VR CPU frame time spikes in multiplayer

[Alonzo]

Hello there!

 

In multiplayer VR, it seems like there is a periodic 'spike' in CPU frame times which leads to a VR headset missing its usual 80 or 90 FPS target. My hardware is an 8086K @ 5.1ghz, RTX 2080, Valve Index headset. This problem has been confirmed by a few other players on different hardware and headsets, and it only seems to occur in multiplayer.

 

Here's an example. This is the "developer frame time graph" from SteamVR, zoomed in. It shows normal CPU frame times (top) of around 5-6ms per frame. Then, a spike (large purple bar, top) to 16ms. According to SteamVR the "Application (Scene)" time is the cause. This causes/coincides with a GPU time spike also (bottom) of 14ms, and the headset goes into reprojection mode for a couple of frames (bottom, red line).

 

 

If we zoom out, this is happening exactly every 5 seconds:

 

 

It looks like this problem does not occur in single player, it's multiplayer only. Some other users have confirmed the problem, and we see it on different servers (Combat Box training server, Combat Box live server, Berloga dogfight server, Wings of Liberty server).

 

I have not confirmed if this is a problem in 2D mode, just VR.

 

One suggestion was this is HUD related, but it seems we get the same problem whether HUD is on or off. Maybe there is something else that happens every 5 seconds on a multiplayer server?

 

More user reports and graphs in this thread on the VR forum:

 

 

Thank you for your time and for the excellent game.

 

[VR-DriftaholiC]

Can confirm this happens to me. 

[Alonzo]

As an update/bump: I have installed a fresh copy of Windows with only IL2 and SteamVR installed, to rule out third party software causing the issue. I still get the same problem. I think there is something in the multiplayer code causing the 5-second stutter.

[GOA_AveFenix506*VR*]

I confirm this problem with my hardware ( Oculus Rift CV1 ).

 

-Karaya-

[E69_Qpassa_VR]

I confirm this is happening using Steam VR, multiplayer only

[ECV56_Lubermatz]

As posted on VR sub-forum, I confirmed this issue too.

[-DED-Rapidus]

@Alonzo, thank you, I'll look too, it turns out on any VR helmet such a problem?

[Alonzo]

13 hours ago, -DED-Rapidus said:

@Alonzo, thank you, I'll look too, it turns out on any VR helmet such a problem?

 

Thank you Rapidus. I have a Valve Index, but other users are reporting the problem also with Oculus Rift S.

 

Here is some testing that Chili did on Combat Box, and some testing also on Berloga server. The frame time spike is exactly every 5 seconds -- maybe in multiplayer there is some kind of periodic message from server to client that causes a CPU spike, or something else the client is doing every 5 seconds, but only in multiplayer.

[ECV56_Lubermatz]

Mine is Rift S.

 

[Ala13_UnopaUno_VR]

Samsung odisey +,I also noticed this failure

[E69_Qpassa_VR]

I think it's independent of the HMD you are using, in my case a Samsung Odyssey

Edited October 27, 2020 by E69_Qpassa_VR

[E69_Qpassa_VR]

8 vs 8 AI single player:

 

 

 

Multiplayer (Wings of liberty, landed)

 

 

Check the CPU Frametime graph

[-DED-Rapidus]

@E69_Qpassa_VR, you need to try on similar models, missions from the liberty server is not the model version that is taken as a sample (8x8). But thanks for the idea, I'll try on the layout.

[TX-Zigrat]

Any progress on this? I am having the same issue, orange spikes. In single player they seem to go away after about 30 seconds except when zooming in; iin multiplayer they keep happening just like Qpassa shows

[TX-Zigrat]

Anyone still monitoring this? The issue still exists, here is a snapshot of my frametimes-- mostly OK and average frametime (71) is very close to the target (72)  but "hitches" that seem to be CPU driven (I have a Ryzen 5600x)

 

[TX-Zigrat]

Bump, anyone monitoring this forum?

[ECV56_Lubermatz]

Hello, Just ended a sortie on Finnish server and this issue is still there, really noticeable and upseting.

Up for a solution to come!

[-DED-Rapidus]

On 2/27/2021 at 5:28 AM, TX-Zigrat said:

Bump, anyone monitoring this forum?

We monitor, only the question is how to create a load for testing in the sterile conditions of the stand.

[ECV56_Lubermatz]

On 3/30/2021 at 3:01 AM, -DED-Rapidus said:

We monitor, only the question is how to create a load for testing in the sterile conditions of the stand.

Are you actually working on having a testing method for this and other issues?

[-DED-Rapidus]

@ECV56_Lubermatz, all your questions and posts in this forum section should be carefully reviewed by me, reproduced, and passed on to the developers if the issue is confirmed and reproduced.