Does this EVGA 3080 Ti just much cooler than expected, or is there another sensor I'm not looking at?

[StoleMyOwnCar]

My other topic is getting a bit packed, so I just decided to make another. This is a very simple topic.

Here's what my temperatures on my open box EVGA 3080 Ti (this is the model) look like after hours of running Red Dead 2 at maxed out settings:

This just seems questionably low to me. I thought the VRAM in particular was supposed to run pretty toasty on this generation, but mine isn't even hitting 90C... and the GPU refuses to even go over 77C while being at only 74% fan capacity.

The load numbers look fine, too, as far as I can tell (this is in Borderless windowed so I'm wondering if that's why it doesn't quite hit 100% GPU usage):

Is there an EVGA application I have to download to get access to more sensors? Does this just have a very good cooler?

Just trying to cover all of my bases with this purchase, thanks.

 

[StoleMyOwnCar]

Actually it doesn't quite hit 100% utilization even in full screen mode, just stays pinned at 99% (at least not in Red Dead 2)... are there any reasons for this? Kind of beside the topic I know, just thought I'd throw it out there.

 

[Furious_Styles]

It's throttling @77C, you can tell by the core clock. You can change that in AB if you want.

 

[NattyKathy]

It's throttling @77C, you can tell by the core clock. You can change that in AB if you want.

the GPUz sensors are reporting "PWR" as the performance limit though and the base thermal limit should be 83°
The 399.3W max board power OP is seeing is right at the 400W Power Limit for this card, so I think it's just a case of hitting PL + a good cooler holding temps.

Actually it doesn't quite hit 100% utilization even in full screen mode, just stays pinned at 99% (at least not in Red Dead 2)... are there any reasons for this? Kind of beside the topic I know, just thought I'd throw it out there.

FWIW I usually see 99% as the max utilization on my cards too, except maybe for synthetics like Furmark.

 

[StoleMyOwnCar]

the GPUz sensors are reporting "PWR" as the performance limit though and the base thermal limit should be 83°
The 399.3W max board power OP is seeing is right at the 400W Power Limit for this card, so I think it's just a case of hitting PL + a good cooler holding temps.


FWIW I usually see 99% as the max utilization on my cards too, except maybe for synthetics like Furmark.

Thanks. I don't really plan to anyway (at least not right now), but do you think this GPU has no room for overclocking, if it hits max board power at stock? Or just hard to tell?

I don't plan to try to run Furmark though, seems pointless to me when no game is going to stress it like that anyway...

It's throttling @77C, you can tell by the core clock. You can change that in AB if you want.

So you think EVGA has set the thermal limit lower than the FE for this model?

 

[motqalden]

More temps in evga precision app

 

[NattyKathy]

Thanks. I don't really plan to anyway (at least not right now), but do you think this GPU has no room for overclocking, if it hits max board power at stock? Or just hard to tell?

I don't plan to try to run Furmark though, seems pointless to me when no game is going to stress it like that anyway...

So you think EVGA has set the thermal limit lower than the FE for this model?

Here's a BIOS dump from the model of card you have -> https://www.techpowerup.com/vgabios/235920/evga-rtx3080ti-12288-210719
83° default TL, 400W default PL. Since TL and PL are usually linked and you're already at ~400W, I still say you're hitting power Limit. Hovering a few watts below TDP is normal behavior for a power-limited GeForce board since there's other hidden PL in addition to the total board power.

Hitting PL doesn't mean you don't have OC headroom- you could keep the same 400W limit and raise the Core clocks to get a little extra speed at the stock power which won't affect temperatures much if at all. Or you could raise core clocks and increase the board power to its 450W max (112% Power Limit) and get even higher clocks, but you'd probably have to crank the GPU fans up since you're already in the high 70s at stock power.

