Yup my PCIE power draw shows 66W in Metro Exodus EE at 3440x1440p. I need to retest my card with other games to see how much actual power it pulls.

I’m honestly really disappointed with this card, considering how limited it is by power. I own a 3080 Tuf non-OC and and 3080 XC3 Ultra and those cards are easily able to pull more than their rated 320W of power and OC decently. This is the first GPU I’ve owned that cannot hit its rated boost clock in a game. I also purchased the XC hybrid kit specifically for this GPU as it was going to go in my personal rig.

So I did some more testing, primarily comparing my 3080 Ti XC3 and my 3080 XC3 Ultra and the 3080 XC3 Ultra will pull more power than it's rated for reliably even if the power slider is not set to the max. In Metro Exodus EE, the 3080 Ti will sit around 320W - 325W with an occasional spike near 330W while also not being able to maintain 1665 mhz. The 3080 XC3 Ultra, on the other hand, will comfortably pull close to 340W with occasional spikes of more than 345W while maintaining or passing its 1775 mhz rated boost clock. 
 
Both cards do use the same cooler, have 2x 8 pins connectors, and a similar PCB design so it is really weird seeing the 3080 Ti XC3 not being able to pull more power than the lower tier card. 
 
3080 Ti XC3 in Metro Exodus EE - https: //imgur. com/ eK0sG1n
3080 XC3 Ultra in Metro Exodus EE - https: //imgur. com/ Fab4LLD

speedysloth

ZheGoose
So I did some more testing, primarily comparing my 3080 Ti XC3 and my 3080 XC3 Ultra and the 3080 XC3 Ultra will pull more power than it's rated for reliably even if the power slider is not set to the max. In Metro Exodus EE, the 3080 Ti will sit around 320W - 325W with an occasional spike near 330W while also not being able to maintain 1665 mhz. The 3080 XC3 Ultra, on the other hand, will comfortably pull close to 340W with occasional spikes of more than 345W while maintaining or passing its 1775 mhz rated boost clock. 
 
Both cards do use the same cooler, have 2x 8 pins connectors, and a similar PCB design so it is really weird seeing the 3080 Ti XC3 not being able to pull more power than the lower tier card. 
 
3080 Ti XC3 in Metro Exodus EE - https: //imgur. com/ eK0sG1n
3080 XC3 Ultra in Metro Exodus EE - https: //imgur. com/ Fab4LLD

Are there any other games that show a similar behavior for you? What is your memory junction temperature while the game is running?

Memory junction temperature for the 3080 Ti sits at 78C - 80C while playing Metro Exodus Enhanced Edition. I did not record the temperature for the 3080 XC3 Ultra memory but I'd assume it's around the same since both have the same cooler. During both tests, the memory was running at stock speed. 
 
As for the below-rated clock speeds, Metro Exodus EE is the only game I have noticed the GPU clocks behaving this way for the 3080 Ti XC3. This behavior also only happens on the 3080 Ti, the 3080 XC3 Ultra does not have any issues maintaining its rated boost clock. It is also the most demanding game that I own followed by Cyberpunk. In Cyberpunk 2077, with Ray tracing enabled and ray-traced lighting set to Ultra, the 3080 TI XC3 sits between 1665 - 1710 mhz. The power draw for both games is around 325W and does not exceed 330W for the 3080 Ti. 
 
Hallmark also mentioned a few posts back that in A Plague Tale: Innocence at 4k Ultra settings, that a 3080 Ti XC3 Ultra could also not maintain its rated boost clock and fell to below 1600 mhz. I don't own that game so I can't test it, unfortunately. 

ZheGoose

speedysloth

ZheGoose
So I did some more testing, primarily comparing my 3080 Ti XC3 and my 3080 XC3 Ultra and the 3080 XC3 Ultra will pull more power than it's rated for reliably even if the power slider is not set to the max. In Metro Exodus EE, the 3080 Ti will sit around 320W - 325W with an occasional spike near 330W while also not being able to maintain 1665 mhz. The 3080 XC3 Ultra, on the other hand, will comfortably pull close to 340W with occasional spikes of more than 345W while maintaining or passing its 1775 mhz rated boost clock. 
 
