Steam Link

Steam Link

Large ~1 second hitch every 60 seconds on Quest 3 with Steam Link
I've been trying to help a Twitch streamer diagnose a recurring hitch issue they have been experiencing with Steam Link while playing games on their Meta Quest 3. She first noticed the issue when playing BeatSaber. Every 45-60 seconds she would experience a ~750ms hitch in gameplay. After some investigation, this issue occurs on other games as well and in SteamVR Home as well, although it seems to occur most frequently in BeatSaber. We suspect it's because there is the most going in in a frame in BeatSaber due to all the on screen effects.

We turned on the SteamLink > Show Debug Graph. Every time a hitch occurs we see this large gap in the WGPRX value (what does this value measure btw?).

Beat Saber Debug Graph[drive.google.com]

We have tried the following settings changes but nothing settings changes but nothing seems to fix this hitch:

  • Verifying the refresh rate is set to 90Hz
  • Setting screen resolution scaling setting to 100% (and even lower)
  • Turning on "Remote Play" > "Low Latency Networking"
  • Making sure OBS wasn't running with hardware encoding
  • Lowering "Target Bandwidth" to 60Mbps

The target bandwidth setting I really thought would be it, but still got the same notch:

Hitch with target bandwidth setting[drive.google.com]

The streamer was previous using the wireless connection to a Wifi6 router provided by their Japanese ISP (PC hardwired to the router), that is out of the room but nearby.

We tried some tests at varying tests at different distances to the router, but basically same result.

We then followed this recommendation and setup a dedicated TP-LINK WiFi 6E AXE5400 router in AP mode in the same room as the PC with line of sight to the headset, but got the same issue.

Wireless AP Setup Diagram[cdn.discordapp.com]

Here is the latest capture playing beatsaber showing the hitch at 20secs and again at 1m20s:

Game Play Hitch Video with Debug Graph[drive.google.com]

The streamers PC specs are as follows:
OS: Windows 11 Home 64-bit
CPU: Intel Core i7-13700 (2.10GHz-5.20GHz) - 16 cores / 24 threads
Graphics Card: NVIDIA GeForce RTX 4070 Ti 12GB GDDR6
RAM: 64GB
SSD: 2TB SSD (NVMe Gen4)

I have been trying to reproduce this same hitch on my PC using the exact same TP-LINK WiFi 6E AXE5400 router in the exact same AP setup and Meta Quest 3, but have been unable to hit the same issue using this set of hardware:
OS: Windows 10 Pro 64-bit
CPU: AMD Ryzen 9 7950X 16-Core Processor 4.50 GHz
Graphics Card: NVIDIA GeForce RTX 3090
RAM: 64GB
SSD: 2TB SSD

At this point I am wondering if there must be some sort if CPU hitch that is causing this issue.

My next steps were to:
  • Update to Win11 on my PC to see if I can repro the issue
  • Have the streamer record an ETW trace on their PC using UIforETW and see if I can see an obvious CPU hitch

If it is helpful I have also attached the streamers latest driver_vrlink.txt log from when they were reproing the issue:
driver_vrlink.txtI think this large "Bad link event" logs (what are these btw?) line up with the hitches they saw:
22:01:07.787 [Info] - Bad link event: d=2.6 ms 70.0 mbit/s -> 70.0 mbit/s, ping 40.0 ms, holdoff 0, auto 0 22:01:07.800 [Info] - Bad link event: d=2.6 ms 70.0 mbit/s -> 70.0 mbit/s, ping 35.0 ms, holdoff 0, auto 0 22:01:31.766 [Info] - Bad link event: d=333.1 ms 70.0 mbit/s -> 70.0 mbit/s, ping 2.9 ms, holdoff 2, auto 0 22:01:33.618 [Info] - Bad link event: d=529.7 ms 70.0 mbit/s -> 70.0 mbit/s, ping 100.0 ms, holdoff 2, auto 0 22:02:13.724 [Info] - Bad link event: d=332.2 ms 70.0 mbit/s -> 70.0 mbit/s, ping 4.2 ms, holdoff 2, auto 0 22:02:22.521 [Info] - Bad link event: d=726.6 ms 70.0 mbit/s -> 70.0 mbit/s, ping 100.0 ms, holdoff 3, auto 0 22:02:28.009 [Info] - Bad link event: d=322.3 ms 70.0 mbit/s -> 70.0 mbit/s, ping 2.7 ms, holdoff 2, auto 0 22:08:32.617 [Info] - Bad link event: d=13.6 ms 70.0 mbit/s -> 70.0 mbit/s, ping 37.6 ms, holdoff 0, auto 0
Any debugging / troubleshooting suggestions would be much appreciated!
[drive.google.com]
< >
Showing 1-8 of 8 comments
I've also encountered this bug, it is incredibly annoying.

