Question RTX 4090 Green screen issue, Windows 11, Happens on deskop idle with nothing open AND under load

shellcool

Distinguished
Jul 24, 2012
33
0
18,540
GPU: ASUS Strix RTX 4090 OC
CPU: AMD 7800X3D
Motherboard: Asus Strix 670E-A
RAM: Corsair CMH32GX5M2B6000Z30K (ram isn't on my motherboard recommended list is the only issue with my build whoops)
PSU: Corsair HX850i
OS: Windows 11 Pro
GPU Driver: GeForce 536.23

ISSUE:
When using my PC randomly the screen will go green and the PC will either sit there for 10 seconds and recover OR the screen goes green for a random amount of time, sometimes about 1 minute, then reboots.

I'm running a brand new RTX4090 and 5800X3D system, The PC even crashes on the desktop idling and doing nothing, if anything it crashes more when it's completly idle on the desktop or doing something tame, like watching youtube.

I decided to check event viewer and found loads of errors every time it crashes, I've listed them at the bottom of this post (only the ones that seem to be relevant)

Within event viewer (Error 3 at the bottom of the post) I get linked to this site by NVIDIA https://nvidia.custhelp.com/app/answers/detail/a_id/3633
The fix Nvidia recommend is

"Increase the timeout value to allow more time to complete more intensive rendering. Microsoft provides information on how to modify the Windows Registry to achieve this."

Has anyone ever gone about doing this before? does it actually fix anything? (found this guide how to do it https://www.drivereasy.com/knowledge/a-tdr-has-been-detected-nvidia-opengl-driver-error-solved/#m2 )
but I'm going to hold of doing this until some of you give me some suggestions

Could this be because of my PCIE Riser Cable? I sadly can't fit my 4090 in my PC without the riser cable so I have no way of attaching it directly to the motherboard without one, however back at my parents I do have my old PCIE 3.0 riser cable which I could try I suppose.

I'm debating spending my entire weekend setting up windows again fresh, It's so much effort to rip my PC off the wall and then apart to do anything though, Thermal ake core P3 ftw though lol.

Any advice would be greatly received as I'm stuck, I am a bit skeptical of trying what Nvidia said in error #3
The GPU seemed fine in my Old setup (plopped it in my 6700k setup for a couple of days using the PCIE 3.0 riser, then when I built the new system with the PCIE 4.0 riser cable it seemed okay for about 2 weeks.
I can push the GPU really hard on Forza Horizon 5 max settings, native 4k everything switched on running at 120hz okay with some crashes and a lot more at 144hz but I thought this was down to a crappy HDMI 2.1 cable, but honestly it's so confusing now if the event viewer issues are true.

I found one forum post saying a Windows 10 update created this issue back in 2020 for someone and the only way to fix it was to use an older version of Windows 10 and disable Windows update (I'm on Windows 11)


I've tried:
When I initially installed Windows I installed it only with the boot drive in (someone said something about Windows installer using other drives for some install data or something weird)
DDU
reinstall Nvidia drivers
switching to the second bios switch on my GPU
updating my GPU's bios from ASUS's site
motherboard bios is the most recent version (1415)
Reseating GPU
Reseating riser cable
reseating RAM
Rtx 4090 firmware (it went from 1.0-1.2 or 1.1-1.2) (Thanks white.a.drew)

I haven't tried:
reseating CPU
reinstalling windows
a DDR5 ram kit on the QVL list for my motherboard




Error #1
in event viewer, around the time of crashes, I get nvlddmkm event ID 0
All the information it gives me is:

"The description for Event ID 0 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.
If the event originated on another computer, the display information had to be saved with the event.
The following information was included with the event:
\Device\Video28
Restarting TDR occurred on GPUID:100
The message resource is present but the message was not found in the message table"


Error #2
Event ID: 4101

"Display driver nvlddmkm stopped responding and has successfully recovered."

Error #3
Event ID: 1
"A TDR has been detected.
The application must close.

Error code: 7

(pid=21192 tid=26964 ld9boxheadless.exe 64bit)
Visit http://nvidia.custhelp.com/app/answers/detail/a_id/3633 for more information."
 
Last edited:
GPU: ASUS Strix RTX 4090 OC
CPU: AMD 7800X3D
Motherboard: Asus Strix 670E-A
RAM: Corsair CMH32GX5M2B6000Z30K (ram isn't on my motherboard recommended list is the only issue with my build whoops)
PSU: Corsair HX850i
OS: Windows 11 Pro
GPU Driver: GeForce 536.23

ISSUE:
When using my PC randomly the screen will go green and the PC will either sit there for 10 seconds and recover OR the screen goes green for a random amount of time, sometimes about 1 minute, then reboots.