Furmark is really only useful for stress testing thermals- it's not a good OC stability tester because it's so power-limited and you're correct that it's def not a good benchmark due to the oddly specific workload tuned for max power/thermals/memory stress

 

[StoleMyOwnCar]

Here's a BIOS dump from the model of card you have -> https://www.techpowerup.com/vgabios/235920/evga-rtx3080ti-12288-210719
83° default TL, 400W default PL. Since TL and PL are usually linked and you're already at ~400W, I still say you're hitting power Limit. Hovering a few watts below TDP is normal behavior for a power-limited GeForce board since there's other hidden PL in addition to the total board power.

Hitting PL doesn't mean you don't have OC headroom- you could keep the same 400W limit and raise the Core clocks to get a little extra speed at the stock power which won't affect temperatures much if at all. Or you could raise core clocks and increase the board power to its 450W max (112% Power Limit) and get even higher clocks, but you'd probably have to crank the GPU fans up since you're already in the high 70s at stock power.

Furmark is really only useful for stress testing thermals- it's not a good OC stability tester because it's so power-limited and you're correct that it's def not a good benchmark due to the oddly specific workload tuned for max power/thermals/memory stress

Thanks for all this information. I'll try playing around with it sometime.

More temps in evga precision app

Is EVGA precision what most people use to overclock these cards? Been a while since I've had EVGA.

 

[GoldenTiger]

Is EVGA precision what most people use to overclock these cards? Been a while since I've had EVGA.

Most still use msi afterburner.

 

[Ducrider748]

Afterburner is usually the go to program for gpu over locking. I personally use Precision X1 because I like the lay out better. User choice is all. That and I can vbios update from there if EVGA has a updated vbios. One other thing is some EVGA cards you can see memory Temps on X1.

To the question that was asked. Get afterburner or X1 and raise the power limit and temp limit. On a benched o erclocked card running benches I raise te.p limit to 90 to 91. Raise fan curve to hit100 percent at 65 degrees. Keep it as cool as possible. Don't worry about gpu usage numbers or watts used. Those numbers are only good to monitor when benching and trying to get the absolute best score.

Most games won't max out the 3080ti to a point where it is running at full 100 percent and can't give anymore. If you are looking for more then a overclock is next. First a better cooling solution will be needed. Hybrid cooler or custom loop. Best part is EVGA is ok with changing cooling solutions and not affecting the warranty.

 

[motqalden]

Thanks for all this information. I'll try playing around with it sometime.

Is EVGA precision what most people use to overclock these cards? Been a while since I've had EVGA.

I use afterburner 99% of the time and Precision to mostly turn off the RGB

 

[motqalden]

Afterburner is usually the go to program for gpu over locking. I personally use Precision X1 because I like the lay out better. User choice is all. That and I can vbios update from there if EVGA has a updated vbios. One other thing is some EVGA cards you can see memory Temps on X1.

To the question that was asked. Get afterburner or X1 and raise the power limit and temp limit. On a benched o erclocked card running benches I raise te.p limit to 90 to 91. Raise fan curve to hit100 percent at 65 degrees. Keep it as cool as possible. Don't worry about gpu usage numbers or watts used. Those numbers are only good to monitor when benching and trying to get the absolute best score.

Most games won't max out the 3080ti to a point where it is running at full 100 percent and can't give anymore. If you are looking for more then a overclock is next. First a better cooling solution will be needed. Hybrid cooler or custom loop. Best part is EVGA is ok with changing cooling solutions and not affecting the warranty.

Yes the FTW3 or higher series does show "memory temps" in Precision but it does NOT show memory hotspot and there can be a huge difference.

 

[StoleMyOwnCar]

I mean I can see the memory temps and hotspot in GPU-Z.

Most games won't max out the 3080ti to a point where it is running at full 100 percent and can't give anymore. If you are looking for more then a overclock is next. First a better cooling solution will be needed. Hybrid cooler or custom loop. Best part is EVGA is ok with changing cooling solutions and not affecting the warranty.

I'm pretty sure I have some headroom considering I'm only running at 77 max and the fan is only at 74%? Or will it saturate very fast? My memory is also only 88 C and I think it's rated for up to 110C now that I did some more research?