I fix it by turning WIFI OFF in the Quest headset and then turning it back ON.
If it doesn't want to turn back on restarting the headset fixes that issue.
Thanks for the suggestion @ghostt! Sadly turning the wireless off and back on again on the Quests didn't seem to help our issue. The same hit reappeared as usual after the wifi restart.

I was finally able to repro the issue on my PC after updating to Win11 (was on Win10)! I was able to catch the hitch in an ETW (Event Trace for Windows) log using the most excellent UIforETW[github.com] capture tool.

The first thing we discovered is that the hitch can happen when discord is running. I saw the hitch twice when discord had a random CPU spike:

Discord Hitch 1[drive.google.com]
Discord Hitch 2[drive.google.com]

I suspect it might be the GameOverlay feature, but I can't verify for sure. Turning it off seems to cause the discord CPU spike to go away? Though more testing is needed.

That said, even with Discord not running at all (exited the program), I still do see this weird lighter hitch that seems to correspond with a 133ms drop out in vrserver.exe

VR Server Hitch[drive.google.com]

It looks like vrserver's GPU usage drops to 0 suddenly despite not having any other obvious preemptions that I can immediately tell from the log (though I am a novice with the Windows Performance Analyzer tool, so entirely possible I am missing something here).
Binarys,

Can you try running a copy of cnping from here: https://github.com/cntools/cnping/releases/tag/1.0.0 to do a high frequency ping your headset, alongside SteamVR (And also try doing while not running SteamVR, but waving your headset around)?

The RX dropout looks much more like it's something to do with your wireless, istead of anything to do with the CPU.

If you can run it alongside and verify you can repro, and see if the hitch is present in cnping, as well as also running with just cnping, we can disambiguate the underlying issue much more easily.
Hey Charlesl,

Thanks so much for reaching out! Thanks for suggesting this tool.

I did 3 tests and am a bit confused by the results.

The first test is playing BeatSaber using SteamLink. You can see the severe hitches I am referring to at 1m5s, 1m25s, 1m46s, 2m06s, 2m26s, 2m46s, 3m10s, and 3m26s.

There are corresponding spikes in the cnping graph at those times. Looks like ping Avg around 6ms with the spikes up around 75ms-90ms.

cnping with SteamVR Link playing BeatSaber[drive.google.com]

The confusing part is that when I stop SteamLink and just watch cnping the average jumps up to around 22ms with max spikes up to 170ms:
cnping without SteamVR[drive.google.com]

I made a third test where I start with SteamLink off, then open the SteamLink app on the Quest BUT not connected to my PC yet, and then close the SteamLink app. While the SteamLink app is running (starting at 4s in the video), we see the avg ping time drop from ~20ms down to 4ms and then back up 20ms after I close it again.

cnpng with and without SteamVR[drive.google.com]

I am not really sure how to compare these results since it seems like the worst spike while playing steamLink is 50ms better then worst spikes when SteamLink isn't running
Binarys,

Thank you for running these tests.

1. In gameplay, it is very interesting that there are corresponding spikes, though they are not as large as I would expect for the situation you are seeing, in that they are typically <50ms, but, that could materially be gaps of up to 70 or 80ms all things added.

2. The spikes you see when not in gameplay are either your desktop or headset doing a wireless search. This is common if there is a wifi menu open somewhere. This data is not representative of what you would see during gameplay. Need to figure out why wifi is scanning then do the comparative test.

3. Can you make your third link "cnpng with and without SteamVR" sharable to all with link?

4. When you do find how to get rid of the scan spikes, be sure to, in your test, move the headset around.

Charles L
Charles,

Thanks for the analysis of the above. And sorry I forgot to make that 3rd video sharable. I just fixed that.

I figured out what was up wireless search spikes. It turns out if you have the settings menu open to the wifi settings panel on the Quest you'll see that behavior, even if you have other windows open. Once you close the settings menu the ping rate will drop down to the expected 4ms avg (and the occasional 70-80ms hitch).

I recorded video of cnping while in the Oculus home while moving the headset around:
cnpng without SteamVR(Fixed)[drive.google.com]

So, given that these spikes do occur with no game running and SteamVR not connected this does seem to be wifi router / Quest related.

I did a bit more digging to see if there were any router config tips for VR streaming and came across this one thread on the Meta forums. One person mentioned they had better luck disabling Wifi6 mode and using 5G instead[communityforums.atmeta.com].

I went ahead and disabled the 6Ghz interface and connected the Quest to 5Ghz on the dedicated TP-LINK AXE5400 router and then two runs thru the same BeatSaber level shown in the first gameplay hitch video:

5G Gameplay Test[drive.google.com]

5Ghz seems to be working great for me. But am happy to try any other settings changes or tests if you were curious to see the results.

