2080 Ti problems

[tylrtexs]

Status: UNRESOLVED

Computer Type: Desktop - Custom Build

GPU: ZOTAC 2080 Ti AMP - 11 GB. Small overclock, 50 to core and 500 to memory. (tried disabling clock and still issues persist)

CPU: I7 9700k Overclocked to 5 Mhz

Motherboard: MSI MPG Z390 Gaming Edge AC LGA1151 latest Bios A.60 8/8/2019

RAM: Kingston Technology HyperX Fury White 16GB 2666MHz DDR4 CL16 DIMM Kit of 2 1Rx8 (HX426C16FW2K2/16) XMP Enabled, no overclock.

PSU: Seasonic 1050W ATX12V/EPS12V 80 Plus Platinum Certified Full Modular Active PFC Snow SILENT-1050

Operating System & Version: Windows 10 clean install 18363.900

GPU Drivers: Nvidia driver 451.48, clean install.

Description of Problem: I have noticed that my GPU has been throttling itself. While playing a very GPU intensive game (ie warzone), the GPU clock starts going up and down sporadically anywhere from 1300 Mhz up to 2k (where it should be.) The temps remain pretty solid around 70 Celsius. I have a liquid cooling loop with two radiators and 6 fans on each Rad. I'm also showing in GPU-Z that there is a perfcap Pwr. In EVGA precision x1, i have the target raised to 110% but it only ever reaches 60% power at most. This causes my games to have rubber band/stutter effect on the fps and its quite unbearable.

Troubleshooting: The only things ive done so far is clean install of drivers and making sure i have optimal performance selected in power options for nvidia control panel. I am honestly open to any ideas. i dont consider myself to be completely naive when it comes to tech but obviously im no expert. Thank you for reading, and responding. Have a great day!

[NelizMastr]

70c on a custom loop is pretty toasty on the GPU side, though, especially at near stock settings. Is the pump working properly? Did you set the pump header to max rather than PWM?

[Eighjan]

What was the previous graphics card...?

[tylrtexs]

2 minutes ago, NelizMastr said:

70c on a custom loop is pretty toasty on the GPU side, though, especially at near stock settings. Is the pump working properly? Did you set the pump header to max rather than PWM?

70c is like at max clock usually. its a 5 speed pump and i believe i have it on setting 2 of 5, 5 being the highest. 

 

2 minutes ago, Eighjan said:

What was the previous graphics card...?

this is the first gpu in this desktop. i built it for the 2080 ti. my old desktop just had a 1080.

[Eighjan]

2 minutes ago, tylrtexs said:

this is the first gpu in this desktop. i built it for the 2080 ti. my old desktop just had a 1080.

I only asked in case you may need to run DDU to remove any other graphics drivers (on-board/Intel?) & reinstall the Nvidia ones.

[tylrtexs]

2 minutes ago, Eighjan said:

I only asked in case you may need to run DDU to remove any other graphics drivers (on-board/Intel?) & reinstall the Nvidia ones.

ahhh gotcha. yeah i ran ddu and reinstalled drivers just to make sure that wasnt an easy fix.

[Eighjan]

1 minute ago, tylrtexs said:

ahhh gotcha. yeah i ran ddu and reinstalled drivers just to make sure that wasnt an easy fix.

You've tried reverting the 2080ti to stock speeds, I guess...?

... and run Windows Update to death...?

[tylrtexs]

Just now, Eighjan said:

You've tried reverting the 2080ti to stock speeds, I guess...?

yes i have

[Eighjan]

3 minutes ago, tylrtexs said:

yes i have

Have you told the BIOS to use Gen.3 PCI-e for graphics, rather than use Auto...?

Edited July 8, 2020 by Eighjan

[tylrtexs]

52 minutes ago, Eighjan said:

Have you told the BIOS to use Gen.3 PCI-e, rather than use Auto...?

yes i've tried auto and setting it to Gen.3. both have the same outcome.

[jones177]

The GPU is not the hottest chip on the PCB of a 2080 ti.  It is the vram and with it the temps can run away and cause issues. 

My FW3 Ultra 2080 ti had this problem before it was fixed with an update and the vram between the GPU and the VRMs would get up to 112c.  

Also the vram does not heat up and cool like the GPU it just gets hotter and hotter. if you can check your cards temps other than the GPU you should do so.

 

Also test your 2080 ti on a game or bench the loads up the 2080 ti. Lots of games don't even at 1440p and some don't even at 4k.

To give you an idea, if I run GTA 5 on ultra at 1440p I only get 36% GPU usage with the power limit at 38%. The only way to load up the GPU is to use the advanced settings but with that 40 frames are lost. 