Both cards do use the same cooler, have 2x 8 pins connectors, and a similar PCB design so it is really weird seeing the 3080 Ti XC3 not being able to pull more power than the lower tier card. 
 
3080 Ti XC3 in Metro Exodus EE - https: //imgur. com/ eK0sG1n
3080 XC3 Ultra in Metro Exodus EE - https: //imgur. com/ Fab4LLD

Are there any other games that show a similar behavior for you? What is your memory junction temperature while the game is running?

Memory junction temperature for the 3080 Ti sits at 78C - 80C while playing Metro Exodus Enhanced Edition. I did not record the temperature for the 3080 XC3 Ultra memory but I'd assume it's around the same since both have the same cooler. During both tests, the memory was running at stock speed. 
 
As for the below-rated clock speeds, Metro Exodus EE is the only game I have noticed the GPU clocks behaving this way for the 3080 Ti XC3. This behavior also only happens on the 3080 Ti, the 3080 XC3 Ultra does not have any issues maintaining its rated boost clock. It is also the most demanding game that I own followed by Cyberpunk. In Cyberpunk 2077, with Ray tracing enabled and ray-traced lighting set to Ultra, the 3080 TI XC3 sits between 1665 - 1710 mhz. The power draw for both games is around 325W and does not exceed 330W for the 3080 Ti. 
 
Hallmark also mentioned a few posts back that in A Plague Tale: Innocence at 4k Ultra settings, that a 3080 Ti XC3 Ultra could also not maintain its rated boost clock and fell to below 1600 mhz. I don't own that game so I can't test it, unfortunately. 

 
I tested A Plague Tale: Innocence but I only have 1440P with all max settings, the weird thing with this game, this game draws the lowest power out of any game I have played with my new card, bust still maintains 1900 Mhz clock.  The power I noticed is around 300-320.  Most of the other games I have tested but still maintain 1900 - 1935 mhz usually pulls around 330-340, for example Battlefield 5 or Call of Duty Cold war.,  Though some games go as low as 1845 mhz.  I don't think I have ever seen below 1800 but I can be wrong since not constantly monitoring.
 
Here are some screenshots with riva tuner stats
 
https://postimg.cc/r0CbYRw3
 
https://postimg.cc/Ln0W7pm9 
 
I also have max frame locked to 141 FPS because it will go much higher then 144 which my gsync monitor is only 144 hz

Tricerakash
I think we 3080ti XC3 ultra owners all seem to agree, this card is very underwhelming for many reasons, so what can we do about it?

 
Hope for a new Bios that will help it, live with it or do a step up to FTW3 which it seems it has their own set of issues lol 

speedysloth
I have game pass so I downloaded Plague Tale. My processor was a bottleneck for the game at 1440p so I used DSR to render at 4k/Ultra. Played the first 15 minutes of the game with an undervolt (1890MHz @ 850mv) :
 
Average Temperature: 82 degrees (70% fan speed)
Average GPU Voltage: 0.824v
Average Power: 341W (usually stayed at 340 with some spikes to 355W).
Average Clock Speed: 1674MHz with a peak of 1860MHz
 
So the clocks are definitely scaled back in this game for me. They were higher at 1440p (~1750Mhz) but my GPU usage was 80%.


Edit:
 
A few more observations:
 
- Changing fan speed to 100% reduced my temperature to around 70 degrees which made the card boost up to 1740MHz on average in a 5 minute run.
- I am getting 95 frames on average @ 4k/Ultra. Limiting the frame rate to 60 allowed me to boost up to the max undervolt that I have set (1890Mhz) and it was locked there for my 5 minute run.
- The power usage with fans at max speed was the same as them at 70%, 341W average.
- Power usage was 270W with the frame rate locked to 60.
 