Thanks,
binarys
Did you manage to figure this out? I am pretty sure I am having the exact same issue.

Just got myself a Quest 3 and tried using Steam Link. Trying to play Beat Saber but my main issue is that every 25-30 seconds there are 2-3 seconds of frame-drops where the picture gets distorted (glitchy) and audio cuts out/speeds up to catch up. Looks likes it goes to low bitrate.

It's quite consistent and never stops. It keeps coming back every 25-30 seconds and it doesn't matter whether I am playing a song or I am just in the main menu.

9800x3d, 64GB ram, 4080, everything installed on an nvme drive. PC connected through ethernet, Quest 3 connected through WIFI 6, 5 ghz.
Apologies for the necro, but I am experiencing the very same issue and given that this topic had developer feedback in the past, I am going to try my luck reviving the discussion.

Similiar to the OP, I am getting frequent dropped frames (pink / purple colored in the SteamVR performance overlay), which coincide with ping spikes to my VR headset (up to 100ms). GPU and CPU utilisation is very low, overall, but there is a certain tendency for the lag spikes to occur when new scenes are rendered - e.g. when teleporting to the ball in Walkabout Mini Golf. The issue also gets worse when moving further away from the router.

Importantly, the issue only occurs with SteamLink. Using Virtual Desktop, even when using SteamVR OpenXR runtime, I get zero dropped frames, whereas I end up with something like 2-3k dropped frames after an hour of Mini Golf when using SteamLink. Also, ping spikes never reach the same levels, when using Virtual Desktop - max was 18ms during tests.

Also, interestingly, I went through a major hardware swap recently, which had no impact whatsoever on the issue.

My setup:

Headset: Quest 2
GPU: AMD Radeon 9070 XT (but had the same issue with an NVIDIA 2080)
CPU: AMD Ryzen 7 9800X3D (but had the same issue with an Intel I5 9600K)
RAM: 32GB DDR5 (but had the same issue with 16GB DDR4)
Harddrive: Sandisk WD_BLACK SN850X NVMe
Network: TP Link Archer A7, connected to the PC via onboard Realtek 2.5Gb via Cat7 cable, and connected to the headset via a dedicated 5Ghz 802.11ac / WiFi 5 channel; Used only for streaming, i.e. not connected to the internet or any other peripherals


Logs (driver_vrlink.txt) are littered with events like these:
Sun Aug 23 2026 11:58:52.856 [Info] - THROTTLE EVENT: 8373 Sun Aug 23 2026 11:58:52.864 [Info] - Bad link event: d=7.8 ms 188.3 mbit/s -> 187.7 mbit/s, ping 7.3 ms, holdoff 1, auto 1 Sun Aug 23 2026 11:58:53.092 [Info] - THROTTLE EVENT: 28239 Sun Aug 23 2026 11:58:53.098 [Info] - Bad link event: d=42.6 ms 188.0 mbit/s -> 185.0 mbit/s, ping 19.3 ms, holdoff 1, auto 1 Sun Aug 23 2026 11:58:53.129 [Info] - Bad link event: d=16.6 ms 184.9 mbit/s -> 184.6 mbit/s, ping 17.3 ms, holdoff 0, auto 1 Sun Aug 23 2026 11:58:53.680 [Info] - THROTTLE EVENT: 33390 Sun Aug 23 2026 11:58:53.685 [Info] - Bad link event: d=55.8 ms 185.7 mbit/s -> 178.7 mbit/s, ping 30.5 ms, holdoff 1, auto 1 Sun Aug 23 2026 11:58:53.768 [Info] - THROTTLE EVENT: 16691 Sun Aug 23 2026 11:58:53.776 [Info] - Bad link event: d=83.1 ms 178.7 mbit/s -> 163.5 mbit/s, ping 32.1 ms, holdoff 1, auto 1 Sun Aug 23 2026 11:58:53.818 [Info] - Bad link event: d=8.3 ms 142.7 mbit/s -> 142.6 mbit/s, ping 17.1 ms, holdoff 0, auto 1 Sun Aug 23 2026 11:58:54.210 [Info] - Bad link event: d=16.7 ms 166.6 mbit/s -> 164.3 mbit/s, ping 15.9 ms, holdoff 0, auto 1 Sun Aug 23 2026 11:58:55.409 [Info] - Bad link event: d=8.7 ms 169.0 mbit/s -> 168.8 mbit/s, ping 8.8 ms, holdoff 0, auto 1

It's wild that this issue seems to persist (for some of us) for more than a year now. Anyone around that could share any insight?
< >
Showing 1-8 of 8 comments
Per page: 1530 50