 

To test my GPU I use Heaven bench to see if the GPU can reach its power target and maintain good clocks. After that I am not worried since I know that my CPU even at 5.1ghz is what is holding it back.

Here is what it looks like testing.

 

 

 

 

[tylrtexs]

32 minutes ago, jones177 said:

The GPU is not the hottest chip on the PCB of a 2080 ti.  It is the vram and with it the temps can run away and cause issues. 

My FW3 Ultra 2080 ti had this problem before it was fixed with an update and the vram between the GPU and the VRMs would get up to 112c.  

Also the vram does not heat up and cool like the GPU it just gets hotter and hotter. if you can check your cards temps other than the GPU you should do so.

 

Also test your 2080 ti on a game or bench the loads up the 2080 ti. Lots of games don't even at 1440p and some don't even at 4k.

To give you an idea, if I run GTA 5 on ultra at 1440p I only get 36% GPU usage with the power limit at 38%. The only way to load up the GPU is to use the advanced settings but with that 40 frames are lost. 

 

To test my GPU I use Heaven bench to see if the GPU can reach its power target and maintain good clocks. After that I am not worried since I know that my CPU even at 5.1ghz is what is holding it back.

Here is what it looks like testing.

 

 

 

 

Thank you so much for the detailed reply. I took off the original casing and fans and attached a water block and with that, all new heating transfer pads on the VRAMS. i'm not sure how i can test the temps of those modules though. So even when i run heaven benchmark, my cards target power still does not go above 60%, and thats with 1440p and high/ultra settings. it will work smoothly for the first few seconds and then starts doing that up and down clock thing. let me open it all and take a screen shot so you can see.

[tylrtexs]

42 minutes ago, jones177 said:

The GPU is not the hottest chip on the PCB of a 2080 ti.  It is the vram and with it the temps can run away and cause issues. 

My FW3 Ultra 2080 ti had this problem before it was fixed with an update and the vram between the GPU and the VRMs would get up to 112c.  

Also the vram does not heat up and cool like the GPU it just gets hotter and hotter. if you can check your cards temps other than the GPU you should do so.

 

Also test your 2080 ti on a game or bench the loads up the 2080 ti. Lots of games don't even at 1440p and some don't even at 4k.

To give you an idea, if I run GTA 5 on ultra at 1440p I only get 36% GPU usage with the power limit at 38%. The only way to load up the GPU is to use the advanced settings but with that 40 frames are lost. 

 

To test my GPU I use Heaven bench to see if the GPU can reach its power target and maintain good clocks. After that I am not worried since I know that my CPU even at 5.1ghz is what is holding it back.

Here is what it looks like testing.

 

 

 

 

https://streamable.com/zf27bb here is a video so you can see it jumping up and down.

[jones177]

2 hours ago, tylrtexs said:

https://streamable.com/zf27bb here is a video so you can see it jumping up and down.

The vram would not heat up that fast and when it causes a down clock it is smooth.

 

It looks like a CPU issue to me and since Heaven uses one thread/core to render you might want to see what your single core score is on Cinebench R15 or 20.

My i7 8086ks did at best 222 in R15 and my i7 8700k usually did around 205. You should be somewhere in between.  

 

[tylrtexs]

3 minutes ago, jones177 said:

The vram would not heat up that fast and when it causes a down clock it is smooth.

 

It looks like a CPU issue to me and since Heaven uses one thread/core to render you might want to see what your single core score is on Cinebench R15 or 20.

My i7 8086ks did at best 222 in R15 and my i7 8700k usually did around 205. You should be somewhere in between.  

 

ill run that and post the results after.

[tylrtexs]

 

28 minutes ago, jones177 said:

The vram would not heat up that fast and when it causes a down clock it is smooth.

 

It looks like a CPU issue to me and since Heaven uses one thread/core to render you might want to see what your single core score is on Cinebench R15 or 20.

My i7 8086ks did at best 222 in R15 and my i7 8700k usually did around 205. You should be somewhere in between.  

 

 

[tylrtexs]

34 minutes ago, jones177 said:

The vram would not heat up that fast and when it causes a down clock it is smooth.

 

It looks like a CPU issue to me and since Heaven uses one thread/core to render you might want to see what your single core score is on Cinebench R15 or 20.

My i7 8086ks did at best 222 in R15 and my i7 8700k usually did around 205. You should be somewhere in between.  

 

and this is the single core results. 

[jones177]

That is low for your CPU but it may be the computer is doing something in the background. It is not low enough to cause the issue that you are having.

Your multi core should be at least 1500.

Here is my test with 2 less cores at 5ghz all cores.

 

Next would be to use MSI Afterburner to see what the CPU is doing when the slowdowns occur. You could use it with Heaven but not if X1 is running. That sometimes causes false reporting that may need the computer to be switched off the clear.