

For example the KP 1kw bios will raise the power limit on pretty much any 3090 whether it is 2x8pin or 3x8pin due to the limits being set sky high. Extreme example but shows what can be done.

What a layman like myself finds hard to understand here though is why a small number of 3080's seem to go up to 450w with this BIOS, but most, going by forums/Reddit complaints, struggle to get over 400w, or splutter up to maybe 420~430w.
 
EVGA released the BIOS so they must have at least, in theory, been happy to try and let these cards draw up to 450w.
 
A lot of people have mentioned power balancing, but I do have to say I don't really know what that is/what handles it. As in, what part of the PCB is the brains behind that. Something at the pins themselves? Is it purely hardware or is software involved?
 
With zero communication from EVGA people like me aren't knowledgeable enough to really get it on our own. So I'm left looking at an MSI BIOS that appears to draw up to 500w on my EVGA card, or the EVGA BIOS which hardwalls at 400w despite EVGA telling me I should be able to see up to 450w.
 
Obviously the safest advice for me is live with it, don't run an MSI BIOS if you aren't happy to take a risk. Which is fair enough, but I'd really have liked EVGA or Jacob to have actually responded about this, not ignore it. They, as in EVGA, released a 450w BIOS for this card, so for anyone to call us, the end users, uppity/unreasonable, for querying this, is unfounded.
 
If they hadn't released this 450w BIOS that would have been an answer in of itself, but they did, so they should at least respond to why some/most of these cards cannot handle it. Or, now as we've found out with this MSI BIOS, "cannot" handle it. While my 1 card is a small sample size, I've not seen a single problem across any games from crashing or anything with the likes of OCP/restarts or shutdowns.
 
Like, customer support actually ghosted me after saying they were away to talk to the BIOS team. That's the first and only time EVGA customer service has ever done that to me. Just adds to the feeling this is an enforced "black out" from up top or something. Which seems way overboard

talon951

carneb
 
After reading your posts about the MSI bios I decided to try it on my 3080 12GB FTW3 card. With the normal bios it will draw 450W under heavy load such as furmark but under normal loads only around 420W. With the MSI Suprim X bios I measured the current on each 8 pin supply with a clamp meter while running furmark and found that it was drawing 163W, 225W, and 83W from input 1,2, and 3 and software reported 63W from the PCIe slot for a total of 534W! As a comparison with the normal EVGA bios I read 135W, 188W, 84W from inputs 1,2,3 and 47W from PCIe slot for a total of 454W. The card will also draw more power under normal use with the MSI bios pulling around 480W instead of 420 in benchmarks.
 

What does GPUZ/HWINFO show for power readings on the EVGA bios?  Curious how the monitoring app readings compare to the clamp meter readings.

When running furmark with the EVGA bios GPU-Z was reporting 156, 152, and 94W. It seems No. 2 draws more power as found by a 3080 Ti owner on another forum, and 1 and 3 draw less power than the actual measured current draw. It is the 12GB bios I flashed I'm not sure the 10GB bios would work.

talon951

carneb

When running furmark with the EVGA bios GPU-Z was reporting 156, 152, and 94W. It seems No. 2 draws more power as found by a 3080 Ti owner on another forum, and 1 and 3 draw less power than the actual measured current draw. It is the 12GB bios I flashed I'm not sure the 10GB bios would work.

Yea that was me (3080ti clamp meter measurements). Interesting that the difference in 8pin #2 is similar (about 30w). But I found #1 and #3 to be fairly accurate on my 3080ti though.

But yea that 3080 really hammers #2 with the MSI bios. Flirting with the 20 amp fuse limit.

Yes I agree, it was pulling 18.8 amps so pretty close to 20. I was at one stage thinking about shunt modding, but I'm glad I didn't just blindly do it without doing the measurements first or pin 2 would have probably gone over the 20 amp fuse limit.

Intoxicus

Audioboxer

talon951
They could definitely fix it by increasing the limits in the bios but it would also drive the 8pin power further out of spec. Maybe that's why they have not done anything too. But that's just speculation on my part.