So the card is boosting fine, the XC3 cooler naturally isn't as good as the FTW3 one so that is holding back the boosting behavior.

so I got lower power , my power stayed 300-320, my average was 307 watts in this game and i held 1905-1935 mhz clock the entire time.  What CPU are you running?  How is it possible my power is much lower but my gpu clocks are much higher average @ 1440p?  My temps were 67.  Is it because my CPU is not a bottleneck and allows card to boost higher?  If power is tied to the boost clocks, then how did I get it 1900+ mhz with so little power I don't get it.  I wonder if it really reporting the right wattage that it is using.

speedysloth

kongfra

speedysloth
I have game pass so I downloaded Plague Tale. My processor was a bottleneck for the game at 1440p so I used DSR to render at 4k/Ultra. Played the first 15 minutes of the game with an undervolt (1890MHz @ 850mv) :
 
Average Temperature: 82 degrees (70% fan speed)
Average GPU Voltage: 0.824v
Average Power: 341W (usually stayed at 340 with some spikes to 355W).
Average Clock Speed: 1674MHz with a peak of 1860MHz
 
So the clocks are definitely scaled back in this game for me. They were higher at 1440p (~1750Mhz) but my GPU usage was 80%.


Edit:
 
A few more observations:
 
- Changing fan speed to 100% reduced my temperature to around 70 degrees which made the card boost up to 1740MHz on average in a 5 minute run.
- I am getting 95 frames on average @ 4k/Ultra. Limiting the frame rate to 60 allowed me to boost up to the max undervolt that I have set (1890Mhz) and it was locked there for my 5 minute run.
- The power usage with fans at max speed was the same as them at 70%, 341W average.
- Power usage was 270W with the frame rate locked to 60.
 
So the card is boosting fine, the XC3 cooler naturally isn't as good as the FTW3 one so that is holding back the boosting behavior.

so I got lower power , my power stayed 300-320, my average was 307 watts in this game and i held 1905-1935 mhz clock the entire time.  What CPU are you running?  How is it possible my power is much lower but my gpu clocks are much higher average @ 1440p?  My temps were 67.  Is it because my CPU is not a bottleneck and allows card to boost higher?  If power is tied to the boost clocks, then how did I get it 1900+ mhz with so little power I don't get it.  I wonder if it really reporting the right wattage that it is using.

I have a ryzen 3600.Im not sure why it's boosting higher with lower power but it was in my case. Locking the frame rate to 60 reduced the power consumption, the gpu usage and boosted the core clock for me.

It card could be reporting the power wrong but I got the same readings in both hwinfo64 and afterburner. All the temperatures and voltages lined up with the high power usage from what I could see. My junction and memory temperatures e.g. were the highest I've seen them since I've gotten the card.

 
I see what you saying about readings in the same across various apps, but was thinking it is reading from the card itself, don't they all read the same sensor?  Maybe the sensor that shows power is buggy on the card itself?  I really don't know just throwing it out there.  My memory and junction temps never got higher then 80 if I recall but I would need to pay more attention and check it again playing the game.  I wasn't actively monitoring memory and junction only the GPU itself.  I can play the game later after work and see what my memory junction gets to out of curiosity, it actually a pretty good game!

Thanks to you guys keeping this alive. I stumbled across this thread and ran Timespy Extreme with GPUz and yeah, I hit 352w for about a half a second in the entire thing. most of the time the card sits at 315-325w. Its the hybrid, so the pump is taking a little, but I have the fans connected to a controller. RGB off.
 
Let me know if theres anything you guys want me to run tonight and I'll provide it to help the cause.

KingEngineRevUp

dalab
Thanks to you guys keeping this alive. I stumbled across this thread and ran Timespy Extreme with GPUz and yeah, I hit 352w for about a half a second in the entire thing. most of the time the card sits at 315-325w. Its the hybrid, so the pump is taking a little, but I have the fans connected to a controller. RGB off.
 
Let me know if theres anything you guys want me to run tonight and I'll provide it to help the cause.

If you can, do timespy graphic test 2, log it in GPU-Z and send the .txt. logs. I think it's good we keep comparing peoples power draws and their average voltages and any other correlations we can find. 
 
I think Speedy is the only one here that averages 345W of power draw but his average voltages aren't that impressive compared to the Ventus we had tested from another user. 
 
Like Grscrxsi was saying, it can be inefficiencies and leaks. But it's kind of crazy for a card to leak that much, where speed draws 30-50W more on average than Kram and Grscrxsi with no real benefit to show for it. Not like the Ventus we saw. 
 