 

[NattyKathy]

Best way to tell is to try it. Just for a quick test, load up AB and set fans to 90-100% and Power Limit to 112% with everything else at stock and see if temperatures and noise still seem ok under load. From there you could apply a modest offset to core and memory clocks (+105 core / +800 memory is pretty safe on GDDR6X Ampere cards) if you wish, and if go up from there if you feel like more in-depth tuning/stability testing.

 

[Ducrider748]

I mean I can see the memory temps and hotspot in GPU-Z.

I'm pretty sure I have some headroom considering I'm only running at 77 max and the fan is only at 74%? Or will it saturate very fast? My memory is also only 88 C and I think it's rated for up to 110C now that I did some more research?

Just because memory can run that hot does not mean it is not throttling back. Hotspot is important if you are really pushing for the limits. At least bump fan speeds up and see where you are. If increased fan speed can not bring down temps then you are having some throttling. AKA not enough cooling to keep up with power limit. If temps go down a little max out the power limit and recheck temps.

 

[StoleMyOwnCar]

I see. I'll experiment around later. I'm just happy enough that this runs at 1860 boost while remaining pretty cool and silent to begin with, considering the price I got it for, but if I can squeeze out a bit of performance that's fine too. Thanks for all of the info here, I'll update it when I try some stuff out.

 

[sardonique]

Question...

I've never seen "PWR" listed as a reason for Performance Cap before. What PSU are you using and PLEASE tell me that you aren't using the PCIe "pigtails" for the 3x 8pin power connectors. On a card like the 3080Ti FTW3, you really should be using 3 separate PCIe cables for each of the 3x 8pins.

 

[NattyKathy]

Question...

I've never seen "PWR" listed as a reason for Performance Cap before. What PSU are you using and PLEASE tell me that you aren't using the PCIe "pigtails" for the 3x 8pin power connectors. On a card like the 3080Ti FTW3, you really should be using 3 separate PCIe cables for each of the 3x 8pins.

You've... never seen a GPU be power-limited before? That just means the board is using the maximum power the BIOS allows.

Good tip about using proper power cables, but the GPU wouldn't have any knowledge of that situation and if it was causing a problem it wouldn't trigger Power Limit perfcap, it would just be causing unexplained instability if the voltage drop across the cables was causing the VRMs extra stress and ripple.

 

[StoleMyOwnCar]

I'm not even sure if this Seasonic PSU can support 3 separate PCIE connectors. It came with 2 PCIE connectors that each had another 8x hanging off of them, so technically it can support a 4x8 pin power connector GPU. I guess what I could try doing instead is shuffling around the connectors. ie connect the two separate PCIE connectors to slots 1 and 2, and then have the 2nd one do its tacked on daisy chain thing over to the third? Because I've read the third one doesn't use up nearly as much wattage as the other two, so that might make more sense. But I'm not sure which ones are 1 and 2, to begin with?

Right now I have the separate connectors on the edges and then the daisy chain in the center, from one of them. Assuming the middle has to be slot 2, obviously that can't be the right way to do it.

Idk I've never had any issues with Seasonic's daisy chains...

 

[StoleMyOwnCar]

Actually I did find another PCIE connector, and I happened to have exactly one more available slot for it, so I went ahead and ran 3 separate connectors to each of the PCIE 8 pin connectors. The perfcap reason didn't change though.

When I started up EVGA Precision, it told me I needed to update the firmware on the GPU, so I let it do so. For some reason it reduced the temperatures by a decent margin.

Next I tried just giving it a 104 MHz bump in Precision without changing the power target... didn't have any issues, it mostly hovered around 1950 Mhz, but oscillated from 1895-1980Mhz. I also changed the fan speed to be a bit more aggressive (84% at 70C and then up to 100% at 83C).... and the GPU mostly remained at 71-72C. Just to try it out I also changed the power target to 110% briefly (but I unlinked it from the temperature target and kept the temperature target at 83C). With the aggressive fan curve, it didn't really reach past 74C and the GPU started hovering around 1960-1980 MHz more often instead of the previous average...