For example the KP 1kw bios will raise the power limit on pretty much any 3090 whether it is 2x8pin or 3x8pin due to the limits being set sky high. Extreme example but shows what can be done.

What a layman like myself finds hard to understand here though is why a small number of 3080's seem to go up to 450w with this BIOS, but most, going by forums/Reddit complaints, struggle to get over 400w, or splutter up to maybe 420~430w.
 
EVGA released the BIOS so they must have at least, in theory, been happy to try and let these cards draw up to 450w.
 
A lot of people have mentioned power balancing, but I do have to say I don't really know what that is/what handles it. As in, what part of the PCB is the brains behind that. Something at the pins themselves? Is it purely hardware or is software involved?
 
With zero communication from EVGA people like me aren't knowledgeable enough to really get it on our own. So I'm left looking at an MSI BIOS that appears to draw up to 500w on my EVGA card, or the EVGA BIOS which hardwalls at 400w despite EVGA telling me I should be able to see up to 450w.
 
Obviously the safest advice for me is live with it, don't run an MSI BIOS if you aren't happy to take a risk. Which is fair enough, but I'd really have liked EVGA or Jacob to have actually responded about this, not ignore it. They, as in EVGA, released a 450w BIOS for this card, so for anyone to call us, the end users, uppity/unreasonable, for querying this, is unfounded.
 
If they hadn't released this 450w BIOS that would have been an answer in of itself, but they did, so they should at least respond to why some/most of these cards cannot handle it. Or, now as we've found out with this MSI BIOS, "cannot" handle it. While my 1 card is a small sample size, I've not seen a single problem across any games from crashing or anything with the likes of OCP/restarts or shutdowns.
 
Like, customer support actually ghosted me after saying they were away to talk to the BIOS team. That's the first and only time EVGA customer service has ever done that to me. Just adds to the feeling this is an enforced "black out" from up top or something. Which seems way overboard

Humor me and post a picture of how your power cables are plugged into the GPU.

?
 
The only way they can be https://imgur.com/a/z0p8FRu
 
Pics of the build from last month. Only changes since then have been minor.

Audioboxer

Intoxicus

Audioboxer

talon951
They could definitely fix it by increasing the limits in the bios but it would also drive the 8pin power further out of spec. Maybe that's why they have not done anything too. But that's just speculation on my part.

For example the KP 1kw bios will raise the power limit on pretty much any 3090 whether it is 2x8pin or 3x8pin due to the limits being set sky high. Extreme example but shows what can be done.

What a layman like myself finds hard to understand here though is why a small number of 3080's seem to go up to 450w with this BIOS, but most, going by forums/Reddit complaints, struggle to get over 400w, or splutter up to maybe 420~430w.
 
EVGA released the BIOS so they must have at least, in theory, been happy to try and let these cards draw up to 450w.
 
A lot of people have mentioned power balancing, but I do have to say I don't really know what that is/what handles it. As in, what part of the PCB is the brains behind that. Something at the pins themselves? Is it purely hardware or is software involved?
 
With zero communication from EVGA people like me aren't knowledgeable enough to really get it on our own. So I'm left looking at an MSI BIOS that appears to draw up to 500w on my EVGA card, or the EVGA BIOS which hardwalls at 400w despite EVGA telling me I should be able to see up to 450w.
 
Obviously the safest advice for me is live with it, don't run an MSI BIOS if you aren't happy to take a risk. Which is fair enough, but I'd really have liked EVGA or Jacob to have actually responded about this, not ignore it. They, as in EVGA, released a 450w BIOS for this card, so for anyone to call us, the end users, uppity/unreasonable, for querying this, is unfounded.
 
If they hadn't released this 450w BIOS that would have been an answer in of itself, but they did, so they should at least respond to why some/most of these cards cannot handle it. Or, now as we've found out with this MSI BIOS, "cannot" handle it. While my 1 card is a small sample size, I've not seen a single problem across any games from crashing or anything with the likes of OCP/restarts or shutdowns.
 
