F.E.A.R. 3: 483 watts peak usage indicated

[80251]

According to HWiNFO64 FEAR3 is spiking to 483 watts peak power draw on my 1080ti at 2164/6026, which is amazing for such an old DX11 game that only fully loads one core and partially loads another.

 

[pavle]

I'd suggest somehow limiting the 1080Ti, so it doesn't fail at such power levels... but you got it O/Ced, so whatever.

 

[80251]

FEAR 3 at 4k resolutions with maxed eye candy can't maintain over 60 FPS on my 1080ti

 

[AlwaysHope]

That's why it's always wise to invest in good quality PSU with high wattage capability, just in case.

 

[Space Lynx]

That's why it's always wise to invest in good quality PSU with high wattage capability, just in case.

I thought modern gpu's and up to date drivers installed already took care of all of this? There is supposed to be protections built in I thought?

 

[AlwaysHope]

I thought modern gpu's and up to date drivers installed already took care of all of this? There is supposed to be protections built in I thought?

In gaming, all that can't account for poor coding.

 

[Space Lynx]

In gaming, all that can't account for poor coding.

a frame limit from rivatuner would stop it in its tracks though? i tend to always have rivatuner set with a frame cap in background. simply because i have seen 3000 fps in a game menu before and my gpu whining, lol. that was many years ago though

 

[AlwaysHope]

a frame limit from rivatuner would stop it in its tracks though? i tend to always have rivatuner set with a frame cap in background. simply because i have seen 3000 fps in a game menu before and my gpu whining, lol. that was many years ago though

I don't use rivatuner, so can't say. But the OP claim of 483w peak usage from his 1080Ti could be a bug in HWiNFO64 ? I've seen up to 550w for my whole system once @ 1440p, that was in The Outer Worlds with its UE4 engine & that was measured from a purpose built electricity consumption device from the wall plug.

 

[wolf]

a frame limit from rivatuner would stop it in its tracks though

Given they use a 1080Ti, you can use the FPS limiter built into Nvidia Control Panel, can be done per game, no third party software required (is a huge bugbear for some it seems)

 

[80251]

The Nvidia CP FPS limiter doesn't seem to work for DX9 games, at least the two I've tried it on: Left 4 Dead 2 and FEAR: Perseus Mandate.

For FEAR 3 the problem isn't outrageously high FPS, it's low FPS. It regularly drops below 60 FPS at 4434x2494 (DSR resolution).

 

[cvaldes]

I don't use rivatuner, so can't say. But the OP claim of 483w peak usage from his 1080Ti could be a bug in HWiNFO64 ?

Yeah, 483W in HWiNFO makes zero sense, most likely a spurious reading or bad sensor.

The PCI limit for an 8-pin power connector is 150W. OP's card -- the 1080 Ti -- only has one 8-pin power connector + one 6-pin power connector and draws 250W. The card would need three 8-pin power connectors plus the 75W from the PCIe slot to provide ~485W.

Oddly, OP wrote "only fully loads one core and partially loads another" which appears to be referring to CPU cores, not GPU load which is a separate matter.

 

[Space Lynx]

Given they use a 1080Ti, you can use the FPS limiter built into Nvidia Control Panel, can be done per game, no third party software required (is a huge bugbear for some it seems)

I always had trouble with nvidia cap limiter not working in some games, and for some reason rivatuner just works in everything

 

[Toothless]

I'd suggest somehow limiting the 1080Ti, so it doesn't fail at such power levels... but you got it O/Ced, so whatever.

A 1080ti, unless shunt modded, will never pull over 300w.

Guys, come on, this is like, basic common sense for Pascal. A Titan Xp can't even hit 300w and it has 10% more cores. HWinfo isn't reading right or the sensor is bugged.

 

[ir_cow]

A 1080ti, unless shunt modded, will never pull over 300w.

Guys, come on, this is like, basic common sense for Pascal. A Titan Xp can't even hit 300w and it has 10% more cores. HWinfo isn't reading right or the sensor is bugged.

Yep. Even the super high end retail cards only have a 350W Retail BIOS. To get around this you need to shunt mod or flash to a unlimited XOC BIOS.

 

[ThrashZone]

According to HWiNFO64 FEAR3 is spiking to 483 watts peak power draw on my 1080ti at 2164/6026, which is amazing for such an old DX11 game that only fully loads one core and partially loads another.