I put it back down to 100% because I didn't want to up the voltages despite the temperature looking good, and now it's basically anywhere between 1895-1980 as before, usually around 1950-1965... I guess I'll just leave it alone for another hour, but it seems stable, at least in Red Dead 2 with everything absolutely maxed out at 3440x1440.

Not sure what this implies, I think there's more room to it but I don't really want it blue screening on me randomly to begin with...

 

[KickAssCop]

Its running much hotter than normal and throttling. Also the FTW3 coolers are some of the worst in the business. I am so sick of my 3090 FTW3 that I only use it to mine now. Will be selling it soon in favor of the 3090 Ti (another EVGA ) that I have coming in.
Most silent cards I used in the 3XXX series generation are Palit Gamerock. Those cards are absolutely insane as they never break 70 C and fans usually run at 40% which means they are extremely silent.

Only card I didn't try this generation is ASUS Strix and heard good things about it but never found it at a price I am willing to pay. Your 3080 Ti is running hotter than normal. Switch to OC bios and be prepared to use ear plugs as the fans go above 2000 RPM.

 

[NattyKathy]

Its running much hotter than normal and throttling. Also the FTW3 coolers are some of the worst in the business. I am so sick of my 3090 FTW3 that I only use it to mine now. Will be selling it soon in favor of the 3090 Ti (another EVGA ) that I have coming in.
Most silent cards I used in the 3XXX series generation are Palit Gamerock. Those cards are absolutely insane as they never break 70 C and fans usually run at 40% which means they are extremely silent.

Only card I didn't try this generation is ASUS Strix and heard good things about it but never found it at a price I am willing to pay. Your 3080 Ti is running hotter than normal. Switch to OC bios and be prepared to use ear plugs as the fans go above 2000 RPM.

I'll note that we don't know what OP's ambient temps and case airflow are like. And losing boost isn't the same as throttling... GPUz is showing pure Power Limit all the way across

 

[StoleMyOwnCar]

I mean it's pretty much silent at stock settings, yeah the fans start getting a little weird (they start "squealing" a bit?) at 75% and up, but I wouldn't really call it "worst in the business"..? Your standards are pretty high, this is one of the coolest and quietest cards I've had in a while...

Here this is what my temperatures looked like at 84% fan speed after some hours of gaming, after I set the boost speed to +104 Mhz (while leaving power target the same).

Here is a boost frequency and temperature graph. It kind of went all over the place but I would say on average it stayed around 1920-1965... was a bit random. The game showed now signs of hiccups.

It actually seems stable at the +100 Mhz target even at the default fan curve, and temperatures haven't really gone anywhere.

I'll note that we don't know what OP's ambient temps and case airflow are like. And losing boost isn't the same as throttling... GPUz is showing pure Power Limit all the way across

Well here is my build at the moment:

The Kraken is in intake, both bottom fans came with this Lian Li case, and are also intake, blowing upwards. I know that the sound card is kind of in the way of one of them, but the air still seems to mostly be making it over there, though I can try taking the sound card out. The top fans and the back fan are exhaust.

Here is a full parts listing:

https://pcpartpicker.com/list/2XrL9r

Ambient temperature... I'm not sure what it is in here. I have the air conditioning set to 69F. It might be a degree or two warmer than that in here when the PC is running full tilt for a while.

 

[StoleMyOwnCar]

This is what the temperatures look like with the default fan curve while putting +100 Mhz target on the GPU... yeah it usually gets up to 1920-1950 and not really +100 in practice, but it's still over 1900 almost all of the time so it's a pretty decent bump and with no temperature or power limit increase...

Is this really considered bad? I mean I don't care that much considering how much I paid for this 3080 Ti, but frankly I'm really shocked. This was one of the coolest and quietest cards I've had in a while and it's staying well under all of its temperature targets so it's just an eye opener for me...