Like, customer support actually ghosted me after saying they were away to talk to the BIOS team. That's the first and only time EVGA customer service has ever done that to me. Just adds to the feeling this is an enforced "black out" from up top or something. Which seems way overboard

Humor me and post a picture of how your power cables are plugged into the GPU.

?
 
The only way they can be https://imgur.com/a/z0p8FRu
 
Pics of the build from last month. Only changes since then have been minor.

Just noticed that it appears every single fan you have on your loops is intake.

Audioboxer

HeavyHemi

Audioboxer

Intoxicus

Audioboxer

talon951
They could definitely fix it by increasing the limits in the bios but it would also drive the 8pin power further out of spec. Maybe that's why they have not done anything too. But that's just speculation on my part.

For example the KP 1kw bios will raise the power limit on pretty much any 3090 whether it is 2x8pin or 3x8pin due to the limits being set sky high. Extreme example but shows what can be done.

What a layman like myself finds hard to understand here though is why a small number of 3080's seem to go up to 450w with this BIOS, but most, going by forums/Reddit complaints, struggle to get over 400w, or splutter up to maybe 420~430w.
 
EVGA released the BIOS so they must have at least, in theory, been happy to try and let these cards draw up to 450w.
 
A lot of people have mentioned power balancing, but I do have to say I don't really know what that is/what handles it. As in, what part of the PCB is the brains behind that. Something at the pins themselves? Is it purely hardware or is software involved?
 
With zero communication from EVGA people like me aren't knowledgeable enough to really get it on our own. So I'm left looking at an MSI BIOS that appears to draw up to 500w on my EVGA card, or the EVGA BIOS which hardwalls at 400w despite EVGA telling me I should be able to see up to 450w.
 
Obviously the safest advice for me is live with it, don't run an MSI BIOS if you aren't happy to take a risk. Which is fair enough, but I'd really have liked EVGA or Jacob to have actually responded about this, not ignore it. They, as in EVGA, released a 450w BIOS for this card, so for anyone to call us, the end users, uppity/unreasonable, for querying this, is unfounded.
 
If they hadn't released this 450w BIOS that would have been an answer in of itself, but they did, so they should at least respond to why some/most of these cards cannot handle it. Or, now as we've found out with this MSI BIOS, "cannot" handle it. While my 1 card is a small sample size, I've not seen a single problem across any games from crashing or anything with the likes of OCP/restarts or shutdowns.
 
Like, customer support actually ghosted me after saying they were away to talk to the BIOS team. That's the first and only time EVGA customer service has ever done that to me. Just adds to the feeling this is an enforced "black out" from up top or something. Which seems way overboard

Humor me and post a picture of how your power cables are plugged into the GPU.

?
 
The only way they can be https://imgur.com/a/z0p8FRu
 
Pics of the build from last month. Only changes since then have been minor.

Just noticed that it appears every single fan you have on your loops is intake.

It is, because it's for 4 rads.
 
Exhaust is at the front because I'm using a Lian Li custom front. If you thought the front fans were also intake, nah they're exhaust, don't worry.
 
Works great for temps on the basis of every rad has the coolest ambient air coming across them. Seeing as all my major components are watercooled, even memory, internal case isn't an issue.
 
And believe it or not, internal case temp isn't an issue, full stop. I monitor the actual temp inside the case with a sensor, let alone obviously look at VRM temps.

I preach the same thing all the time. Been doing the same for about a decade.
 
A rad with intake air is a happy rad.  It won't turn into a furnace in there as some people believe like a hot blow dryer right before the fuse trips.

I researched those filters at one point. They do work well to filter out dirt, but hurt airflow significantly I think. If your priority is keeping out dust then they are a good choice. If you're more interested in the best temps then they are probably not the way to go.

Almost any filter that has a finer mesh and hence traps more dirt will be more restrictive. It's a trade off for performance vs dust filtering. You could also bump your fan speed up if you don't mind a little more noise to get the same temps.

talon951
Almost any filter that has a finer mesh and hence traps more dirt will be more restrictive. It's a trade off for performance vs dust filtering. You could also bump your fan speed up if you don't mind a little more noise to get the same temps.

That would make sense. Instead of buying the whole kit, I might just buy the pieces needed to do the back of the case.
 
