cancel
Showing results for 
Search instead for 
Did you mean: 

High Latency Spikes causing FPS tanking RTX 3080, Gaming A and Nvlddmkm.sys

Adrian1983
Level 10
Ok so it's not something that seems to happen with all games but some are doing this and it is annoying the hell out of me since day 1 setting this up I am noticing this, Playing Borderlands 3 yesterday once it completely froze my system for a second with the GPU tanking performance.

Ok so I am on the Gaming A which has 4 M2 Gen 4 NVME slots, I have 2 slots populated with 2x Samsung 980 Pro's Gen 4 and 2 slots populated with 2x Samsung 970 Evo Plus's Gen 3, Nowhere in the Bios manual does it state that having any particular or all 4 slots populated will it share bandwidth with anything else on the PCI chipset pipeline according to the manual and to be honest I don't expect that when it is PCIE 5.0 capable and nothing using PCIE 5.0 yet and certainly not in my system.

Anyway after doing a bit of my own research ie checking the Frame times in HWInfo after that massive freeze reported frame times in the 200ms+ range in HWInfo for the GPU, Soooo I then ran Latencymon to see and no surprise the Nvidia driver literally within mins casusing tons of page faults and high spikes within minutes of just sat on the desktop, However I don't want to blame the Nvidia driver just yet as it could be some other issue with the motherboard causing a conflict so I ran MSInfo and checked the Conflicts/Sharing in hardware resources and surprise surprise my 3080 seems to be sharing resources with NVM Controller and Intel PEG10 (Whatever that is???) At least that's what it looks like otherwise why would all 3 devices be in the Conflicts/Sharing fields in the first place, There shouldn't be anything in there with a happy functioning system so to speak that I know of.

Images attached...93371

Upon a little more Googling I found others with the same issue and some blaming the Z690 Chipset has issues and Intel blaming motherboard manufacturers, Basically finger pointing at each other, If this is the case this is not good and will never get resolved with this chipset and will only get worse with faster GPU's chewing more of the PCIE 4.0 Bandwidth when the next gen GPU's are available so my question is how do I remove/stop these conflicts in Windows or is this not possible with Windows being automated with IRQ's and resources going forward?
4,773 Views
15 REPLIES 15

ahfoo
Level 13
It very normal business practise pointing finger to each other working within. So end of the day as a consumer who need to wait for explaination and solution. By the time they fixed the issue, more issue arise and it go on and on and never end till new generation cpu, chipset and mobo release.

ahfoo wrote:
It very normal business practise pointing finger to each other working within. So end of the day as a consumer who need to wait for explaination and solution. By the time they fixed the issue, more issue arise and it go on and on and never end till new generation cpu, chipset and mobo release.

Yeah I must admit can't see anyone admitting fault here whether it's Intel or the Mobo manufacturers, Guess we'll never know but either way there's a bottleneck in the pipeline and the GPU is being starved of data somewhere along the line, Whether it's the CPU or the Chipset at fault (Or both) this is not right and it shouldn't be happening.

People will automatically blame Nvidia but it's another service which is causing the high Nvidia spikes, I don't believe personally it is the NVidia driver it's either the M2's on the same pipeline or the amount of them or chipset or both.

Squall_Leonhart
Level 8
the nvidia driver doesn't cause page faults, this is a disregardable value since 99% of them are thrown by the Windows defender active scanner.

excessive dpc latency is due to bus errors, remove or get a gen 4 qualified riser.

anything under 600 is perfectly fine operation, this is a transition spike between low power bus state and full power bus state.

Your picture has no conflicts.

Squall Leonhart wrote:
the nvidia driver doesn't cause page faults, this is a disregardable value since 99% of them are thrown by the Windows defender active scanner.

excessive dpc latency is due to bus errors, remove or get a gen 4 qualified riser.

anything under 600 is perfectly fine operation, this is a transition spike between low power bus state and full power bus state.

Your picture has no conflicts.


Well I don't have a riser so if it is a bus error then it's still a fault of the chipset or the motherboard.

I did have a Cablemod PCIE Gen 4.0 Riser which I removed weeks ago to eliminate that but still the same issue.

I have been gettting over 2000 errors sometimes in 5-10 m, Not always but sometime it spikes and quickly shoots up to over 2000.

Adrian1983
Level 10
So is this acceptable? Doing a simple render test in GPU Z to check the bandwidth test and then doing a quick test with Crystal disk mark almost instantly bangs me into the thousands? And this is with the GPU not doing anything intensive just the question mark bandwidth test in GPUz and a quick M2 Read test in Crystal disk mark.

This is not normal behaviour looking at that it's a motherboard or motherboard driver issue.93392

Adrian1983 wrote:
So is this acceptable? Doing a simple render test in GPU Z to check the bandwidth test and then doing a quick test with Crystal disk mark almost instantly bangs me into the thousands? And this is with the GPU not doing anything intensive just the question mark bandwidth test in GPUz and a quick M2 Read test in Crystal disk mark.

This is not normal behaviour looking at that it's a motherboard or motherboard driver issue.93392


This tool is not meant to be run on a busy system, its meant ot be run on an idle system.

Busy includes any vendor software that updates cheesey rgb components (especially RGB RAM)

STARRAIN_ROG
Customer Service Agent
Hi Adrian1983,
Do you mean have the FPS dropping and system frozen issues while gaming?
May I have the model name, bios version and bios settings of your motherboard and the driver version of your RTX 3080?
Thank you.

STARRAIN@ROG wrote:
Hi Adrian1983,
Do you mean have the FPS dropping and system frozen issues while gaming?
May I have the model name, bios version and bios settings of your motherboard and the driver version of your RTX 3080?
Thank you.


Yes randomly, Not in all games but most, Symptom I can see with the OSD of Afterburner GPU power and usage fluctuates and drops momentarily game stutters and recovers, Could happen after half an hour or could happen at anytime to be honest, Temps are good everywhere, GPU never over 55c Hotspot 65-70c GPU Vram in some games 60-68c, 12900K Temps in games 35-44c all good,

Something is making the GPU starve of data and after running the latency DPC monitoring program I can see why.

Here are the details as follows, I have tried multiple Bios's for the board and Nvidia drivers still the same.

The board in question is the Rog Strix Z690 Gaming A with 4 M2 Gen NVME slots.

Motherboard Bios version 1304

RTX 3080 Current driver version 512.95, Happens with previous drivers also,

Here are my bios settings in a text file on Google drive.

https://drive.google.com/file/d/1gfhtM9PV9_pVkYzd0JyXQSZg2HqomZT0/view?usp=sharing

Thanks in advance!

Adrian1983
Level 10
This is after 5 mins of running the latency monitor and running 1 crystal disk mark test on and Read/Write on 1 test not all 4.

There is always errors in device about WMI and the CPU being in a reduced performance state for so many seconds so I also tried High Performance mode and CPU un park utilities to eliminate that also and still the same.93399934009340193402