Question What could cause RTX 4080 to not use full power in games?

Bullszeye

Distinguished
Mar 7, 2013
40
2
18,535
Hey all, I have an issue with my less than half a year old GPU. Using DDU to wipe drivers and reinstalling makes it work until the next time I restart my PC. The issue is that in games the GPU uses less than 60% TDP depending on the game, on some games even less than 10% if they're less graphically demanding. When I try Furmark, I get a higher score than the average 4080 so I can't RMA it due to it performing as expected on benchmarks. This issue is very weird, I literally tried everything and am out of ideas... Thanks!

GPU model is ROG-STRIX-RTX4080-O16G-GAMING



https://ibb.co/XXyZWTW - afterburner from a game where i get terrible performance
https://ibb.co/Kx8LJpH - afterburner after wiping drivers with DDU and reinstalling, issue happens again on the next restart

GPU-Z:
https://ibb.co/jkHp2sJ - with issue
https://ibb.co/Y26vqhF - without issue

Interesting thing is, when it is running in x1 3.0, furmark doesnt kick the value up, but the GPU scores as it should anyways.

Full pc specs:
windows 11
coolermaster h500p mesh case
asus rog strix z690-e
i9 12900k - stock
NZXT C850P
32 gb ddr5 6000 mhz ram
980 pro 1 tb in pcie 5.0 slot
kingston ssd 1 tb
860 evo 512 gb
two WD hdds, one 4TB other 8 TB
asus rog strix rtx 4080 OC edition


What I've concluded and tried so far after much research:
The gpu sets itself to 3.0 1x for some reason, pressing question mark doesnt help. Going into safe mode -> DDU -> restart -> reinstall drivers makes the GPU work properly. On the second restart after this, gpu sets itself back to 1.0x
GPU is in top slot, it is in 8x (When working properly) due to sharing bandwidth with an M2. NVIDIA settings is in maximum performance mode.
Tried GPU bios update, MOBO bios update.
Mobo settings are all set correctly.
I've checked device manager for PCI bus driver and it had an AMD driver there. I uninstalled that, and windows replaced it with PCI Express Root Complex. It did nothing, issue is still there.
Installing mobo chipset again, did nothing.

Someone asked me to check if its in MSI or IRQ mode, i dont really know how to check that, but according to this:
"Go to Device Manager. Click in menu "View -> Resources by type". Expand "Interrupt request (IRQ)" node of the tree. Scroll down to "(PCI) 0x... (...) device name" device nodes. Devices with positive number for IRQ (like "(PCI) 0x00000011 (17) ...") are in Line-based interrupts-mode. Devices with negative number for IRQ (like "(PCI) 0xFFFFFFFA (-6) ...") are in Message Signaled-based Interrupts-mode. " - my gpu is in (16) so im guessing msi mode is not on?

Thanks!
 

Bullszeye

Distinguished
Mar 7, 2013
40
2
18,535
See first reply it mentions your board.


Try use the second m2 slot, leave top m2 vacant and see if that helps.
I can try doing this later as I have to buy specific screwdrivers to take the m2 cover out, however from what i understand running my gpu at 4.0 8x should be just marginally worse than 16x, but that is not the issue i have. My issue is that the gpu is running at 3.0 1x for some odd reason. using DDU to reinstall drivers sets it back to 4.0 8x, but after another restart its back at 1x...
 

boju

Titan
Ambassador
My issue is that the gpu is running at 3.0 1x for some odd reason. using DDU to reinstall drivers sets it back to 4.0 8x, but after another restart its back at 1x...

My reply was implying regarding that, troubleshooting if this is a bug regarding top m2 slot with a different gen ssd. Maybe, maybe not. We'll see if your graphics card still drops to 1x after moving ssd.
 
  • Like
Reactions: Bullszeye

Bullszeye

Distinguished
Mar 7, 2013
40
2
18,535
My reply was implying regarding that, troubleshooting if this is a bug regarding top m2 slot with a different gen ssd. Maybe, maybe not. We'll see if your graphics card still drops to 1x after moving ssd.
Oh, okay, I'll get back to you when i do that. I did some testing while looking at hwinfo. These are my findings.
Gpu will randomly run at 8x 4.0 or 1x 4.0 or 1x 3.0 after each restart.
Screenshots:
https://ibb.co/g7ZDyPd - running properly
https://ibb.co/LJj94Qd - stuck at 1x 4.0
 
Last edited:
Jul 7, 2022
539
521
1,760
Hey all, I have an issue with my less than half a year old GPU. Using DDU to wipe drivers and reinstalling makes it work until the next time I restart my PC. The issue is that in games the GPU uses less than 60% TDP depending on the game, on some games even less than 10% if they're less graphically demanding. When I try Furmark, I get a higher score than the average 4080 so I can't RMA it due to it performing as expected on benchmarks. This issue is very weird, I literally tried everything and am out of ideas... Thanks!

