Rift S issue - frequent position resetting

[RampantMark]

I'm hoping you guys might be able to help with an issue that is occurring in my Rift S.

I mainly play Elite Dangerous, in a seated position, but this issue isn't restricted to this game, or in seated versus standing games...

As far as I can tell, I have the most up to date MB BIOS and drivers, and most current Windows update.

I have disabled power save to USB and to the Rift S USB hub, and I am using a powered USB hub to connect the Rift S to a USB 3.0 (maybe even 3.1) port.

I'm running a custom OC Vega 56 within Wattman to 950 memory speed/1650 GPU speed, but the issue is also happening in non-OC mode.

I'm guessing there's an occasional glitch in either the USB or DP pipeline, because the way the fault manifests is that the colour in the Rift S display gets "washed out" and "pastel" for a spilt second before the headset repositions itself to one of 2 rough positions, either to the standing default position, (which makes the view originate in the game character's lap, or repositions to a seated level ,but a few inches backwards, so I can see my game character's chest in front of my eyes. It's a bit weird, but I can also say it's annoying as hel to have to keep pressing the key to "reset HMD" on a continual basis. I'm totally in the dark as to what might be glitching out the USB or DP connectivity...

If anyone here has any experience of this particular glitch, or something similar, I'd be grateful for any suggestions.

Thanks

 

[RampantMark]

MB is Gigabyte AORUS Elite (rev 1.0) running the penultimate BIOS - F41. There is only one newer BIOS is F42.c and that centres on ABBA.

USB port being used is a blue tongued USB 3.0 is on the back panel of the AORUS, intermediate powered USB hub is a TP-Link 7-port powered hub model UH700. The Rift S does work, so USB hub "compatibility" is perhaps not the issue, it's just that when in use the position needs to be reset. Sometimes once or twice per game session, sometimes 20+ times in the first 5 minutes... Although in Beat Sabre I tend to just accept the new perspective so that it doesn't keep jumping and ruining the flow.

 

[RampantMark]

Sorry, just after reading your signature. The problems you are describing are some of the symptoms of the USB not working correctly. Even when you use incompatible ports, the Rift S does work, but, starts having issues, losing tracking, losing frame of reference, height from the ground etc. What I would do first is plug out everything from the USB ports and only connect the Rift S to one of the USB 3 port on the back panel. The four USB 3 ports are the ones between the HDMI connector and the Lan connector on your motherboard.

Then start the Oculus software and go into settings and reset the floor height.

I'll do that when I get the time, but I'll also "repair" the Oculus software in between removal and reconnecting.

 

[RampantMark]

Use a tool called USBTreeView to see if you can locate the rift into a USB port that is on it's own Root Hub or even better a controller with nothing else plugged in with it.

Great, thanks for the tip.

 

[RampantMark]

I've had some mixed results.

First I removed the Rift S, and repaired the Oculus software.
At the same time I also opened the case and installed an additional fan, as an exhaust in the roof vent, to hopefully control case temperatures a little bit better. Before this I only had 1 exhaust fan in the rear, and this was on the 120mm AiO CPU cooler. Temp peaks were "OK" at ~70ish CPU, ~80ish Vega56 memory in normal high load Elite 3-D space-based gaming, but peaked higher (and precipitated jerkiness and frame-drops) when on planet surfaces where there were lots and lots of render objects. If you're familiar with Elite, I'm talking about "Crystal Shard Biological sites", which clearly put a great deal of additional stress on the GPU and GPU memory as well as the CPU.... but that's kind of an aside.

After doing those things I hooked up the Rift S, using a different Display Port and different USB 3.0 port, and went in the game, with temperature monitor software displayed on the monitor, which I was checking occasionally, just to make sure the additional fan wasn't stealing all the CPU cooler air and causing the CPU temps to increase.
Sadly, the game and entire system crashed 3 times, going into shut-down. Ouch. Pretty sure it wasn't caused by temperature spikes, but can't be sure because I wasn't looking at the temps when it crashed, so conceivably could've been an uncontrolled, divergent spike.
Weirdly, all 3 shut-downs were when I was docked in a station, so there wasn't a lot of 3-D rendering going on or much load on the GPU/CPU!


I spent a bit of time in the BIOS to make sure the fan profiles were all set "appropriately" and that the fans would ramp up with temps, but that all seemed fine. Made a few subtle tweaks while I was there, however.
I also re-validated Elite Dangerous game files, to be sure the game code wasn't responsible for crashes? Also swapped the Display Port to the one that I hadn't tried before (Vega 56 Pulse has 3 x DP).