I've tried running my setup without any dust filters and I just don't notice a difference. But that's maybe just because of 4 rads and using Noctua fans that can probably handle restriction pretty well.
 
My environment is not really dusty or anything, but even if you clean often you still get dust/lint/hairs floating about. 4 intakes, no matter your environment, is just going to be dust magnet for the insides/radiator fins.
 

kraade
I thought my lian li filters were bad restrictive when using on the floor with intake fans, but having like the side filters and having a gap before the filters made a huge difference in not affecting the flow. I made spacers for the floor fans then scratched that idea because the floor got a 360 SE and seems to be fine as either intake or exhaust. If I was to try an intake again, I might try a foam filter or thin carbon like on an air purifier.

I've always read having your case on the floor is like the worst thing you can do for dust lol. So yeah, on the floor I would definitely be using filters of some sorts for intakes.

Well, I am another EVGA 3080 Ti FTW3 disappointed buyer you can add to the list.  The card almost never breaks 400w even under stressful test conditions that pull other cards to that limit.  It is not "normal" behavior, even if it does sound nice (or sound true) to say that cards "don't need to stay operating at the very limit just because that is the stated limit". That might be true, but the behavior of these cards in the past is that they DO operate close to the power limit when your power limit slider is at 100%.  And when you have a bios that goes beyond 100%, the average power draw goes up!
 
Not so with these cards, and I would say it is a defect.  I even have voltage headroom (supposedly the voltage limit is 1.09v on these) at like 1.05v when hitting the 400W power limit.  So my card NEVER goes beyond 1.05 or so, even if I lock the voltage higher in Afterburner.  This is because its hitting the 400W power limited sooner than the voltage limit.
 
I wish EVGA would release a BIOS with this fixed.  But, since they are no longer in the GPU game, they probably will do nothing.  I consider it to be false advertising.  The reason I got this was because of the stated power limit of 450W lending to better overclocking headroom.  Well, based on what I've seen others go through when discussing this with EVGA, it seems like pretty terrible and apathetic customer service.

HeavyHemi

mattschlosser
Well, I am another EVGA 3080 Ti FTW3 disappointed buyer you can add to the list.  The card almost never breaks 400w even under stressful test conditions that pull other cards to that limit.  It is not "normal" behavior, even if it does sound nice (or sound true) to say that cards "don't need to stay operating at the very limit just because that is the stated limit". That might be true, but the behavior of these cards in the past is that they DO operate close to the power limit when your power limit slider is at 100%.  And when you have a bios that goes beyond 100%, the average power draw goes up!
 
Not so with these cards, and I would say it is a defect.  I even have voltage headroom (supposedly the voltage limit is 1.09v on these) at like 1.05v when hitting the 400W power limit.  So my card NEVER goes beyond 1.05 or so, even if I lock the voltage higher in Afterburner.  This is because its hitting the 400W power limited sooner than the voltage limit.
 
I wish EVGA would release a BIOS with this fixed.  But, since they are no longer in the GPU game, they probably will do nothing.  I consider it to be false advertising.  The reason I got this was because of the stated power limit of 450W lending to better overclocking headroom.  Well, based on what I've seen others go through when discussing this with EVGA, it seems like pretty terrible and apathetic customer service.

I don't have any problem at all hitting 450 watts. It's not a BIOS problem. How about some more details on the rest of your system and what you're using to load the system?
MSI Kombustor stress test should load the GPU to 450 watts using the default stress test. Make sure you have Vsync and any frame limiters disabled. You should be seeing around 140 FPS with the default settings.

 You're right, it does appear that the program using to load the GPU is the difference.  I used Superposition and 3dMark and Furmark at various settings trying to get one that would load the GPU completely (basically setting resolution and graphics quality to ultra or extreme to draw more watts).  On previous cards (not 3080 ti), Superposition and 3Mark would push the GPU to its max, no so with this 3080 Ti.  Since it seemed to be capping out at 400w, I assumed I was having the same issue that others have stated, all else seeming to be exact or equal in terms of similar symptoms.
 
