• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Need Help :(

Associate
Joined
31 Mar 2010
Posts
106
I'm using MSI Afterburner graphs to diagnose frame drops in VR on a 3080Ti. I can't see anything wrong with what the card is doing, other than Voltage Limit moving between 0 and 1 on occaision, but I am testing at a 2.0 pixel density on a Rift-S. Native resolution sits at 0, but I guess this is because the card is barely awake.

I look at the CPU core usages and it seems that the 4250mhz boost drops to 3600-3400mhz in a spike, which appears to represent the frame drop/stutter. I'm unsure why and unsure if this is normal, because I never had this problem before on older cards so never bothered to check. These boost spikes appear uniform over a number of cores/threads, which I assume is what the particular game is using.

Other than reported application frame drops the in-hmd debug tools are telling me that CPU time is 3-5 ms and GPU around the same (high res 80hz). However, even at 0.5 pixel density I am having these frame drops, but the CPU appears more stable at this resolution with less drops.

I've had a smooth ride with this system on a GTX 1080, but since the 3080Ti it has become a nightmare.

If somebody could help, because despite my many years of PC usage, I feel like I am out of my depth on this latest graphics hardware.

CPU temps are good and in-game < 60c (56-59c) on a Noctua d15 cooler running stock speeds.

I don't know if the CPU is the problem, but it's the only thing I can see that has a noticable change.
 
Last edited:
Also what PSU? I had odd behaviour when moving from 1070 -> 3070 due to the power supply not being man enough for the job.
 
What's the CPU? Modern graphics cards need a 'modern CPU' to get the best out of them.
I agree, but I knew I wouldn't be running the card to its full potential, but at 80hz on a 2k HMD the CPU should be enough. On monitor it will spit out 250-300 fps with full RT and ultra in Doom Eternal with 20-30% spare on the GPU, so I know there's lots of room for improvement on optimised games. Cyberpunk with full RT ultra will hammer the CPU, but TBH, I only brought this card to continue VR development. A new system was something I was going to get next year.

Cpu is a Ryzen 7 3700x

I take it these cpu clock speed drops aren't normal? I am even having dropped frames in VR template projects that could run on a potato.

Also what PSU? I had odd behaviour when moving from 1070 -> 3070 due to the power supply not being man enough for the job.
Corsair 850w with two dedicated 150w connectors. HWInfo reports power is OK.

It's only a CX model, though, and a little old. It's just that this will happen even if the card is only pulling 150w in low resolution HMD tests, which never go over 80fps.

Also, different VR titles seem to act differently. Alyx is consistently better, and the real problem titles are UE4 titles, but even Onward (Unity) can do it, only a lot less frequently and usually when I am moving around and working the renderer.
 
Last edited:
From your other thread where you mention you were using PBO and had turned Rebar on, have you tried turning both of these off?

Assuming you have the latest Ryzen drivers, latest bios etc, Windows power plan set to Performance etc?

No other devices in the PC that could be causing latency issues (capture devices, additional sound cards etc)?

Might be worth posting some pictures of your bios settings e.g. CPU/Memory/Overclocking options - may be something you've overlooked/set slightly wrong
 
From your other thread where you mention you were using PBO and had turned Rebar on, have you tried turning both of these off?

Assuming you have the latest Ryzen drivers, latest bios etc, Windows power plan set to Performance etc?

No other devices in the PC that could be causing latency issues (capture devices, additional sound cards etc)?

Might be worth posting some pictures of your bios settings e.g. CPU/Memory/Overclocking options - may be something you've overlooked/set slightly wrong
Yes, PBO is off, Rebar is off and I am even trying it at PCIE3 and 4. There's nothing else connected except KB/Mouse and monitors. Power plan is currently Ryzen High Performance, but none of them make a difference. I've tried with default bios settings and I only have the memory set at manufacturer settings. All drivers and bios is up to date. However, the tech support at Gigabyte (both mobo and card are GB) told me to use the latest beta bios.
 
Last edited:
I did another system reset. Removed two memory sticks (down to 16gb) running them at default non-XMP settings and everything else, except hot plug moved to disabled, at defaults.

This is the core clock spikes on the CPU:
clock-spikes.jpg


When the game is idle and HMD removed (gone into standby) these clocks are fixed at 4275mhz (no spikes). The spikes only seem to happen when I am running around, rotating or just looking around. I'm unsure if this is the cause, the symptom or something else.
 