This morning, 45 minutes of gaming produced no crashes.
There was also far greater stability of the headset position and orientation. I think 1 or 2 HMD resets in total, at the start of a session and it was pretty solid.

However. There is another annoying fault. The 3-D image colour is permanently washed out and pasty. The same kind of pasty colour that the image would display beforehand just a split second prior to each erroneous HMD position reset that I was experiencing.
Funny thing is that the "in station" text is vivid orange and the colour is well saturated and lovely to look at. However, flying around is pasty and unsaturated. The dashboard colour ought to be the same as the vivid orange docked interface, and, while a few days ago I was looking in awe at the rendering of the Orion Nebula, which was beautiful crimson colours and life-like 3-dimensional in image quality, today the same object is a pasty yellowish smear out of the cockpit window. Real shame that it's come to this. (and perhaps this is the image quality some users were complaining of when they swapped their Rift CV2 for a Rift S - when they complained the "blacks" were no longer "black"...?)


So I'm now a bit stuck where to go from here.

 

[RampantMark]

I don't think the headset is faulty, since it is proven that it can support vivid colours and also able to retain robust orientation and position. So it is most likely something to do with my computer setup, other hardware or software environment issue.

A bit of progress pinpointing the problem...

Of the 3 DP ports on the Vega 56 Pulse graphics card, I was using the outermost, with USB via a powered hub. I changed to other DP ports and also experimented with different USB ports without the powered hub in the loop.

The upshot now is that, to play Elite Dangerous:

1. on one DP port, the visuals in the Rift S in 3D gameplay are permanently washed out. It plays fine and the repositions are almost eliminated.
2. on the other 2 DP ports, the visuals offer much improved richness in colour. However, the game crashes randomly and the entire system powers down ungraciously.

This problem *appears to be* isolated to the game Elite Dangerous. It's perhaps to early to be conclusive, but 30 minutes in Beat Sabre - connected to the same combination of DP and USB that crashed the last session of Elite - there was no problem evident.

Next step is to try some other VR games to see what occurs running those. I have Skyrim and F1 2015 that I can try for that...

If anyone has any bright ideas, I'd be very grateful

 

[RampantMark]

And lo, it came to pass that the problem has been rectified.
Possibly.
Probably.

It appears to be that the Radeon Wattman profile I was running may have been responsible.

Stepping it to "turbo" automatic profile seems to have done the trick.

No more position shifts and no more crashes and I've been enjoying Elite Dangerous in vivid technicolour detail. Lovely job.

 

[RampantMark]

Glad you got it sorted!!

Thanks dude.

I was running a mild WattMan overclock profile that a friend who also has a Vega 56 Pulse sent me. Simple memory speed increase to 950MHz. It was actually running really well with everything else including benchmarks and stress testing. Turns out that Elite is just a finickity game.
I'll experiment with WattMan again some time, only next time it'll be with the power turned up a bit and probably with a mild undervolt on the GPU itself.

 

[RampantMark]

I am kicking myself here!! It's one of the first things I normally recommend, remove any overclocks.

The weird thing is that the Vega 56 manual settings I was using are "modest". Not only that - they worked previously with Elite Dangerous in VR on Rift S, although sometimes with the associated VR HMD position shifting and resetting that precipitated this post... while also working absolutely spot on with everything else I use the system for, either in VR or in 2-D gaming. In fact, the overclocks actually improved the performance of Shadow of War significantly, to the point where the system ran cooler with the overclock than without at the same game settings!

Now I have a quandary. Is it my system at fault here? (Have I damaged my Vega 56? (unlikely)) Is the game code in Elite Dangerous somehow bugging my hardware out and shutting my system down? It is a strange place in the game that the recent shutdowns are occurring - whilst docked in a station - which ought to be a lower graphics load than flying around inside the station and while flying over and docking at planetary bases... This morning I watched the VDDR temperature rise to above 80degC whilst sat stationary in dock, despite having a reasonably high manual Vega 56 fan curve set and only a very small increase of memory freq to 800.

It is just odd.

I'm waiting for Black Friday/Cyber Monday to upgrade to a 5700 XT Nitro+, but still not sure what to do in the meantime. Perhaps I ought to lodge a support ticket with Frontier

 

[RampantMark]

In the custom settings of your Vega card, try the setting the power limit to max but leave the rest at stock and see if that sorts your issue. If that seems to solve the issue, we can try undervolting your card then. I know it sounds counter productive, but, underclocking Vega actually gets more performance with less noise, heat and general problems. (for most owners!!)

