Question 2080ti Power Draw at Idle

[Fluffly Fluff]

I have two Asus 2080ti Turbo cards in SLI and their GPUs are drawing 40 watts and 20 watts at idle--this is when at the desktop with nothing open--not even a browser. Albeit I am running a 60 Hz 4k monitor, but I have seen numbers showing others running a 60Hz 4K monitor plus a second standard size monitor with only one 2080ti and their power draw on the GPU for the two monitors is only 24 watts. 60 watts total for one 4k monitor is a far cry from 24 watts for one 60 Hz 4k monitor plus a second standard resolution monitor.

I also use the same monitor with another computer that has two 2080s in SLI, and the power draw of the two 2080s at idle is half. There is also data showing that typically a 2080ti GPU uses one watt less power than a 2080 GPU at idle, so what the heck?

It just seems very abnormal. I know more higher end cards draw more power under load which is accounted for by higher performance, as in high FPS and such. But typically a 2080ti at idle draws one watt less than a 2080 at idle.

Here is a quote from the Nvidia forum:

My 2080ti is idling at 18W with extended desktop configuration over 2 displays and only 8W with a single monitor.

And from the EVGA forum:

My 2080 TI FTW3 Ultra idles at 16 w also.

 

[dotas1]

Please do not create a new thread for the same issue.

[SOLVED] - 2080 vs 2080ti Idle temperatures

I have two very similar rigs with relatively comparable builds (that is, similar airflow, cases and fans, similar CPUs, identical AOI coolers, etc). One has two 2080s in SLI and the other two 2080tis in SLI. It seems very odd that the lesser of the set, the 2080s idle at around 34 degrees...

forums.tomshardware.com

In your threads you're saying about similar systems and you started with idle temperatures and then move to peak temperatures. Unless you start being very specific, we can only guess.
Post full system specs of both systems. Driver versions, BIOS version, Windows version, any overclock applied. Be specific what games are you trying to play at what resolution, with what settings, what fps you get (mean and max) and what peak temperatures you get.
That is the only way we can really help.

 

[Fluffly Fluff]

Well, I'm not sure if I've pinned down one side of the problem or not. GPU-Z and HWMonitor have differing numbers for GPU power draw. GPU-Z breaks the numbers down more specifically, hence maybe the GPU-Z power draw numbers are more in line with the numbers others have posted about their power draw. So maybe (I'm not sure) what I think is a power draw issue is not one after all. Regardless, HWMonitor is showing the power draw of the main 2080ti as around 40 watts at idle, while it is showing the main 2080 as around 20 watts at idle. This is very strange, as multiple articles show that their total power draw at idle ought to be very similar.

As for my posts being the about same issue--I apologize. But at first I thought the core of the issue was a cooling issue so posted about the temperatures. But then later I thought it was a power issue, hence the different threads. Sometimes a different angle gets people with the right knowledge/info/fix to contribute.

In a nutshell, the 2080s in SLI are doing everything that has exactly the same system demands about 5 or 6 degrees (or higher) cooler than two 2080s in SLI, from when under load to being idle from about a minute after start up. I don't understand why this is, but realize the power draw almost across the board is higher with the 2080tis. But honestly, I don't see why this is. Apparently falsely, I thought higher end cards were better in a few ways, and also more efficient under the same load.

As for specifics, all four cards are Asus Turbo versions with no overclock. The Nvidia drivers, Window 10 drivers, and Intel drivers are all the latest as available via GeForce Experience, Windows updater and Intel Driver & Support Assistant and AMD's website (for chipset drivers). I check every day at startup for and update all drivers. BIOSes are the latest for a Rampage VI Extreme Omega (with the 2080s) and a Crosshair VIII Formula (with the 2080tis) motherboards as posted on the Asus webpages. The CPUs are an Intel 9980xe and a Ryzen 3950x.