 

[Schro]

Is this really considered bad? I mean I don't care that much considering how much I paid for this 3080 Ti, but frankly I'm really shocked. This was one of the coolest and quietest cards I've had in a while and it's staying well under all of its temperature targets so it's just an eye opener for me...

I'd say your experience is within the ballpark of expectations for that card. Also, keep in mind that different games, resolutions, feature settings (RT/DLSS) and even different areas within games can end up producing different temperature results and different boost clock results.

I saw 78.3C in Cyberpunk 2077 (1440p, no RT/DLSS) at full stock configuration on an open test bench, 73C with fans at 100% with the power limit maxed, +160 GPU and +1400 on memory with that card.

 

[StoleMyOwnCar]

I'd say your experience is within the ballpark of expectations for that card. Also, keep in mind that different games, resolutions, feature settings (RT/DLSS) and even different areas within games can end up producing different temperature results and different boost clock results.

I saw 78.3C in Cyberpunk 2077 (1440p, no RT/DLSS) at full stock configuration on an open test bench, 73C with fans at 100% with the power limit maxed, +160 GPU and +1400 on memory with that card.

Thanks, I guess I'll need to get Cyberpunk 2077 for testing sometime. I've been meaning to play it anyway... but idk if I wanna pay full price for it just to use it as a test bench

Could someone please tell me what this sound is?


I tried turning the GPU fans up to 100% and am noticing this really annoying, oscillating... kind of metallic high pitched buzzing sound? I guess that's the best way I can explain it? The fans at 100% themselves sound pretty much fine, but I don't understand what this sound is. It's not the GPU's coil while, because the GPU has never exhibited any coil whine under load or not under load. It only starts up when just the GPU fans are getting up past 77% or so. At stock voltages, they don't ever need to do that to begin with so I've never noticed it. Is it a busted bearing or something?

Edit: Testing them individually, I think they all do it... I guess they just used fans that are fine at up to 77% and then just start getting very annoying past that. I've never heard of a bearing that sounds this bad at 100%, so I can see what you were saying KickAssCop

And actually happens at lower RPM's too, just a lot less noticeable now that I inspect it closer.

 

[Tactlesss]

Sounds like something is loose, needs to be lubricated and a stray wire is rubbing against the fan blades

 

[StoleMyOwnCar]

I'm going to try a few different things:
1. Look at the GPU and see if any of the fan screws are loose or maybe the backplate screws or something else loose, causing vibration.

2. Take it over to a friend's house and see if it's related to the power circuitry of my home, because I started noticing that my old computer is suddenly exhibiting this sound... the 8700k that I took from the living room and moved over to the exercise room also has a pretty similar sound profile right now and I can't figure out where it's coming from. It's never had coil whine before, and not sure why it would start now. It might be the old Corsair 850HX that I moved into it, because that's literally the only thing that's different in that build right now.. could it be causing distant sockets to cause coil whine, too? I guess I'll also try switching that computer off and trying other outlets as well.

Oh and I guess I'll try turning off my aquarium canister filters as well just in case.

3. If neither of these things show anything, I'll contact EVGA.

4. If they don't respond favorably, I might take it back to Microcenter. Although it works at stock settings right now while being silent, and it was only 910$ so the price was great, frankly this lets out a pretty annoying sound at anything even slightly above the stock fan temps. So that means if the ambient temps go up at all, there goes the silence. Not sure though, honestly. if it works at stock and mostly boosts to 1900s without any voltage tweaking and stays pretty cool, it's difficult to commit to returning it considering how good of a deal I got it for... but if EVGA says no without any investigation, that means they were never planning to honor an open box card for warranty to begin with, which is just gonna be pointless, and at that rate I'd rather pay a bit more for the 6950XT that's going to come out soon, or just stick with my 2080 till the next gen I guess. While this GPU really does some great numbers and looks amazing, I can live without it still.

Thanks for all the input in this topic, everyone.

 

[GoldenTiger]

I'm going to try a few different things:
1. Look at the GPU and see if any of the fan screws are loose or maybe the backplate screws or something else loose, causing vibration.