Hi,
I wouldn't believe that
I'd at least confirm with an older more appropriate version for 10 series like v6.08 portable

Lots of 1080ti were made some with 2x8pin and some with 1-8 1-6pin so which do you actually have but none will crank that much power I don't believe just by gaming I mean it's not amazon gaming is it

HWiNFO

HWiNFO: Free software download for olderv versions of windows.

www.fosshub.com

 

[cvaldes]

Lots of 1080ti were made some with 2x8pin and some with 1-8 1-6pin so which do you actually have but none will crank that much power I don't believe just by gaming

Unfortunately OP didn't share which specific model 1080 Ti they own. 8+6 pin power was for the reference card.

OP also didn't state whether or not any actual problems manifested. Just a high reading (that makes no sense).

 

[Toothless]

Unfortunately OP didn't share which specific model 1080 Ti they own. 8+6 pin power was for the reference card.

OP also didn't state whether or not any actual problems manifested. Just a high reading (that makes no sense).

I pushed my MSI 1080ti with 2x8 power and only managed 280w peak. That's power limit+voltage bumped, and no throttle. I seriously can't see the 400w+ readings being legit.

 

[cvaldes]

I pushed my MSI 1080ti with 2x8 power and only managed 280w peak. That's power limit+voltage bumped, and no throttle. I seriously can't see the 400w+ readings being legit.

I know, I know. That's why I specifically called out a spurious reading or bad sensor. I have a 2070 Super that has 8+6 and is rated for 225W. I know OP's card is Pascal not Turing but there's no way OP can get to ~480W in normal operation.

Unfortunately OP has provided very little detail about their hardware and more importantly whether or not there were any actual consequences so we're left just guessing.

 

[80251]

The 1080ti in question is a MSI 1080ti Duke that has the Asus Strix XOC VBIOS flashed (no power limits with this VBIOS). HWiNFO64 also indicates the two 8-pin PCIe connectors are also experiencing significant voltage drops at this indicated power draw:
GPU 8-pin#1 min.: 11.48V, GPU 8-pin#2 min.: 11.512V

 

[ShiBDiB]

The 1080ti in question is a MSI 1080ti Duke that has the Asus Strix XOC VBIOS flashed (no power limits with this VBIOS). HWiNFO64 also indicates the two 8-pin PCIe connectors are also experiencing significant voltage drops at this indicated power draw:
GPU 8-pin#1 min.: 11.48V, GPU 8-pin#2 min.: 11.512V

I just checked every ASUS ROG Strix 1080Ti vbios in the database and the highest W limit was 330... Your sensor is faulty (or you're using some awful bootleg VBIOS).

 

[ir_cow]

I just checked every ASUS ROG Strix 1080Ti vbios in the database and the highest W limit was 330... Your sensor is faulty (or you're using some awful bootleg VBIOS).

Op said it was flashed to the XOC BIOS. Those are unlimited and kinda not safe for air cooling.

 

[c2DDragon]

There is always this solution to avoid any GPU throwing FPS to the moon :

Rivatuner Statistics Server 7.3.6 - Download For Windows

Free RivaTuner Download For you Windows and Mac. 100% Free and Safe. Get the Latest version now and go-to solution for top-tier file protection.

www.rivatuner.org

 

[80251]

Op said it was flashed to the XOC BIOS. Those are unlimited and kinda not safe for air cooling.

I strapped a 127x38mm and a 127x50mm fan along with 50mm of shrouds to the stock heatsink, replaced the stock TIM with conductonaut and have nearly the entire backplate covered with thermal pads. It never goes above 64°C even on the hottest summer days, but I can't maintain the 2164 core clocks during the summer months.
FWIW, I've been running this 1080ti with the ASUS XOC VBIOS practically ever since I bought it (because the stock VBIOS would hit power limits even if the thermals were under control), the MSI Duke 1080ti has a very robust VRM circuit and a largish heatsink.

 

[ir_cow]

I strapped a 127x38mm and a 127x50mm fan along with 50mm of shrouds to the stock heatsink, replaced the stock TIM with conductonaut and have nearly the entire backplate covered with thermal pads. It never goes above 64°C even on the hottest summer days, but I can't maintain the 2164 core clocks during the summer months.
FWIW, I've been running this 1080ti with the ASUS XOC VBIOS practically ever since I bought it (because the stock VBIOS would hit power limits even if the thermals were under control), the MSI Duke 1080ti has a very robust VRM circuit and a largish heatsink.

You can do whatever you want with your hardware. I'm just saying there is a reason why retail cards do not have unlimited power options besides just for cooling reasons. I remember seeing a ASUS XOC BIOS for the 2080 Ti that had the voltage stuck at 1.25v. These are meant for internal lab use and or actual LN2 runs. Because when the card dies, no one cares in the lab

 

[80251]

@ir_cow
Is it still possible for end users to get a copy of the ASUS XOC BIOS for the 2080ti? I'd be interested to see what watercoolers have been able to achieve with that BIOS on their 2080ti's.