It can just be... The Ventus is just a better designed board compared to the XC3. Now obviously we have a poor pool of data, just one Ventus tested and a few XC3s. 
 
But I think we have had enough XC3s tested here to see a correlation, they have ranges of power draws, have similar average clocks in PL loads and similar average voltages. They don't perform as well as a 3080 Ti FE, which is sad. 

 
Aside from not seeing full power , is that what is really causing slow boost speeds?
 
Is it really that bad compared to the FE, here is one review:
 
https://www.thefpsreview.com/2021/06/24/overclocking-nvidia-geforce-rtx-3080-ti-founders-edition/
 
His average boost is 1804 mhz before overclocking.  In games the lowest I have seen is 1845 mhz boost and I am at stock but I have seen it boost constantly as high as 1950 mhz especially in games like call of duty cold war it pegs at 1950 from what I recall but can run another test later tonight as well. In A Plague Tale: Innocence I get constant 1950 and that only at watts of 330-338.  Red dead redemption 2 is constant 1845 mhz for me I noticed that last night and never saw it dip at all.   What i don't really understand enough about the technology why the boost changes per game.  Like why is call of duty my highest boost, and Red Dead redemption 2 is only 1845?  I would think it would be the same all the time if my thermals are under 70 which they are,
 
 
I haven't even tried any overclocking not sure I will bother.  I am not saying people here are not having issues, obviously they are and some are much worst then others, but can we outright say all FE outdo XC3?  Maybe the FE cards that reviewers get are the best ones sent from nvidia?  Maybe it just has to do with silicon lottery?  just like CPU silicon lottery.  As far as I know they are only rated for 1625 MHZ so anything above that is extra right?  I just wonder if power has to do with performance if my power in Plague Tale is only 300-320 with average of 307, but I am getting 1905-1920 boost speeds?  My performance is just as good even though according to sensors I am not getting 350 Watt.  Are you saying if I got the full 350 Watt my boost will then go up to maybe 2100?  I mean if that is the case then yeah I want my full 350 Watt LOL
 
 
 
 

Just received a new bios for the 3080 Ti XC3 Hydro Copper. Will do some testing shortly.

KingEngineRevUp

kram36
Just received a new bios for the 3080 Ti XC3 Hydro Copper. Will do some testing shortly.

Good luck, I hope it's good news!

Have a bunch of stuff to do today. Hope to get to it by the time EVGA closes today to get the info back to EVGATech_LeeM.

kongfra

KingEngineRevUp

dalab
Thanks to you guys keeping this alive. I stumbled across this thread and ran Timespy Extreme with GPUz and yeah, I hit 352w for about a half a second in the entire thing. most of the time the card sits at 315-325w. Its the hybrid, so the pump is taking a little, but I have the fans connected to a controller. RGB off.
 
Let me know if theres anything you guys want me to run tonight and I'll provide it to help the cause.

If you can, do timespy graphic test 2, log it in GPU-Z and send the .txt. logs. I think it's good we keep comparing peoples power draws and their average voltages and any other correlations we can find. 
 
I think Speedy is the only one here that averages 345W of power draw but his average voltages aren't that impressive compared to the Ventus we had tested from another user. 
 
Like Grscrxsi was saying, it can be inefficiencies and leaks. But it's kind of crazy for a card to leak that much, where speed draws 30-50W more on average than Kram and Grscrxsi with no real benefit to show for it. Not like the Ventus we saw. 
 
It can just be... The Ventus is just a better designed board compared to the XC3. Now obviously we have a poor pool of data, just one Ventus tested and a few XC3s. 
 
But I think we have had enough XC3s tested here to see a correlation, they have ranges of power draws, have similar average clocks in PL loads and similar average voltages. They don't perform as well as a 3080 Ti FE, which is sad. 

 
Aside from not seeing full power , is that what is really causing slow boost speeds?
 