2. Take it over to a friend's house and see if it's related to the power circuitry of my home, because I started noticing that my old computer is suddenly exhibiting this sound... the 8700k that I took from the living room and moved over to the exercise room also has a pretty similar sound profile right now and I can't figure out where it's coming from. It's never had coil whine before, and not sure why it would start now. It might be the old Corsair 850HX that I moved into it, because that's literally the only thing that's different in that build right now.. could it be causing distant sockets to cause coil whine, too? I guess I'll also try switching that computer off and trying other outlets as well.

3. If neither of these things show anything, I'll contact EVGA.

4. If they don't respond favorably, I might take it back to Microcenter. Although it works at stock settings right now while being silent, and it was only 910$ so the price was great, frankly this lets out a pretty annoying sound at anything even slightly above the stock fan temps. So that means if the ambient temps go up at all, there goes the silence. Not sure though, honestly. if it works at stock and mostly boosts to 1900s without any voltage tweaking and stays pretty cool, it's difficult to commit to returning it considering how good of a deal I got it for... but if EVGA says no without any investigation, that means they were never planning to honor an open box card for warranty to begin with, which is just gonna be pointless, and at that rate I'd rather pay a bit more for the 6950XT that's going to come out soon, or just stick with my 2080 till the next gen I guess. While this GPU really does some great numbers and looks amazing, I can live without it still.

Thanks for all the input in this topic, everyone.

I wouldn't go so far as #4. Your card was a great deal.... If you have higher ambient Temps later, just set the quiet speed on the fans as the max and the card will lower its clocks and voltage a tad, leaving you with a fast silent card still. It doesn't sound like anything is wrong at all with your card.

 

[bluestang]

I use the EVGA PX1 software.

Lower you TDP/Power Level to 75%-80% and then do +50 or +100 on the core and +250 or +500 on the memory. 100% power gains you nothing if temps are limiting you clocks. Lower temps = higher clocks in my experience and I usually have my 3080ti FTW3 fully loaded doing DC work when not gaming or using the PC.

 

[StoleMyOwnCar]

I wouldn't go so far as #4. Your card was a great deal.... If you have higher ambient Temps later, just set the quiet speed on the fans as the max and the card will lower its clocks and voltage a tad, leaving you with a fast silent card still. It doesn't sound like anything is wrong at all with your card.

Well I mean did you see that youtube video I linked above? Are you sure that high pitched sinusoidalish noise is a normal sound for a GPU fan to make? I did some physical tests and it doesn't look like anything physical that I can see is causing it. I tightened the screws on each GPU fan slightly but it did nothing. It reverberates within the tempered glass case so it ends up quite loud.

I suppose yeah at stock (and even slightly above stock) it does run just fine and is pretty quiet though...

I use the EVGA PX1 software.

Lower you TDP/Power Level to 75%-80% and then do +50 or +100 on the core and +250 or +500 on the memory. 100% power gains you nothing if temps are limiting you clocks. Lower temps = higher clocks in my experience and I usually have my 3080ti FTW3 fully loaded doing DC work when not gaming or using the PC.

So lower TDP while keeping the same boost clock? Will that actually work...? Couldn't hurt

 

[chameleoneel]

My other topic is getting a bit packed, so I just decided to make another. This is a very simple topic.

Here's what my temperatures on my open box EVGA 3080 Ti (this is the model) look like after hours of running Red Dead 2 at maxed out settings:
View attachment 470726

This just seems questionably low to me. I thought the VRAM in particular was supposed to run pretty toasty on this generation, but mine isn't even hitting 90C... and the GPU refuses to even go over 77C while being at only 74% fan capacity.

The load numbers look fine, too, as far as I can tell (this is in Borderless windowed so I'm wondering if that's why it doesn't quite hit 100% GPU usage):
View attachment 470727

Is there an EVGA application I have to download to get access to more sensors? Does this just have a very good cooler?

Just trying to cover all of my bases with this purchase, thanks.