GPU model is ROG-STRIX-RTX4080-O16G-GAMING



https://ibb.co/XXyZWTW - afterburner from a game where i get terrible performance
https://ibb.co/Kx8LJpH - afterburner after wiping drivers with DDU and reinstalling, issue happens again on the next restart

GPU-Z:
https://ibb.co/jkHp2sJ - with issue
https://ibb.co/Y26vqhF - without issue

Interesting thing is, when it is running in x1 3.0, furmark doesnt kick the value up, but the GPU scores as it should anyways.

Full pc specs:
windows 11
coolermaster h500p mesh case
asus rog strix z690-e
i9 12900k - stock
NZXT C850P
32 gb ddr5 6000 mhz ram
980 pro 1 tb in pcie 5.0 slot
kingston ssd 1 tb
860 evo 512 gb
two WD hdds, one 4TB other 8 TB
asus rog strix rtx 4080 OC edition


What I've concluded and tried so far after much research:
The gpu sets itself to 3.0 1x for some reason, pressing question mark doesnt help. Going into safe mode -> DDU -> restart -> reinstall drivers makes the GPU work properly. On the second restart after this, gpu sets itself back to 1.0x
GPU is in top slot, it is in 8x (When working properly) due to sharing bandwidth with an M2. NVIDIA settings is in maximum performance mode.
Tried GPU bios update, MOBO bios update.
Mobo settings are all set correctly.
I've checked device manager for PCI bus driver and it had an AMD driver there. I uninstalled that, and windows replaced it with PCI Express Root Complex. It did nothing, issue is still there.
Installing mobo chipset again, did nothing.

Someone asked me to check if its in MSI or IRQ mode, i dont really know how to check that, but according to this:
"Go to Device Manager. Click in menu "View -> Resources by type". Expand "Interrupt request (IRQ)" node of the tree. Scroll down to "(PCI) 0x... (...) device name" device nodes. Devices with positive number for IRQ (like "(PCI) 0x00000011 (17) ...") are in Line-based interrupts-mode. Devices with negative number for IRQ (like "(PCI) 0xFFFFFFFA (-6) ...") are in Message Signaled-based Interrupts-mode. " - my gpu is in (16) so im guessing msi mode is not on?

Thanks!
Sounds like a CPU bottleneck. What game are you playing?
 

boju

Titan
Ambassador
Oh, okay, I'll get back to you when i do that. I did some testing while looking at hwinfo. These are my findings.
Gpu will randomly run at 8x 4.0 or 1x 4.0 or 1x 3.0 after each restart.
Screenshots:
https://ibb.co/g7ZDyPd - running properly
https://ibb.co/LJj94Qd - stuck at 1x 4.0

Yeah i see it. No idea why it does that. Searching about it mentions Ms game pass, don't think that fits your issue lol. Windows was freshly installed on this system yeah? Just checking.

Hopefully second m2 slot does something positive until you find out exactly why or maybe future bios update might address.

Sounds like a CPU bottleneck. What game are you playing?

Doubt that.
 
Jul 7, 2022
539
521
1,760
Hey all, I have an issue with my less than half a year old GPU. Using DDU to wipe drivers and reinstalling makes it work until the next time I restart my PC. The issue is that in games the GPU uses less than 60% TDP depending on the game, on some games even less than 10% if they're less graphically demanding. When I try Furmark, I get a higher score than the average 4080 so I can't RMA it due to it performing as expected on benchmarks. This issue is very weird, I literally tried everything and am out of ideas... Thanks!