Is it really that bad compared to the FE, here is one review:
 
https://www.thefpsreview.com/2021/06/24/overclocking-nvidia-geforce-rtx-3080-ti-founders-edition/
 
His average boost is 1804 mhz before overclocking.  In games the lowest I have seen is 1845 mhz boost and I am at stock but I have seen it boost constantly as high as 1950 mhz especially in games like call of duty cold war it pegs at 1950 from what I recall but can run another test later tonight as well. In A Plague Tale: Innocence I get constant 1950 and that only at watts of 330-338.  Red dead redemption 2 is constant 1845 mhz for me I noticed that last night and never saw it dip at all.   What i don't really understand enough about the technology why the boost changes per game.  Like why is call of duty my highest boost, and Red Dead redemption 2 is only 1845?  I would think it would be the same all the time if my thermals are under 70 which they are,
 
 
I haven't even tried any overclocking not sure I will bother.  I am not saying people here are not having issues, obviously they are and some are much worst then others, but can we outright say all FE outdo XC3?  Maybe the FE cards that reviewers get are the best ones sent from nvidia?  Maybe it just has to do with silicon lottery?  just like CPU silicon lottery.  As far as I know they are only rated for 1625 MHZ so anything above that is extra right?  I just wonder if power has to do with performance if my power in Plague Tale is only 300-320 with average of 307, but I am getting 1905-1920 boost speeds?  My performance is just as good even though according to sensors I am not getting 350 Watt.  Are you saying if I got the full 350 Watt my boost will then go up to maybe 2100?  I mean if that is the case then yeah I want my full 350 Watt LOL
 
 
 
 

I think it's both factors, the card not being able to pull more power and silicone lottery. I feel like for my card at least, I lost the silicone lottery as my card struggles in metro Exodus EE to hit its rated 1665mhz boost clock, and even in Cyberpunk I can barely maintain 1700 mhz. My card also sits around 70C, so temperatures are definitely not an issue. All my testing was done at 1440p ultrawide (3440x1440p) so the card is definitely being pushed more than regular 1440p but not as hard as 4k. 

kongfra

KingEngineRevUp

kongfra

KingEngineRevUp

dalab
Thanks to you guys keeping this alive. I stumbled across this thread and ran Timespy Extreme with GPUz and yeah, I hit 352w for about a half a second in the entire thing. most of the time the card sits at 315-325w. Its the hybrid, so the pump is taking a little, but I have the fans connected to a controller. RGB off.
 
Let me know if theres anything you guys want me to run tonight and I'll provide it to help the cause.

If you can, do timespy graphic test 2, log it in GPU-Z and send the .txt. logs. I think it's good we keep comparing peoples power draws and their average voltages and any other correlations we can find. 
 
I think Speedy is the only one here that averages 345W of power draw but his average voltages aren't that impressive compared to the Ventus we had tested from another user. 
 
Like Grscrxsi was saying, it can be inefficiencies and leaks. But it's kind of crazy for a card to leak that much, where speed draws 30-50W more on average than Kram and Grscrxsi with no real benefit to show for it. Not like the Ventus we saw. 
 
It can just be... The Ventus is just a better designed board compared to the XC3. Now obviously we have a poor pool of data, just one Ventus tested and a few XC3s. 
 
But I think we have had enough XC3s tested here to see a correlation, they have ranges of power draws, have similar average clocks in PL loads and similar average voltages. They don't perform as well as a 3080 Ti FE, which is sad. 

 
Aside from not seeing full power , is that what is really causing slow boost speeds?
 
Is it really that bad compared to the FE, here is one review:
 
https://www.thefpsreview.com/2021/06/24/overclocking-nvidia-geforce-rtx-3080-ti-founders-edition/
 
His average boost is 1804 mhz before overclocking.  In games the lowest I have seen is 1845 mhz boost and I am at stock but I have seen it boost constantly as high as 1950 mhz especially in games like call of duty cold war it pegs at 1950 from what I recall but can run another test later tonight as well. In A Plague Tale: Innocence I get constant 1950 and that only at watts of 330-338.  Red dead redemption 2 is constant 1845 mhz for me I noticed that last night and never saw it dip at all.   What i don't really understand enough about the technology why the boost changes per game.  Like why is call of duty my highest boost, and Red Dead redemption 2 is only 1845?  I would think it would be the same all the time if my thermals are under 70 which they are,
 