In my experience, Nvidia cards stop boosting much over the advertised boost clock, around 74C. The upper thermal limit for harder throttling is 83C, but the boost will stop pushing higher than advertised, around 74C. Cool the card under 74C, and you will usually see boosting higher than advertised.

As noted, you can use tools like Afterburner, to try and increase the soft cap on boost clocks. But IME, it doesn't always work.

 

[mnewxcv]

Man, stop watching Temps and tweaking clocks and fans. Just run the card. Temps are fine and performance is fine. The card has a 3 year warranty so just enjoy it and don't worry about it.

 

[GoldenTiger]

Man, stop watching Temps and tweaking clocks and fans. Just run the card. Temps are fine and performance is fine. The card has a 3 year warranty so just enjoy it and don't worry about it.

This.

 

[KickAssCop]

For reference I have a MSI Suprim X 3080 Ti that runs at 73 C, 86 C hotspot and about 84 C memory (max temps). Fan speed hovers around 70% or under 2000 RPM.

Not the best I used but decent and better than my EVGA 3090. I am hoping my incoming EVGA 3090 Ti does better given how massive that cooler is.

 

[StoleMyOwnCar]

Man, stop watching Temps and tweaking clocks and fans. Just run the card. Temps are fine and performance is fine. The card has a 3 year warranty so just enjoy it and don't worry about it.

I mean... you're probably right, but that's a really weird thing to say for Hardforums of all places...

Anyway, I've been doing some testing and this card at even 82-84% fans basically never even hits 70C. I upped the base Mhz by 100 and basically came out to this... I left it running Cyberpunk 2077 (which is on sale at Humble store for 30$ at the moment by the way, good timing lol) overnight while I slept:

At 81-84% fan speed I basically can't hear it over my game, and it's running incredibly cool and boosting to about 1950-1980 on average. Haven't had any crashes after hours of playing, though sometimes I get this weird pause (very rare, like once every several hours). I kind of wish I could set a hard cap on speed, I don't need it occasionally boosting to over 2GHz like that, I think that's what the hiccups are at.

This is what the temps look like while actively playing for a few hours just now:

I think overall I was a bit too hasty and paranoid since it was open box... this is a fine card. The cooler's a bit wonky at higher fan speeds but it definitely works. And at 81-84% it's still easy to tolerate while keeping the thing pretty cool. For 900 bucks I can't really complain, it performs quite well even if it didn't overclock. Which I guess it does...

I don't think I'm gonna try Furmark though, heard it can still damage GPUs...

Edit: Question, what's the VOp in the screenshot above? That's a new one. I didn't jack up the maximum power so I don't get it...

Edit: Oh that's with DLSS, it boosts lower without it... oh well.

 

[WilyKit]

having a hard time determining if you’re troubleshooting something or thinking aloud at this stage

 

[Tactlesss]

having a hard time determining if you’re troubleshooting something or thinking aloud at this stage

It's better to think about threads like this one as a monologue, and everyone else like a Greek Chorus.

 

[Mchart]

Man, stop watching Temps and tweaking clocks and fans. Just run the card. Temps are fine and performance is fine. The card has a 3 year warranty so just enjoy it and don't worry about it.

Yep. IMO, no reason to overclock these latest cards. You aren't getting anything you'll notice out of it. It's not like you're overclocking a Ti4200 to Ti4600 speeds.

 

[mhenley]

Mchart is right, overclocking doesn't really have the intended effect because we hit power limits or thermal limits... BUT, undervolting is the new overclocking. Lowers power and heat so that those are no longer preventing your max boost. Here's a quick guide: https://www.reddit.com/r/nvidia/comments/koub76/3_ways_to_undervolt_in_msi_afterburner_for_3080/

I have my EVGA 3080 12GB running at .912v, GPU temp 79c (Mem 84c), default fan curve, and it boost to 1980mhz steadily while playing, occasionally dipping to 1950mhz in longer sessions. My PerfCap has a lot more grey than green too. This was tested with Cyberpunk 2077 at max detail and ray tracing with DLSS set to quality.