I'm running a brand new RTX4090 and 5800X3D system, The PC even crashes on the desktop idling and doing nothing, if anything it crashes more when it's completly idle on the desktop or doing something tame, like watching youtube.

I decided to check event viewer and found loads of errors every time it crashes, I've listed them at the bottom of this post (only the ones that seem to be relevant)

Within event viewer (Error 3 at the bottom of the post) I get linked to this site by NVIDIA https://nvidia.custhelp.com/app/answers/detail/a_id/3633
The fix Nvidia recommend is

"Increase the timeout value to allow more time to complete more intensive rendering. Microsoft provides information on how to modify the Windows Registry to achieve this."

Has anyone ever gone about doing this before? does it actually fix anything? (found this guide how to do it https://www.drivereasy.com/knowledge/a-tdr-has-been-detected-nvidia-opengl-driver-error-solved/#m2 )
but I'm going to hold of doing this until some of you give me some suggestions

Could this be because of my PCIE Riser Cable? I sadly can't fit my 4090 in my PC without the riser cable so I have no way of attaching it directly to the motherboard without one, however back at my parents I do have my old PCIE 3.0 riser cable which I could try I suppose.

I'm debating spending my entire weekend setting up windows again fresh, It's so much effort to rip my PC off the wall and then apart to do anything though, Thermal ake core P3 ftw though lol.

Any advice would be greatly received as I'm stuck, I am a bit skeptical of trying what Nvidia said in error #3
The GPU seemed fine in my Old setup (plopped it in my 6700k setup for a couple of days using the PCIE 3.0 riser, then when I built the new system with the PCIE 4.0 riser cable it seemed okay for about 2 weeks.
I can push the GPU really hard on Forza Horizon 5 max settings, native 4k everything switched on running at 120hz okay with some crashes and a lot more at 144hz but I thought this was down to a crappy HDMI 2.1 cable, but honestly it's so confusing now if the event viewer issues are true.

I found one forum post saying a Windows 10 update created this issue back in 2020 for someone and the only way to fix it was to use an older version of Windows 10 and disable Windows update (I'm on Windows 11)


I've tried:
When I initially installed Windows I installed it only with the boot drive in (someone said something about Windows installer using other drives for some install data or something weird)
DDU
reinstall Nvidia drivers
switching to the second bios switch on my GPU
updating my GPU's bios from ASUS's site
motherboard bios is the most recent version (1415)
Reseating GPU
Reseating riser cable
reseating RAM

I haven't tried:
reseating CPU
reinstalling windows
a DDR5 ram kit on the QVL list for my motherboard




Error #1
in event viewer, around the time of crashes, I get nvlddmkm event ID 0
All the information it gives me is:

"The description for Event ID 0 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.
If the event originated on another computer, the display information had to be saved with the event.
The following information was included with the event:
\Device\Video28
Restarting TDR occurred on GPUID:100
The message resource is present but the message was not found in the message table"


Error #2
Event ID: 4101

"Display driver nvlddmkm stopped responding and has successfully recovered."

Error #3
Event ID: 1
"A TDR has been detected.
The application must close.

Error code: 7

(pid=21192 tid=26964 ld9boxheadless.exe 64bit)
Visit http://nvidia.custhelp.com/app/answers/detail/a_id/3633 for more information."
It maybe the 4090 firmware have you tried flashing it?
 
GPU: ASUS Strix RTX 4090 OC
CPU: AMD 7800X3D
Motherboard: Asus Strix 670E-A
RAM: Corsair CMH32GX5M2B6000Z30K (ram isn't on my motherboard recommended list is the only issue with my build whoops)
PSU: Corsair HX850i
OS: Windows 11 Pro
GPU Driver: GeForce 536.23

ISSUE:
When using my PC randomly the screen will go green and the PC will either sit there for 10 seconds and recover OR the screen goes green for a random amount of time, sometimes about 1 minute, then reboots.

I'm running a brand new RTX4090 and 5800X3D system, The PC even crashes on the desktop idling and doing nothing, if anything it crashes more when it's completly idle on the desktop or doing something tame, like watching youtube.

I decided to check event viewer and found loads of errors every time it crashes, I've listed them at the bottom of this post (only the ones that seem to be relevant)

Within event viewer (Error 3 at the bottom of the post) I get linked to this site by NVIDIA https://nvidia.custhelp.com/app/answers/detail/a_id/3633
The fix Nvidia recommend is

"Increase the timeout value to allow more time to complete more intensive rendering. Microsoft provides information on how to modify the Windows Registry to achieve this."