 
I haven't even tried any overclocking not sure I will bother.  I am not saying people here are not having issues, obviously they are and some are much worst then others, but can we outright say all FE outdo XC3?  Maybe the FE cards that reviewers get are the best ones sent from nvidia?  Maybe it just has to do with silicon lottery?  just like CPU silicon lottery.  As far as I know they are only rated for 1625 MHZ so anything above that is extra right?  I just wonder if power has to do with performance if my power in Plague Tale is only 300-320 with average of 307, but I am getting 1905-1920 boost speeds?  My performance is just as good even though according to sensors I am not getting 350 Watt.  Are you saying if I got the full 350 Watt my boost will then go up to maybe 2100?  I mean if that is the case then yeah I want my full 350 Watt LOL
 

Well lets put it this way. The 3080 Ti FE out of the box will run at TDP of 350W. With slider slid to 114%, that's now 400W. 
 
XC3 users don't really touch that 350W and if they do, it seems to not do anything. The XC3 is also supposed to be able to access 366W when the slider is slid to 105% but that doesn't really seem to do anything either. Their average clocks are stuck below 1800 MHz, even dropping to the 1600s sometimes. 
 
Take note that these are in situations where games need more than 300W of power draw.
 

• Assassin's creed Valhalla doesn't run the card more than 310W from what I have seen. All 3080 Ti will behave the same here.

• Horizon Zero Dawn pushes the card at 375W, XC3 users will feel performance drops here.

• Cyberpunk 2077, Resident Evil Village both with ray tracing on will consume more than 400W if it can. XC3 users will see a even bigger performance drop compared to higher TDP cards. 

 
i dont own cyberpunk, re village, or horizon dawn (played this on ps4)  , but maybe i will buy cyberpunk even though i heard the game sucks just to see what i get performance wise, I am curious to see what it looks like on a top end machine. 
 
 
Also just to add that is what I don't get this reviewer is gettign 350W out of the box but his average is still only 1804 mhz before overclocks which is worst then what I am getting so far from the games I tested, but I do need to test more.  

He is pro testi the card where you'd be power limited. That's the point of a good benchmark, to our your card in a situation it'll more than likely hit a limit.

You don't need cyberpunk, just do TimeSpy or port royal like everyone else has been doing here. It'll even report your average clocks in the run.

Look at gsrcrxsi results here https://www.3dmark.com/spy/21525507

His average clock is 1761 and that's with +120 overclock, if he didn't have the +120 then he'd probably average 1640 MHz.

KingEngineRevUp

kongfra

KingEngineRevUp

kongfra

KingEngineRevUp

dalab
Thanks to you guys keeping this alive. I stumbled across this thread and ran Timespy Extreme with GPUz and yeah, I hit 352w for about a half a second in the entire thing. most of the time the card sits at 315-325w. Its the hybrid, so the pump is taking a little, but I have the fans connected to a controller. RGB off.
 
Let me know if theres anything you guys want me to run tonight and I'll provide it to help the cause.

If you can, do timespy graphic test 2, log it in GPU-Z and send the .txt. logs. I think it's good we keep comparing peoples power draws and their average voltages and any other correlations we can find. 
 
I think Speedy is the only one here that averages 345W of power draw but his average voltages aren't that impressive compared to the Ventus we had tested from another user. 
 
Like Grscrxsi was saying, it can be inefficiencies and leaks. But it's kind of crazy for a card to leak that much, where speed draws 30-50W more on average than Kram and Grscrxsi with no real benefit to show for it. Not like the Ventus we saw. 
 
It can just be... The Ventus is just a better designed board compared to the XC3. Now obviously we have a poor pool of data, just one Ventus tested and a few XC3s. 
 
But I think we have had enough XC3s tested here to see a correlation, they have ranges of power draws, have similar average clocks in PL loads and similar average voltages. They don't perform as well as a 3080 Ti FE, which is sad. 

 
Aside from not seeing full power , is that what is really causing slow boost speeds?
 