The game is World of Warcraft and the max FPS is set to max 60 on all levels with various settings basically in the middle for both systems--and exactly the same. The resolution is 4k. In one test at around exactly the same time (and ambient temperature), the 2080s reached 58 degrees max, with the card fans near max. The 2080tis reached 63 degrees. Honestly, I would expect the opposite. And while these temps are far from unhealthy in both cases, I simply don't understand why the 2080tis are getting hotter under the exact same load, and hence have less headroom. The case that the 2080tis are in is about 20% smaller though (Thermaltake View 31 and Thermaltake View 71), but don't think that explains the difference. But I do wonder if airflow is the core of the cause--but all four of the cards are Turbo cards. In the smaller case (with the 2080tis) I have a 200 mm fan plus a 140 mm fan at the front, and with the larger case I have three 140 mm fans. All are Noctua 140 mm industrial grade fans, and the 200 mm fan is a typical one, that turns at about 720 rpms +/- 20 rpms. Otherwise the cooling in both systems is identical--both with Ryujin 360 AOI cpu coolers, both with a 140 Noctua at the back and a 120 mm or 140 mm at the bottom of the case. Both have high end PSUs (albeit different makers and slightly different otherwise).

I am tempted to swap the cards out from one system to the other to see if that changes anything, but would really and seriously prefer not to do this, and hope to get to the core of the cause for the discrepancy another way.

 

[Deleted member 1560910]

Is you computer running normal ? Why be so concern with such things that really dont matter. It is what it is

 

[MMorris666]

yeah, seems very anal about one thing yet hasn't bothered to do a simple search on the gpu specs. thats the only thing that seems strange to me.

 

[Zerk2012]

Well, I'm not sure if I've pinned down one side of the problem or not. GPU-Z and HWMonitor have differing numbers for GPU power draw. GPU-Z breaks the numbers down more specifically, hence maybe the GPU-Z power draw numbers are more in line with the numbers others have posted about their power draw. So maybe (I'm not sure) what I think is a power draw issue is not one after all. Regardless, HWMonitor is showing the power draw of the main 2080ti as around 40 watts at idle, while it is showing the main 2080 as around 20 watts at idle. This is very strange, as multiple articles show that their total power draw at idle ought to be very similar.

As for my posts being the about same issue--I apologize. But at first I thought the core of the issue was a cooling issue so posted about the temperatures. But then later I thought it was a power issue, hence the different threads. Sometimes a different angle gets people with the right knowledge/info/fix to contribute.

In a nutshell, the 2080s in SLI are doing everything that has exactly the same system demands about 5 or 6 degrees (or higher) cooler than two 2080s in SLI, from when under load to being idle from about a minute after start up. I don't understand why this is, but realize the power draw almost across the board is higher with the 2080tis. But honestly, I don't see why this is. Apparently falsely, I thought higher end cards were better in a few ways, and also more efficient under the same load.

As for specifics, all four cards are Asus Turbo versions with no overclock. The Nvidia drivers, Window 10 drivers, and Intel drivers are all the latest as available via GeForce Experience, Windows updater and Intel Driver & Support Assistant and AMD's website (for chipset drivers). I check every day at startup for and update all drivers. BIOSes are the latest for a Rampage VI Extreme Omega (with the 2080s) and a Crosshair VIII Formula (with the 2080tis) motherboards as posted on the Asus webpages. The CPUs are an Intel 9980xe and a Ryzen 3950x.

The game is World of Warcraft and the max FPS is set to max 60 on all levels with various settings basically in the middle for both systems--and exactly the same. The resolution is 4k. In one test at around exactly the same time (and ambient temperature), the 2080s reached 58 degrees max, with the card fans near max. The 2080tis reached 63 degrees. Honestly, I would expect the opposite. And while these temps are far from unhealthy in both cases, I simply don't understand why the 2080tis are getting hotter under the exact same load, and hence have less headroom. The case that the 2080tis are in is about 20% smaller though (Thermaltake View 31 and Thermaltake View 71), but don't think that explains the difference. But I do wonder if airflow is the core of the cause--but all four of the cards are Turbo cards. In the smaller case (with the 2080tis) I have a 200 mm fan plus a 140 mm fan at the front, and with the larger case I have three 140 mm fans. All are Noctua 140 mm industrial grade fans, and the 200 mm fan is a typical one, that turns at about 720 rpms +/- 20 rpms. Otherwise the cooling in both systems is identical--both with Ryujin 360 AOI cpu coolers, both with a 140 Noctua at the back and a 120 mm or 140 mm at the bottom of the case. Both have high end PSUs (albeit different makers and slightly different otherwise).