Has anyone ever gone about doing this before? does it actually fix anything? (found this guide how to do it https://www.drivereasy.com/knowledge/a-tdr-has-been-detected-nvidia-opengl-driver-error-solved/#m2 )
but I'm going to hold of doing this until some of you give me some suggestions

Could this be because of my PCIE Riser Cable? I sadly can't fit my 4090 in my PC without the riser cable so I have no way of attaching it directly to the motherboard without one, however back at my parents I do have my old PCIE 3.0 riser cable which I could try I suppose.

I'm debating spending my entire weekend setting up windows again fresh, It's so much effort to rip my PC off the wall and then apart to do anything though, Thermal ake core P3 ftw though lol.

Any advice would be greatly received as I'm stuck, I am a bit skeptical of trying what Nvidia said in error #3
The GPU seemed fine in my Old setup (plopped it in my 6700k setup for a couple of days using the PCIE 3.0 riser, then when I built the new system with the PCIE 4.0 riser cable it seemed okay for about 2 weeks.
I can push the GPU really hard on Forza Horizon 5 max settings, native 4k everything switched on running at 120hz okay with some crashes and a lot more at 144hz but I thought this was down to a crappy HDMI 2.1 cable, but honestly it's so confusing now if the event viewer issues are true.

I found one forum post saying a Windows 10 update created this issue back in 2020 for someone and the only way to fix it was to use an older version of Windows 10 and disable Windows update (I'm on Windows 11)


I've tried:
When I initially installed Windows I installed it only with the boot drive in (someone said something about Windows installer using other drives for some install data or something weird)
DDU
reinstall Nvidia drivers
switching to the second bios switch on my GPU
updating my GPU's bios from ASUS's site
motherboard bios is the most recent version (1415)
Reseating GPU
Reseating riser cable
reseating RAM

I haven't tried:
reseating CPU
reinstalling windows
a DDR5 ram kit on the QVL list for my motherboard




Error #1
in event viewer, around the time of crashes, I get nvlddmkm event ID 0
All the information it gives me is:

"The description for Event ID 0 from source nvlddmkm cannot be found. Either the component that raises this event is not installed on your local computer or the installation is corrupted. You can install or repair the component on the local computer.
If the event originated on another computer, the display information had to be saved with the event.
The following information was included with the event:
\Device\Video28
Restarting TDR occurred on GPUID:100
The message resource is present but the message was not found in the message table"


Error #2
Event ID: 4101

"Display driver nvlddmkm stopped responding and has successfully recovered."

Error #3
Event ID: 1
"A TDR has been detected.
The application must close.

Error code: 7

(pid=21192 tid=26964 ld9boxheadless.exe 64bit)
Visit http://nvidia.custhelp.com/app/answers/detail/a_id/3633 for more information."
Rtx firmware/uefi update https://www.nvidia.com/content/Driv...ter_1.2-x64.exe&firmware=1&lang=us&type=Other
 
Hey cheers for replying, I did this yesterday, but forgot to add it to my list of things tried, I think i'm going to have to shove my PCIE 3.0 riser back in and see if the issue disappears for now then 🙁
If the 3.0 riser makes a difference and running the card in 3.0 works then I would almost bet the card is faulty or the mobo isn't transferring the 4.0 data properly causing drop outs in communication with the gpu and mobo
 
If the 3.0 riser makes a difference and running the card in 3.0 works then I would almost bet the card is faulty or the mobo isn't transferring the 4.0 data properly causing drop outs in communication with the gpu and mobo
Had a quick and simple check, I know it's a 3.0 device but I've shoved my GTX 1070 into the system and not changed anything else, going to see if the cable is fine as it is, if this works, i'll go into the BIOS and see if i can set the PCIE slot to run at 3.0 speeds, if that fixes it then the riser isn't properly running at the PCIE 4.0 speeds, I have owned 3 PCIE 3.0 risers since 2017, so they are delicate/randomly fail

My PC has literally always been bolted to a brick wall and never touched, I find it funny stuff still breaks lol
 
Had a quick and simple check, I know it's a 3.0 device but I've shoved my GTX 1070 into the system and not changed anything else, going to see if the cable is fine as it is, if this works, i'll go into the BIOS and see if i can set the PCIE slot to run at 3.0 speeds, if that fixes it then the riser isn't properly running at the PCIE 4.0 speeds, I have owned 3 PCIE 3.0 risers since 2017, so they are delicate/randomly fail

My PC has literally always been bolted to a brick wall and never touched, I find it funny stuff still breaks lol
It's because you're buying 3.0 risers for a 4.0 card that would be your issue
 
It's because you're buying 3.0 risers for a 4.0 card that would be your issue
No I own a brand new 4.0 riser that i bought with the 4090

