Closing SteamVR with OC CPU leads to WHEA

[mochuelorojo]

Hi all.

 

Recently I updated my system with a new CPU liquid cooler, an asus aorus 1080Ti and an SSD hardrive. I also decided to do a “mild” overclock with the wizard of my BIOS. Now I’m running my i7 7820X 3,6Ghz @4,6Ghz. It’s great cause now I get 45 -70 fps even in career mode (settings in high) vs the previous 25-45fps (set in balanced) previous to OC. I’ve checked the integity of my OC with stress tests like Blender and several benchmarks; I had no problems: high frames and no overheating problems so I decided start playing and enjoying… Well, at least I thought so because here comes the “but”: when I take my oculus headset off and the steamVR turns off automatically (or when I do not have HMD tracking), I get a WHEA uncorrectable error (also known as BSOD) that reboots my computer. Always. No matter how long I’m playing. It’s not even necesary to open any game. Just the steamVR app, and when the HMD tracking is gone previous to automatic turn off (in red), the BSOD again… (ONLY when overclocked) At 3,6Ghz there are no problems so I decided a softer OC at 4.0Ghz: same problem, WHEA uncorrectable error again…  According to coolermaster power supply calculator I need 625 Watts to run this rig with no problems: my power supply is 750 Watts (I think it’s pretty weird that I pass a stress test with no faults but when I CLOSE the steamVR it crashes…) If I play games with oculus home platform there are NO problems either.

 

I’ve been looking for info in several forums. There are a lot of reports about crashing steamVR app, but they are usually when starting, not when after playing during hours with no issues, it crashes and reboots your computer when you take your VR headset off.

 

I updated my BIOS, firmware, motherboard, USB 2.0 and 3.0 drivers. I also checked connecting both, sensors and headset, in different 2.0 and 3.0 USB ports with no luck either. I’ve reinstaled steam, steam VR and oculus software as well. Nothing seems to help. When I un-overclock the problem is gone but I’d like to keep my CPU at least at 4,0 Ghz since the gain of fps is noticiable. If I run “whocrashed”, it says the crash is related to “hal.dll” and kernel. Can’t give you much more info… Any ideas? Any advice will be welcome

 

Jordi

[Wolf8312]

Well I dont know what the hell I'm talking about really so take that on board, but I've had that Whea error message too when stress testing my own OC.

 

Plus if it only happens when you are OC'd then it seems to me almost certain that its due to the overclock. To be honest in game BOS doesent seem to put too much temp on my own OC but you might consider doing what I do.

 

I use air cooling and dont like like to run my OC 24/7 just in game when playing BOS or DCS. As in VR its pretty tiring anyway with a headset I rarely play for more than an hour at a time and then switch off the OC and rest, but temps at their absoloute highest only get to 60, and thats just spikes mainly, on average more like 40-50.

 

Are you minitoring the temps in game while you're playing?  I actually find the game loading up is more stressful for the CPU than actually playing the game. With the software I use I can turn it on right at the beggining of the actual flight if I so wish and also turn fans to max. Anyway using this method I've never had any BSOD problems, I'm using ASUS suite 3 btw. 

 

Oh by the way I also got BSOD's using my bios to give me an auto OC! Had better luck with ASUS suite using 'fast' rather than 'extreme' settings. ASUS suite 3 has an auto tool but also has all the options to set the OC manually from windows without having to use go into the bios. It's really a great peice of software, though serious OC nerds would scoff at the notion!

[SCG_Fenris_Wolf]

You might want to run that overlock through stress testing during the night first. 

 

Out of ideas what you can try with that cpu tbh, I can only recommend again to you to exchange it for a one better suited for today's gaming. 

 

What you could try however is to install the Oculus Tray Tool, and then check "set OVRserver priority to high". Then start Oculus Home, minimize that, then SteamVR, and then test how it reacts. Maybe it crashes because it's not getting enough resources. But that's a far stretch towards a BSOD...

 

But to be honest, after a quick Google search, it sounds like you went far beyond what the CPU allows for overclocking. And going higher with the voltage to support the strong overclock may create temp spikes, especially if you have the Load Line Calibration on Auto (put it to 0 or 1 if you can), and fry your vcore regulator / vrm or even the CPU in the socket.

 

Anyway, good luck mate. 

Edited March 25, 2018 by SCG_Fenris_Wolf

[mochuelorojo]

Well, temperatures never reach more than 50 ºC in game. I have my corsair h115i set to the max cooling rpms. I guess the problem is the OC too, but I hope that it was a steamVR software issue. You know, something I could fix changing something inside a config file since the thing is that it happens with any game in this platform. I don't think is an IL2 issue. If I could start IL2 with oculus home app it wouldn't  happen. As I said it's not even necesary to play any game in steamVR. Just opening it and inducing the "HMD not found" message it triggers the WHEA... 

[SCG_Fenris_Wolf]

Just fyi, as you know Corsair h115i is a watercooler, so it doesn't cool the vrm (the voltage regulation unit next to the CPU) on the motherboard. Try to put LLC to 0, run stress test, then to 1, or even to 2.

 

 

Entiendo que no es facil, pero lo mejor recomendación es que tu canjees lo por otro procesador 

Edited March 25, 2018 by SCG_Fenris_Wolf

[mochuelorojo]

19 hours ago, SCG_Fenris_Wolf said:

Just fyi, as you know Corsair h115i is a watercooler, so it doesn't cool the Vcore, the voltage regulation unit next to the CPU on the motherboard. Try to put LLC to 0, run stress test, then to 1, or even to 2.

 

No luck. The same problem with any value... I just wanna play. I think I'll heed your advice and I'll try to buy a 4,6 Ghz single core (or higher).

 

Thank you guys