Is it really that bad compared to the FE, here is one review:
 
https://www.thefpsreview.com/2021/06/24/overclocking-nvidia-geforce-rtx-3080-ti-founders-edition/
 
His average boost is 1804 mhz before overclocking.  In games the lowest I have seen is 1845 mhz boost and I am at stock but I have seen it boost constantly as high as 1950 mhz especially in games like call of duty cold war it pegs at 1950 from what I recall but can run another test later tonight as well. In A Plague Tale: Innocence I get constant 1950 and that only at watts of 330-338.  Red dead redemption 2 is constant 1845 mhz for me I noticed that last night and never saw it dip at all.   What i don't really understand enough about the technology why the boost changes per game.  Like why is call of duty my highest boost, and Red Dead redemption 2 is only 1845?  I would think it would be the same all the time if my thermals are under 70 which they are,
 
 
I haven't even tried any overclocking not sure I will bother.  I am not saying people here are not having issues, obviously they are and some are much worst then others, but can we outright say all FE outdo XC3?  Maybe the FE cards that reviewers get are the best ones sent from nvidia?  Maybe it just has to do with silicon lottery?  just like CPU silicon lottery.  As far as I know they are only rated for 1625 MHZ so anything above that is extra right?  I just wonder if power has to do with performance if my power in Plague Tale is only 300-320 with average of 307, but I am getting 1905-1920 boost speeds?  My performance is just as good even though according to sensors I am not getting 350 Watt.  Are you saying if I got the full 350 Watt my boost will then go up to maybe 2100?  I mean if that is the case then yeah I want my full 350 Watt LOL
 

Well lets put it this way. The 3080 Ti FE out of the box will run at TDP of 350W. With slider slid to 114%, that's now 400W. 
 
XC3 users don't really touch that 350W and if they do, it seems to not do anything. The XC3 is also supposed to be able to access 366W when the slider is slid to 105% but that doesn't really seem to do anything either. Their average clocks are stuck below 1800 MHz, even dropping to the 1600s sometimes. 
 
Take note that these are in situations where games need more than 300W of power draw.
 

• Assassin's creed Valhalla doesn't run the card more than 310W from what I have seen. All 3080 Ti will behave the same here.

• Horizon Zero Dawn pushes the card at 375W, XC3 users will feel performance drops here.

• Cyberpunk 2077, Resident Evil Village both with ray tracing on will consume more than 400W if it can. XC3 users will see a even bigger performance drop compared to higher TDP cards. 

 
i dont own cyberpunk, re village, or horizon dawn (played this on ps4)  , but maybe i will buy cyberpunk even though i heard the game sucks just to see what i get performance wise, I am curious to see what it looks like on a top end machine. 
 
 
Also just to add that is what I don't get this reviewer is gettign 350W out of the box but his average is still only 1804 mhz before overclocks which is worst then what I am getting so far from the games I tested, but I do need to test more.  

He is pro testi the card where you'd be power limited. That's the point of a good benchmark, to our your card in a situation it'll more than likely hit a limit.

You don't need cyberpunk, just do TimeSpy or port royal like everyone else has been doing here. It'll even report your average clocks in the run.

Look at gsrcrxsi results here https://www.3dmark.com/spy/21525507

His average clock is 1761 and that's with +120 overclock, if he didn't have the +120 then he'd probably average 1640 MHz.

 
I will run more tests later tonight and post

gsrcrxsi
I also received the second beta BIOS. I think this one will be a winner for most people.
 
Long story short, I'm seeing full power limit reached (~366W), and I don't think it's a coincidence that I'm seeing up to ~75W PCIe slot power now instead of being capped at 66W. Seeing higher clock speeds and voltages than before, which again supports my comments earlier that voltage and clocks were being clamped by the artificial power limiting at the PCIe slot.
 
Timespy Compares:
https://www.3dmark.com/compare/spy/21819943/spy/21898806
 
Port Royal Compares:
https://www.3dmark.com/compare/pr/1148574/pr/1154079
 
GPUz logs and other stuff:
https://drive.google.com/file/d/1ATyn1SCwISDkOu6xgZCWnrGzaR9vA-CB/view?usp=sharing
 
 

This is freaking awesome. Big win for Kram and you not relenting on this.

We need a FTW3 fix then, because we're not hitting a PCI-E power above 55-60 on average.