GPU model is ROG-STRIX-RTX4080-O16G-GAMING



https://ibb.co/XXyZWTW - afterburner from a game where i get terrible performance
https://ibb.co/Kx8LJpH - afterburner after wiping drivers with DDU and reinstalling, issue happens again on the next restart

GPU-Z:
https://ibb.co/jkHp2sJ - with issue
https://ibb.co/Y26vqhF - without issue

Interesting thing is, when it is running in x1 3.0, furmark doesnt kick the value up, but the GPU scores as it should anyways.

Full pc specs:
windows 11
coolermaster h500p mesh case
asus rog strix z690-e
i9 12900k - stock
NZXT C850P
32 gb ddr5 6000 mhz ram
980 pro 1 tb in pcie 5.0 slot
kingston ssd 1 tb
860 evo 512 gb
two WD hdds, one 4TB other 8 TB
asus rog strix rtx 4080 OC edition


What I've concluded and tried so far after much research:
The gpu sets itself to 3.0 1x for some reason, pressing question mark doesnt help. Going into safe mode -> DDU -> restart -> reinstall drivers makes the GPU work properly. On the second restart after this, gpu sets itself back to 1.0x
GPU is in top slot, it is in 8x (When working properly) due to sharing bandwidth with an M2. NVIDIA settings is in maximum performance mode.
Tried GPU bios update, MOBO bios update.
Mobo settings are all set correctly.
I've checked device manager for PCI bus driver and it had an AMD driver there. I uninstalled that, and windows replaced it with PCI Express Root Complex. It did nothing, issue is still there.
Installing mobo chipset again, did nothing.

Someone asked me to check if its in MSI or IRQ mode, i dont really know how to check that, but according to this:
"Go to Device Manager. Click in menu "View -> Resources by type". Expand "Interrupt request (IRQ)" node of the tree. Scroll down to "(PCI) 0x... (...) device name" device nodes. Devices with positive number for IRQ (like "(PCI) 0x00000011 (17) ...") are in Line-based interrupts-mode. Devices with negative number for IRQ (like "(PCI) 0xFFFFFFFA (-6) ...") are in Message Signaled-based Interrupts-mode. " - my gpu is in (16) so im guessing msi mode is not on?

Thanks!
Have you tried backing up your C: drive then reinstalling a fresh copy of windows 11? I always do this when making a big hardware change like a graphics card. If the other answers these gentle people offer do not work, I suggest starting again with a fresh windows install.

BTW: for a quick and cheap NVME screwdriver, on your way home from work, stop at a Walgreens or CVS pharmacy and grab the cheapo eyeglasses screwdriver set they usually have around the checkout area.
 

boju

Titan
Ambassador
Have you tried backing up your C: drive then reinstalling a fresh copy of windows 11? I always do this when making a big hardware change like a graphics card.

Different take on things but such extremes shouldn't be necessary. Motherboard change then yeah, you would reinstall Windows. Different chipsets, registry entries, drivers etc, wouldn't play nice together. DDU should be sufficient in most cases changing GPU brands but then again you could be on to something, if have the patience.
 

Bullszeye

Distinguished
Mar 7, 2013
40
2
18,535
Yeah i see it. No idea why it does that. Searching about it mentions Ms game pass, don't think that fits your issue lol. Windows was freshly installed on this system yeah? Just checking.

Hopefully second m2 slot does something positive until you find out exactly why or maybe future bios update might address.



Doubt that.
After much pain and suffering due to the behemoth gpu taking so much space I finally managed to switch them. Interestingly, my pc boots faster now and the issue was fixed!! I've been trying to reproduce it for 5 restarts now and it works! Thanks so much man! I never would have came to the idea to do something like this!

Also as far as the fresh install goes... ummm.... this windows installation is one that was cloned from 2 previous SSDs and is now on its third completely new configuration. I will definitely do a fresh reinstall eventually, but the most important thing is that the issue at hand was fixed! Thanks!
 
Jul 7, 2022
539
521
1,760
Different take on things but such extremes shouldn't be necessary. Motherboard change then yeah, you would reinstall Windows. Different chipsets, registry entries, drivers etc, wouldn't play nice together. DDU should be sufficient in most cases changing GPU brands but then again you could be on to something, if have the patience.
Like I said, if your and others potential fixes don’t work, the ultimate fix is fresh install.

I’ve been building computers since I was 8 years old and I’ve always done a file backup then fresh install of windows whenever I upgrade a major part (gpu, cpu, motherboard) and I have never had any issues in 25 years. It might be overboard but it’s nice knowing there are no gremlins in the system sort of speak