I'm using MSI Afterburner graphs to diagnose frame drops in VR on a 3080Ti. I can't see anything wrong with what the card is doing, other than Voltage Limit moving between 0 and 1 on occaision, but I am testing at a 2.0 pixel density on a Rift-S. Native resolution sits at 0, but I guess this is because the card is barely awake.

I look at the CPU core usages and it seems that the 4250mhz boost drops to 3600-3400mhz in a spike, which appears to represent the frame drop/stutter. I'm unsure why and unsure if this is normal, because I never had this problem before on older cards so never bothered to check. These boost spikes appear uniform over a number of cores/threads, which I assume is what the particular game is using.

Other than reported application frame drops the in-hmd debug tools are telling me that CPU time is 3-5 ms and GPU around the same (high res 80hz). However, even at 0.5 pixel density I am having these frame drops, but the CPU appears more stable at this resolution with less drops.

I've had a smooth ride with this system on a GTX 1080, but since the 3080Ti it has become a nightmare.

If somebody could help, because despite my many years of PC usage, I feel like I am out of my depth on this latest graphics hardware.

CPU temps are good and in-game < 60c (56-59c) on a Noctua d15 cooler running stock speeds.

I don't know if the CPU is the problem, but it's the only thing I can see that has a noticable change.
Maybe @melmac can offer some suggestions as I know he is a frequent user of VR.
 
Are you getting frame drops outside VR?
Not that I have noticed, but much of my time has been on this problem of late. Simple source games, no. Cyberpunk seems to be running as expected, even at 4K full RT Ultra, but I don't like the game so it's just a benchmark. The odd thing is is that it behaves differently on different VR games that use a different engine. Alyx is relatively stable, but can bug out a bit when my collision capsule is interacting with some of the world (complex walls, certain corner spots etc), but for overall rendering issues they're not present, even at 1.75SS and everything ultra. Fallout4 VR with mods (a game I gave up on with the gtx1080) also seems consistent and I may have one or two framedrops a minute, which I put down to the game (I was running it maxed out, so that's probably the CPU on an old engine and perhaps the rendering pipeline, too). All Unreal Engine games behave the same, where some are worse than others. Serious Sam VR is practically perfect.

I can even get it to drop in Oculus Home's default template, which uses UE4.

I don't have many modern UE4 monitor games, but the ones I tried seem to work OK at all resolutions.

Onward (Unity game) was also dropping some frames, but far less frequently. I would expect perfection on a game that is Quest2 focused, so while it's not exclusively UE4, it's that engine that seems to be the most damaged in VR on this card/drivers. I have tried earlier drivers but it made little or no difference.

I can't see how this is a hardware issue if some work (relatively well) and others don't. I can only believe it's the drivers for this card. Nvidia just told me I will bottleneck at 1080p so I need the latest everything, so that's a no go. Afterwards I looked up Nvidia Support and came across this: https://www.youtube.com/watch?v=NbNZxX2MIFM
Maybe @melmac can offer some suggestions as I know he is a frequent user of VR.
I'd appreciate any help with this.

Oculus got back to me again, but they only offered suggestions with DPI scaling the exe and moving the window between my monitors. It didn't help, but I have to admit, they have been very responsive.
 
Last edited:
You say your ram is running at stock, what speed is stock? My system goes to pot in games, minimums mainly (iirc), when my ram was 2666 or below (tried with multiple sets of ram) This is with a 3700x and vega64 both underwater. The ram issue followed the cpu/gpu from an x370 gigabyte gaming 5 motherboard to an x470 tomahawk max II. Anything above 2666 it goes away.

possibly completely unrelated but thought it worth mentioning.
 
have you tried DDU to remove the nvidia driver and do a fresh install of the latest drivers?

Memory speed might be a thing too, check it hasnt dropped back to 2133.

I'd try a fresh install of windows if you havent already, and check its all up to date
 
Maybe @melmac can offer some suggestions as I know he is a frequent user of VR.

I hope I can help.

I'd appreciate any help with this.

Oculus got back to me again, but they only offered suggestions with DPI scaling the exe and moving the window between my monitors. It didn't help, but I have to admit, they have been very responsive.

Have you windows 11?

and are you experiencing stuttering or lag?
 
Back
Top Bottom