I am tempted to swap the cards out from one system to the other to see if that changes anything, but would really and seriously prefer not to do this, and hope to get to the core of the cause for the discrepancy another way.

In a nutshell your looking for problems that are not their!

EDIT I don't even know why your looking at numbers under when the systems are not under load.
Different systems different cards, and your also expecting software to be 100% accurate.

 

[Fluffly Fluff]

But double the power draw at idle is extremely strange. How can double the power draw at idle be wrong when the temperatures are also significantly higher? I can see a 1 or 2 watt difference, which would be in line with what multiple articles test and show, but double?

EDIT: Well, why should the one rig be noisy with fans twice as loud as with the other system with nothing even open? And a more expensive system to boot!

EDIT2: "yeah, seems very anal about one thing yet hasn't bothered to do a simple search on the gpu specs. thats the only thing that seems strange to me."

But I have done numerous searches. Multiple articles explain and show with tests that the power draw of a 2080 at idle is very similar to a 2080ti. The power draw of a 2080ti at idle is NEVER double that of a 2080. NEVER. Here is an example of a page that shows the power draw ought to be similar at idle, not double!! https://www.anandtech.com/show/14663/the-nvidia-geforce-rtx-2080-super-review/15

The computer with the 2080tis is a lot noisier than the other one. So it is also a practical matter as well. Plus, higher temperatures mean higher wear and tear.

 

[dotas1]

https://www.techpowerup.com/247985/...consumption-tested-better-but-not-good-enough

https://www.nvidia.com/en-us/geforce/forums/geforce-graphics-cards/5/284861/rtx-2080-idle-power-consumption-high/

https://www.nvidia.com/en-us/geforce/forums/geforce-graphics-cards/5/285220/rtx-2080-multi-monitoring-idle-powerbug-and-more/

There are hundreds more threads about your issue. I guess you didn't search well enough to find them? Or you simply wanted a new answer beside the thousands given in those threads.

 

[Fluffly Fluff]

https://www.techpowerup.com/247985/...consumption-tested-better-but-not-good-enough

https://www.nvidia.com/en-us/geforce/forums/geforce-graphics-cards/5/284861/rtx-2080-idle-power-consumption-high/

https://www.nvidia.com/en-us/geforce/forums/geforce-graphics-cards/5/285220/rtx-2080-multi-monitoring-idle-powerbug-and-more/

There are hundreds more threads about your issue. I guess you didn't search well enough to find them? Or you simply wanted a new answer beside the thousands given in those threads.

I've seen all of those and tinkered as some advised, but with no fix. Hence, all they do is only point to an issue--which people repeatedly tell me I don't have. The article I linked also points to an obvious issue. Hence, you are merely preaching to the choir. My reason for posting on the forum is looking for a solution--not denials there is an issue or critiques to the effect that I have not previously looked for a fix. To be honest, I have spent well over ten hours at this just searching and tinkering, not to mention about $300 hoping for a fix already via a new AOI cooler that I'd hoped would improve airflow and solve the problem--which also took a few hours to install. Never mind having thoroughly removed every spec of dust inside the case and having blown out the cards to clear away every last spec of dust. It is obvious something is wrong.

Those links are also to problems that literally surfaced years ago. Is it reasonable to assume driver issues have not yet been sorted out? I'd also add that most of the people complaining in those links are worried about power consumption in the 20 watt range, when using multiple monitors (which I am not doing), etc, and not the 60+ watt range--which is what I am seeing at idle.


Lastly, when I turn on the rig with the two 2080ti the fans run high, the thing pumps a ton of hot air into the room at idle. And with the ambient temperature at 30 degrees already it is a nuisance and a problem on multiple levels.

 

[dotas1]

I never said you don't have some kind of problem. I never denied that it is an issue.

The links I provided clearly shows that some drivers did work as intended and some others made it worse. Newer drivers were reported to have that issue do it's not safe to assume it's been fixed after all this time.