I would leave the memory clock at defaults for the moment.

I've done this and with the power slider max'ed out, it has survived the space station interface without shutdown, although it is very clear to see (using CPUID HWMonitor) that VDDC VR temps in particular climb very quickly whilst idling, and docked in a station. Most concerning, so it might actually have been VDDC VR or MVDD VR temperature spikes that caused system shutdown. Even during combat with multiple NPC opponents doesn't create as much temperature, or flying around a rotating station - which relate to much higher graphics rendering loads going through the GPU. It's weird behaviour.

I wonder what these component names actually refer to, and what their safe/rated operating temperatures might be?

 

[RampantMark]

So, it works properly now after setting the power limit to max?

The VDDC VR and MVDD VR are rated to 125 degrees. I think temps are around 80 degrees are perfectly normal for those sensors under load.

It's early days, yet - but I *may* have found a solution...

In the carbuncle of Windows 10 OS , I've switched GAME MODE off in its entirety, and this simple switching has enabled me to run a short, preliminary run of docking at a station and making a few in-game station-menu selections without the VDDC rising above 63degC (GPU and memory temps stable around the low-mid-50's), *with memory freq set at 850 in Wattman.*

I sincerely hope that his is the switch that disables what appears to be a bug with this game in particular...


Further test and checking later today/tomorrow. Wish me luck

 

[RampantMark]

Game mode causes people problems in normal games too, so switching it off is definitely worth a try.

Good luck!!

Why don't we all get told about stuff like this?
Either from MS (who ought to know about stuff like this because they own the carbuncle of software, which has turned from a "reasonable" OS into an Apple-style Malware environment!)…
Or from gaming and software developers? In this case it could be any or all of: Frontier Developments (Elite Dangerous developer/publisher - with native/in-house VR support); AMD (graphics card supplier and publisher/owner of Radeon Wattman software); Oculus (VR third party supplier and VR/driver software owner).

Or even from knowledgeable PC suppliers?

Any one of the above could take a bit of ownership and spread the word. Instead, I found this tiny nugget in a single small thread a game forum!

 

[RampantMark]

Because in scheme of things the Game bar problem doesn't affect many people.

As for your other suggestions, that's partly my fault, you see turning off any overclock is a well known troubleshooting step. It should have been one of the first things I told you. Playing games will eventually show up an unstable overclock.

Definitely not your fault, matey.

I've completed a fair test now with memory speed up to 900 again. No shutdown. No uncontrolled overheating of GPU components. The only significant change being the switch of Win 10 Game Mode to off.

Multiple visits to the docked station interface board and temps stabilise at around 68degC, where prior to the switch it was 80 and increasing rapidly.
I'd say that was a software environment problem - not a Wattman overclocking problem. Something in the OS creating a conflict with game code or Oculus VR code... not an unstable overclock, particularly since absolutely everything else took the profile and worked it with stability.

Of course, there is the possibility that it isn't the Game Mode switch change, but something else entirely that has changed in the background that is responsible for the recent lack of shutdowns, but for me, Win 10 Game Mode still looking like the prime suspect.

I'll keep up with further updates as I build up the Wattman profile again to what it was before.

I was also beginning to suspect a PSU issue with my 600W 80+Bronze, but happy that it might not be that, either

 

[RampantMark]

I think you have it sorted. It is unusual to see the game bar cause that big jump in performance. I wonder what caused it?

It's always a relief to find out that you don't have to replace any hardware.

How is the Vega 56 handling VR?

VR with Vega 56 is lovely.
Graphics settings in Elite Dangerous are not very special, and not as high as I'd expected after using a friends CV1 on a Ryzen3 1600/Vega 56 combination, who runs much higher settings, but I guess the Rift S higher resolution pegs things back a little for the same smoothness.

I'm planning to upgrade to RX5700XT Nitro+ around Black Friday/Cyber Monday, when I would expect VR to go from "lovely" to "VERY lovely".

 

[RampantMark]

Turns out OCUK built ny PC with an 850W PSU.

Well done to OCUK.

Sadly, all the steps I'd taken above still didn't solve my issue of shutdowns...

I was forming the conclusion that it was a power issue, anyway, since the computer simply shuts down, without writing any debug information, etc... Which is looking more and more like a power supply issue - and the computer not being served enough wattage, so falls of the cliff and shuts down. No reboot. No simple software crash. Nothing but a hard power down.