I put the GTX 1070 in place of the 4090 on the 4.0 riser and the system ran fine for 5 hours, just put the 4090 back in the 4.0 riser and it's flickering green/freezing green again, so next fault finding step im putting my PCIE slot in 3.0 mode in bios, see what happens with the 4090


(yes i do own a 3.0 riser, it's currently in a box next to the PC)
 
No I own a brand new 4.0 riser that i bought with the 4090

I put the GTX 1070 in place of the 4090 on the 4.0 riser and the system ran fine for 5 hours, just put the 4090 back in the 4.0 riser and it's flickering green/freezing green again, so next fault finding step im putting my PCIE slot in 3.0 mode in bios, see what happens with the 4090


(yes i do own a 3.0 riser, it's currently in a box next to the PC)
It sounds like you found the problem either the mobo can't handle 4.0 data properly or the riser is faulty what brand is the riser.... and please don't say thermaltake
 
It sounds like you found the problem either the mobo can't handle 4.0 data properly or the riser is faulty what brand is the riser.... and please don't say thermaltake
Sadly I have to use Thermaltake risers with my Thermaltake case, I bought an AsusROG one in 2020 and it didn't fit, had to return and buy a Thermaltake one, they're great once you get one that works they stay working (for the most part, just don't touch the bloody things)
 
Sadly I have to use Thermaltake risers with my Thermaltake case, I bought an AsusROG one in 2020 and it didn't fit, had to return and buy a Thermaltake one, they're great once you get one that works they stay working (for the most part, just don't touch the bloody things)
Thermaltake riser are basically the equivalent to buying the cheapest generic riser on the market they don't know how to make them right I don't understand how they have such a big problem producing a quality one but they are garbage straight out sorry to say
 
Just commenting to say I put the GPU directly in the motherboard and it still has all the same issues.

I even RMA'd the GPU and they ran it fine for 70+ hours with no issues so they sent it back, I got it and now it crashes even more AND in TWO of my completely different PC's.

I'm clueless...
 
What are the psus your using in all the systems
I've tried a 1000watt be quiet gold rated unit which is brand new out the box with about 5 hours run time

And I also had the graphics card run fine for a few weeks with my 850 watt Corsair HX850i

I also plugged just the GPU into the 1000 watt unit and powered the rest of the system from the 850watt unit, i reallllllly disagree that it's a power related issue seeming the system can run games for 5 hours fine then sit on desktop for 30 mins while I watch TV and just suddenly have this GPU driver crash
 
I've tried a 1000watt be quiet gold rated unit which is brand new out the box with about 5 hours run time

And I also had the graphics card run fine for a few weeks with my 850 watt Corsair HX850i

I also plugged just the GPU into the 1000 watt unit and powered the rest of the system from the 850watt unit, i reallllllly disagree that it's a power related issue seeming the system can run games for 5 hours fine then sit on desktop for 30 mins while I watch TV and just suddenly have this GPU driver crash
Yeah no I agree there just wanted to double check its 100% not a psu issue. At this point I would presume the one thing left if I didn't miss it the mobo would be the last link to issues. Not liking timings somewhere
 
I've tried a 1000watt be quiet gold rated unit which is brand new out the box with about 5 hours run time

And I also had the graphics card run fine for a few weeks with my 850 watt Corsair HX850i

I also plugged just the GPU into the 1000 watt unit and powered the rest of the system from the 850watt unit, i reallllllly disagree that it's a power related issue seeming the system can run games for 5 hours fine then sit on desktop for 30 mins while I watch TV and just suddenly have this GPU driver crash
This could be your system trying to over over clock the gpu causing the crash... or the ram not liking the coms between it and the gpu, or the cpu not talking right with the pcie bus
 
This could be your system trying to over over clock the gpu causing the crash... or the ram not liking the coms between it and the gpu, or the cpu not talking right with the pcie bus
I had the PC running 100% fine with a GTX1070 for 10 days or more while the 4090 was with Scan.co.uk technicians.

My ram isn't on the QVL so I may need to try returning that and swapping it out for a kit that is on the QVL
but is there any reason that ram would work fine with a 1070 but not a 4090?

I'm under the same assumption as you, something today with the PCIE bus because the 4090 would be running PCIE 4.0, however, I did lock the PCIE bus at 3.0 speeds in bios for the 4090 and still get the same crash, I could try PCIE 2.0 or 1.0 as a trial but I might be better off trying the second PCIE slot on my MB, which is PCIE 4.0 x16 so it's still full speed.

Do you know if having it in this different PCIE slot would be using a completely different PCIE lane within the CPU? assuming something is wrong with the physical PCIE slot or the PCIE lanes on the CPU that somehow doesn't effect the GTX1070 even though that should be using all 16 lanes, albeit at a much slower speed (PCIE gen3 not 4)