But when I used OCCT to load the GPU yesterday it pushed the power to 445w or so and held that for a while, so there is no real "limit" at 400w.  Maybe its just coincidence that the majority of stress test applications, benchmarks, and real-world gaming are only pushing this card to around 400w.  But I can confirm now that my card will reach 450w and hold it, with the right test -- I've only found OCCT to do that, but I will try Kombuster as you mentioned.

ty_ger07
But in the game where it was limited to 400 watts, what was the "perfcap reason"? Was it "PWR"? If so, you can't blame vsync or computer setup or power supply or anything else other than the card itself. If vsync or framerate limiter was keeping the card from using more power, the perfcap reason wouldn't be "PWR".

Please verify the perfcap reason. It is the metric in this discussion which matters the most.

It isn't fair to use a power virus and then say that a power virus proves that the card is working as intended, when you game as intended with a perfcap reason of "PWR". The reason you bought the card was to game. It should balance and power limit well enough to reach its full advertised potential in the application intended. That is the expectation set forth.

Hemi likes to ignore the details when it comes to this matter. He loves proving that there is one situation where the card uses more power, even though that situation suits no one and the explanation ignores all other details on the matter.

I will need to look and see if there even was any perfcap reason.  I do not know.
 
But using OCCT to get the card to pull 450w does prove that its possible to hit 450w and sustain it, and that there isn't a hard power limit at 400w.  How would it allow a power virus to pull 450w but know to limit a game to 400w (presuming that game SHOULD be pulling more than 400w)?

ty_ger07

mattschlosser

ty_ger07
But in the game where it was limited to 400 watts, what was the "perfcap reason"? Was it "PWR"? If so, you can't blame vsync or computer setup or power supply or anything else other than the card itself. If vsync or framerate limiter was keeping the card from using more power, the perfcap reason wouldn't be "PWR".

Please verify the perfcap reason. It is the metric in this discussion which matters the most.

It isn't fair to use a power virus and then say that a power virus proves that the card is working as intended, when you game as intended with a perfcap reason of "PWR". The reason you bought the card was to game. It should balance and power limit well enough to reach its full advertised potential in the application intended. That is the expectation set forth.

Hemi likes to ignore the details when it comes to this matter. He loves proving that there is one situation where the card uses more power, even though that situation suits no one and the explanation ignores all other details on the matter.

I will need to look and see if there even was any perfcap reason.  I do not know.
 
But using OCCT to get the card to pull 450w does prove that its possible to hit 450w and sustain it, and that there isn't a hard power limit at 400w.  How would it allow a power virus to pull 450w but know to limit a game to 400w (presuming that game SHOULD be pulling more than 400w)?

Because the card doesn't power balance properly; if you find the right power virus which loads up the right thing, you could reach the power limit before the power balance issue plays its role. It's been proven plenty of times, but Hemi won't look at anything other than his perspective.

We know that the card can use 450 watts, but often we see that power balancing problems causes one input to limit everything way before the total expected power limit.

Hemi's approach is that if he can show that one scenario doesn't exhibit the problem, it is no problem. When I push him on the subject of intended use, disregard of the other indications, and falling below expectations, his response is 'maybe that is the way it is supposed to work'. Yeah, right. Every other video card is history didn't exhibit this behavior, the problem exists primarily along lines of product segmentation, and EVGA had a special RMA program to specifically attempt to resolve this issue with the higher-end models due to overwhelming outcry, but ignored trying to improve the lower-end models which would cut into the performance marketing of the higher-end models.

IF your card is prematurely limited in performance due to an unexpectedly low power perfcap event, seeing your card reach full expected perfomance in one situation is not an indication that everything is fine. All it does is mock you about a level of performance which you will not obtain for your normal intended application.

I ran the Superposition benchmark with GPU-Z sensors open -- you are right, it was PerfCap = pwr.  So even though it hovered around 400W the whole time, it was hitting the performance cap due to power when it should have another 50W of headroom.  So, what's the recourse? Will EVGA replace if you RMA?  Seems like you are correct and that our cards don't perform at the stated power limit according to what is expected.