So I got into the case and checked the PSU was indeed the 850W that OCUK said they'd put in my build. Of course it was. (Kolink 80+Bronze modular.)

While I was in the case I started to get a bit suspicious of the single power cable feeding both power ports of the Veg 56. One of those cables with 2 power sockets on the end, but power coming from the one cable. I noted there was an available PCIE 8pin socket on the PSU. Which gave me an inkling of an idea. I got the PC boxes out of the loft and had a rummage around for power cables. Sure enough, there's an 8pin PCIE power lead. (Again, with 2 sockets on one end...) Did a bit of cable re-routing and installed the extra cable...

And now I've got the Vega 56 supplied by 2 discrete power cables.

After a bit of digging on t'internet I discovered that the 8 pin PCIE "standard" is to support 150W power draw. This added to my loose suspicions about it being a power supply issue?

A little bit of testing so far.
With standard default settings in Wattman - Played the game that was suffering shutdowns and went to the parts where shutdowns were most prevalent. No shutdowns so far. Max power draw reported by Radeon and HWMonitor software (both coherent) was 180W.
Decided to then push a bit by placing the power slider in Wattman up to max (+50%).
Max reported power draw 232W. No shutdowns so far.
Added a custom fan profile to Wattman, to keep the temperatures a little bit more in check. Nothing very aggressive, just to take the edge off the 70deg GPU and 76 VDDC VR maximums noted so far at sustained 180-200W+ power draw.
That didn't promote any instant shutdowns either.
Final stage so far has been to increase the memory speed to 850MHz.
No shutdowns after a couple of attempts at the high-stress environment for a couple of minutes.

So this time I'm tentatively hopeful that in providing a two cable power supply I might be on to something?

Hopeful.

 

[RampantMark]

sounds like you may have sorted it, but i had the resetting issue using the rift s, where i'd get a grey out/ wash out and then it would reset. I did some profiling and found out my cpu was bottle necking, was running an i3 81000 and upgraded to i5 9400f and have not had any problems since

Thanks

I'm still not sure it's sorted. Had another shutdown this morning, in exactly the same place in the game that has caused these issues. Seems less prevalent, but still not there yet...

The trial continues.

 

[RampantMark]

Dear chap - what is pancake mode?

 

[RampantMark]

Sorry dude!! It's a term that has come in since the arrival of VR. It means normal gaming on a monitor. You know on the flatscreen? Flat = Pancake

HaHa lol yes. Apologies for being slow on the uptake.

I haven't tried this particular game in pancake mode, but yesterday tried the pancake game Shadow of War, and you guessed it - it performed shutdowns. Twice. Not long after starting into the game. Everything is set to defaults.

Today's endeavour did the following:

CCleaner - full registry clean out.

Sought out AMD Ryzen chipset drivers - my god they're not easy to find on the internet and no way could I find them on AMD sites. Eventually discovered the install files on guru3D. All installed. Tried the Rift game. Accessed the particular area that is prevalent of shutdowns - with the HMD perched on the top of my head so I could monitor the pancake projection and HWMonitor and Radeon monitor all at the same time. Whereas previously this game selection had the GPU drawing over 170Watts of power (up to max 181/182Watts in default trims) this time it was drawing a lot less power - between 80 and 130 Watts on average...

I've also used some powershell commands to check my windows system files and all appears fine.

 

[RampantMark]

Still having shutdown issues - and every time it's associated with high levels of system power draw during gaming.

Can't even run fully though the Benchmark in Shadow of War - got a shutdown just as frame rate dropped from initial numbers of 120fps down through 90-ish, indicating that it's in the high-GPU-load scenes at the end of the benchmark that's causing the hard shutdown. No BSOD and no restart - just a simple abrupt power-down.

So today I bit the bullet and contacted OCUK for an RMA. It gets picked up on Tue next week.

Because it is hard shutdowns I'm experiencing, it really all points toward a failing PSU that isn't able to deliver the required power to the system. It might not be that, but that's gotta be the prime contender. It may well be some other component, such as GPU itself or even MB or CPU, but at least it's going back for a proper tech inspection.

Fingers crossed it's something simples.

 

[RampantMark]

Sorry for the late reply, I wasn't around yesterday.

You are returning the PSU?

I hope it's faulty!! It would be great to finally get a resolution for your shutdown problem.

Not just the PSU, but the entire system.