If I remember correctly the 60w you mention is for both cards and the links I provided do have 27-29w idle per card with one monitor. So you clearly have the idle power bug.
Temps do rise faster with higher watts and fans will spin faster to cool down the cards, hence the higher noice.

My recommend is to try some of the drivers that were mentioned as fixing the bug but with only 1 card at a time. Try first card no1, then no2, then both. Then share your findings.

 

[Fluffly Fluff]

Thank you for your reply. Incidentally, I have been looking closer at this issue and testing in game while playing Wow. Apparently, reducing the power target has shown some reduction of power use in game for other people. Hence I tried this.

The result was interesting. When limiting the Power Target for the 2080ti cards to 40% I saw the GPU power draw in game of the main card average and roughly stay at around 100 watts. However when limiting the Power Target for the 2080 to only 59% the power the GPU power draw averaged around 50 watts. But all of the in game settings are exactly the same, so this does not make sense.

As for installing older Nvidia drivers I think you can only install one in the system and not for each card individually. However, please correct me if I'm wrong. Also, I have done a bit of the ground work to figure out how to install older drivers but still don't know how. I will get on that though. An Nvidia driver problem would explain this big time.

 

[dotas1]

You can't install different driver per card. I meant to install an older known to be working driver.
Use DDU following these instructions:

How To - How to perform a CLEAN install of your video card drivers.

Updated 9-19-19 How to do a CLEAN install of your graphics card drivers using the Wagnard tools Display Driver Uninstaller A major issue on many systems is the presence of multiple graphics card drivers or drivers that have been incompletely uninstalled followed by the installation of...

forums.tomshardware.com

The result was interesting. When limiting the Power Target for the 2080ti cards to 40% I saw the GPU power draw in game of the main card average and roughly stay at around 100 watts. However when limiting the Power Target for the 2080 to only 59% the power the GPU power draw averaged around 50 watts. But all of the in game settings are exactly the same, so this does not make sense.

You are comparing different systems with different GPUs. 2080tis are power hungry and not power efficient. I see no serious problem there.

 

[Fluffly Fluff]

I contacted Nvidia about the high power draw. The rep is going to contact the development team about it and get back to me. Regardless of what random people think, there is no reason for a such a ridiculously high power draw at idle, and there is simply no need to use such a power hungry heat factory when not completely necessary--which is next to never--unless and until the problem is addressed properly.

 

[dotas1]

Are you willing to try each card as I suggested? Different drivers?

 

[Fluffly Fluff]

To be honest, I really appreciate your replies in this thread. I also want to install older drivers to test them but do not know how. Is there a way to get your hands on them? The Nvidia website only has the latest ones.

I also contacted Nvidia support via their chat support. A technician went through some checks. There is no problem when gaming. He then had me do a clean uninstall and then install the latest driver complete with some fixes to it. The process made no different to my idle power draw though. He then got in contact with an engineer and the engineer said that the power draw was "not a problem." It is a problem though, as the rig is literally pumping a ton of heat into my home every minute it is on when the ambient temperature here is already above 30 degrees C.

Anyhow, I replied to these guys that the power draw at idle ought to be close to the power draw of my 2080 SLI set up, but that it was double. I also pointed out that others running a single 2080 ti and a 4k 60Hz monitor were only seeing around 26 watts of power draw while I am seeing 40+ watts at idle (and this only includes the power draw from the main card--add in 20+ more watts for the second card in SLI), which is considerably higher.

Note that I am a little worried that older drivers might not jive well with the latest Windows version and updates, and also the latest AMD chipset drivers plus the latest Asus Bios for my motherboard. Not to mention that it might cause games to crash, as I have seen that happen when running out of date Nvidia drivers. Hence, even though I have actively looked for how to get my hands on older drivers, I gave up after a while. All of this in mind, I have simply decided to sit on my hands for now, and wait for the next Nvidia driver and/or Bios. I sent a feedback message to Nvidia complaining about how much heat my cards are generating at idle which is heating up my home to some extent. Hopefully they pair down the power draw and/or implement some way for users to do this in the future. Until then, or until the fall, when the weather cools off, I am just going to use another computer. There is no reason to turn on a blast heater in your home in summer either.