EVGA RTX 3090 FTW3U Power Balancing response from ECGA_JacobF

lobstar

Limp Gawd
Joined
Nov 24, 2004
Messages
371
Looks like they are finally acknowledging the issues many people experience with much lower than anticipated power draw on these cards. Specifically model 24G-P5-3987-KR.

The specific post on the EVGA forums can be found here in this thread.
 
Interesting.... I have a P5 version (which looks to be what Jacob is mentioning in his corrected post)... but I have no issues and never have using the stock vBios. Wondering if I should just be happy mine works, or if I should email and be proactive about it?
 
Interesting.... I have a P5 version (which looks to be what Jacob is mentioning in his corrected post)... but I have no issues and never have using the stock vBios. Wondering if I should just be happy mine works, or if I should email and be proactive about it?
Define 'works'. My cards have functioned fine in games. Are you saying you are able to reach 450w usage on the OC bios and 500w on the XOC? Both of my RMA'd cards have worked fine but are 1) power limited at basically all clocks and 2) drawing in excess of 75w on the PCIe connector which is out of spec for PCIe power limits. If your card consistently draws around 66w on the PCIe then you probably have a functioning card. If you notice it going above 80w, you likely have a card affected by this issue. I'm worried about the long term effects of drawing 80-90w over a part designed for 75w.
 
Define 'works'. My cards have functioned fine in games. Are you saying you are able to reach 450w usage on the OC bios and 500w on the XOC? Both of my RMA'd cards have worked fine but are 1) power limited at basically all clocks and 2) drawing in excess of 75w on the PCIe connector which is out of spec for PCIe power limits. If your card consistently draws around 66w on the PCIe then you probably have a functioning card. If you notice it going above 80w, you likely have a card affected by this issue. I'm worried about the long term effects of drawing 80-90w over a part designed for 75w.
I use the OC vbios with full slider at 107% (450W I believe). I get 449~450W on the Power Draw max and about 68~74W on the PCIe slot max. It's the same across pretty much any game or benchmark for me if it loads the card enough. I don't think I have ever logged my PCIe above 75~76W in anything.
 
Very interesting. The first batch of 3090 FTW3 HydroCoppers are supposed to go out tomorrow to those already in queue (which I'm a part of) so I'm interested to see how my 3090 FTW3 will fair.

Doesn't look like there is a custom BIOS being made. From what I've gathered in that thread it looks like EVGA is asking people to RMA their cards (for free) in which they'll return a 3090 FTW3 with the XOC BIOS already installed. I'm guessing they are testing these 3090 FTW3s to see they aren't being affected by the power balancing issues and sending them back as the RMA replacement.
 
Last edited:
I'm guessing they are testing these 3090 FTW3s to see they aren't being affected by the power balancing issues and sending them back as the RMA replacement.
I agree that's probably the easiest way out of this. I could see a scenario also where they don't want to release a fixed bios into the wild that would adversely affect or even possibly damage cards functioning properly like Sly's.
 
Looking at the EVGA thread, it sounds to me like they got a bunch of faulty chips that control the power circuitry and have no way to tell which ones have issues without manually testing them.

And the messed up ones can have performance issues as well.

If it was only a vBIOS issue, then all cards would have the issue and the updated BIOS would fix all cards. This is not the case apparently.

HRRRMMMM.
 
I'm half tempted to try the 500W vbios as I have temperature headroom, but at the same time, I got a good thing going and don't want to risk screwing it up for a minor bump in benchmarks! 😆
 
Looking at the EVGA thread, it sounds to me like they got a bunch of faulty chips that control the power circuitry and have no way to tell which ones have issues without manually testing them.

And the messed up ones can have performance issues as well.

If it was only a vBIOS issue, then all cards would have the issue and the updated BIOS would fix all cards. This is not the case apparently.

HRRRMMMM.
Again, I agree with the sentiment but it's REALLY strange to me that the XC3 bios seems to correct the issue on some cards.
 
I agree that's probably the easiest way out of this. I could see a scenario also where they don't want to release a fixed bios into the wild that would adversely affect or even possibly damage cards functioning properly like Sly's.
What I also wonder is (from reading the evga forums); does it also matter where the P5 version was manufactured? Mine was China, but it looks like there are more power issues/failures with the Taiwan versions... maybe there was a bad batch of chips in the mix causing the issues and a specific factory got more of that bad batch? It's anyone's guess, but at least evga is working on it... I know people are upset (I would be too after it took so long to get the damn card to begin with). But evga, in my experience, generally does what it can to make things right if you have issues. They honestly may be just as much in the dark as us and are trying to figure it out without causing a huge fuss about it until they know what is actually wrong.
 
What I also wonder is (from reading the evga forums); does it also matter where the P5 version was manufactured? Mine was China, but it looks like there are more power issues/failures with the Taiwan versions... maybe there was a bad batch of chips in the mix causing the issues and a specific factory got more of that bad batch? It's anyone's guess, but at least evga is working on it... I know people are upset (I would be too after it took so long to get the damn card to begin with). But evga, in my experience, generally does what it can to make things right if you have issues. They honestly may be just as much in the dark as us and are trying to figure it out without causing a huge fuss about it until they know what is actually wrong.
I think it would be very unlikely after 6 months that they don't know the cause of both the power balancing and the red light of death issues. The voltage controller appears to be the culprit for both issues based on community investigation. If it was a simple fix I think they would've released it by now. I'm guessing the fix will be switching from an analog voltage controller to a digital one that can redistribute power draw between the 8-pin connectors and PCI-E slot. As I've mentioned previously in different threads; I wouldn't be surprised if we will see a PCB revision soon.
 
I'm lost, I thought the complaints were, we're supposed to be able to pull 500 Watts with FTW3 Ultra's. And even with the beta OC bios they weren't? thus reducing the OC capability.

just regular OC BIOS, not the beta OC one posted in that evga thread.
RegularXOC.PNG

I'm half tempted to try the 500W vbios as I have temperature headroom, but at the same time, I got a good thing going and don't want to risk screwing it up for a minor bump in benchmarks! 😆
 
I'm lost, I thought the complaints were, we're supposed to be able to pull 500 Watts with FTW3 Ultra's. And even with the beta OC bios they weren't? thus reducing the OC capability.

just regular OC BIOS, not the beta OC one posted in that evga thread. View attachment 337090
Nothing to be confused about. This is a hardware issue that affects some cards and not others. Likely due to the analog voltage controller that cannot adjust power draw distribution based on components (not all components are made equally). Cards with digital voltage controllers do not have this issue.
 
I'm lost, I thought the complaints were, we're supposed to be able to pull 500 Watts with FTW3 Ultra's. And even with the beta OC bios they weren't? thus reducing the OC capability.

just regular OC BIOS, not the beta OC one posted in that evga thread. View attachment 337090
False; the 3090 FTW3 Ultra "base bios"; on either the normal or OC Bios only allows 107% Power Slider; which is 450W maximum. Power Slider at 100% is 420W maximum.

The "500W Bios" is for overclockers with more headroom; but at the same time, I think evga was trying to see if that would fix the issues being reported. The start of the issue was people unable to even pull close to 450W on the stock bios and maxing out the PCIe power before the 8-Pin even kicked in properly. Was 100% a balancing issue, no matter the power limit, from my understanding anyway...
 
I've had my request approved already and they will contact me when I am to send in my card. I'm surprised how quick they got back but I did email within minutes of Jacob's post.
 
Back
Top