My current build is mostly new, CPU (Ryzen 7 5800x), RAM (Kingston 64Gb DDR4 3200), Mobo (Asus Crosshair VIII Dark Hero), Video Card (Asrock Phantom Gaming RX6900XT) and NVMe drive (2TB Samsung 970 Evo Plus) are all less than 12 months old, and some are less than 6 months. The only "old" parts of the system are the PSU (AX760i that still seems ok), the case (Corsair 900D supertower) and the SSDs and HDD. I'm running a Noctua NH-D15 Cooler and not overclocking. Though I am running the RAM with with DOCP settings (3200Mhz instead of 2000Mhz).
The PC was running stably with no crashes for several months until the end of last year. Sometime in early January I installed new audio drivers from the Asus site and around the same time I switched my old Samsung 28" 4k monitor for an MSI Optix MPG321UR-QD on Displayport. The new monitor supports 48-144Hz VRR and I've got it set to 120Hz refresh in Windows (10 Pro 64bit). I also have a second monitor (Dell S3221QS) running on HDMI at 60Hz. I should also say that I updated Radeon drivers at some point in January too. Though I'm not sure whether that was before or after the crashes started.
Since then I've had numerous random hard lockups (only) while gaming (World of Warcraft). The PC completely locks up with repeating audio playing over headphones or the Dell speakers (depending on which I'm routing audio through). When I say complete lock up I mean nothing responds, mouse won't move, can't alt-tab or bring up task manager, have to power off the PC to reboot.
All the temps and voltages appear to be fine, I've had Corsair iCUE running in the second monitor and nothing appears out of the ordinary in that regard. CPU temps around 56-60, ditto video card. PSU temps in the mid 30s and power draw around 300W on a 760W PSU. NVMe drive temps are high 40s or low 50s. The only temperature that seems a bit iffy is the southbridge on the mobo. The heatsink on it feels quite warm. Though I can't say exactly what temperature it is running at.
My initial thought was that something weird was happening with the audio drivers. Something I thought was backed up when I saw they'd been pulled from the Dark Hero support site. So I rolled back to the previous drivers but still got random lockups. Including one that was so bad Windows was corrupted enough I had to format the NVMe and reinstall from scratch.
So now I have a completely clean install of Windows 10 Pro 64bit with latest AMD drivers, the "good" realtek audio drivers and all windows updates done - and I'm still getting the random crashes. I've literally just pulled the PC apart and tried reseating everything on the off chance that something was "loose" but I'm not holding my breath.
One thing I've noticed during a couple of the crashes is that the screen may flicker black and come back a few seconds before the lockup. I checked the event logs and I can't see anything like a "driver stopped responding" error before the crash. Basically the only error showing in the logs is "windows restarted after an unexpected shutdown".
So, any thoughts? Could it be related to running two monitors with vastly different refresh rates? While it is summer here it's not super hot and the temps I'm monitoring in the PC don't seem that bad either. Just in windows typing this the CPU is sitting at 34.8c and the video card at 39c (though it has yet to crash in Windows). I can run a Memtest but I'd be surprised if the RAM which was good for six months spontaneously decided to go bad. Is there anything else I should be installing that I've overlooked? I'm pretty sure Ryzen Master was installed alongside the Radeon drivers. I can try either stress testing the system with 3dMark or play another game for hours on end. But it's a bit difficult because the crashes appear to be random. The longest I've gone since the start of the year without a crash is probably 4 or 5 days but other times I've had two on the same day a few hours apart.
Should add, that even though I have crash dumps turned on I can't find one afterwards.
I had similar issues on my Power Color Demon RX 6900 XT.
The PC would just lock up and only a hardboot would reset it and get me video signal.
I used a Group Policy Editor to block Windows Update form downloading their AMD display drivers (Windows 10 Pro 64 21H2).
I found that if I set my GPU fans manually with a flat line not a curve to run constantly at 75% I haven't experienced any crashes or freeze ups for a month.
Also try running a command prompt as admin sfc /scannow to check your OS for errors.
It may also help to run a command prompt as admin CHKDSK C:/f for a quick check of your storage drives for errors.
Always restart PC after command prompt action so that any repairs are completed.
My GPU issues like this started about a year ago to and same problems with different GPU's. I was running 2 MSI Armor OC RX 590's in crossfire.
I also found that after any large windows update it's good to run an sfc /scannow for some reason I almost always find some missing files after large updates. I even do this after GPU and CPU chipset updates.
I already used GPE to block Windows updates so it's not that. Frustratingly every time there is a lockup I have to reload my saved settings though because the Radeon drivers go back to default. I set my fan curve so they never shut down but start low and ramp up quickly from 50c. I'm also sure it's not vid card temps because of the fan settings. The card is so noisy I can tell when it's above 70% fan speed easily.
Brand new install had crashes within a few days. Seemed unlikely but I ran SFC and DISM already with no notable problems. I haven't run chkdisk on this build. But when I ran it on the old windows install that's what literally killed windows and necessitated a format and reinstall. So I'm a little reluctant lol
I found a discussion on the Asus ROG forums where someone was getting random crashes and lockups that they tracked to running everything off the same power strip. Which is something I've been doing. So I'm trying their solution - splitting the load over multiple power points. No crashes/lockups in the last 12 hours. So far so good. If this fixes it, my guess would be that running everything last year was close to the limit on the power strip but switching to the high refresh rate monitor bumped up the power draw just enough. Will update once I see how this goes.
I just remembered another community member having issues with no signal black screens and freezes running an RX 6800 XT.
They found somewhere that if your using a multi rail PSU that they changed where they plugged the PCIEx8 power cables into the PSU so they each ran on a different rail.
They tried that out and it worked. If I remember correctly it had something to do with a more even power load distribution.
This actually makes since. I've worked with a lot of high end car and concert audio equipment in the past and power load management is crucial.
I don't know if this will help you but I'm running a Thermaltake Toughpower 1200 watt multi rail. My GPU was acting up when I first installed it. I changed the connection ports on my PSU which has 6 8 pin PCIE ports and the power up and reboot cycle that would happen every 3 to 5 minutes stopped and hasn't come back.
Good luck I know how frustrating it is when you finish a new build and the PC just doesn't behave right especially how much it costs just to build a decent rig these days.
Splitting the load between different power points didn't help. It was looking good for a couple days, but I've just had two crashes within a couple hours while gaming tonight. And for the record the screen isn't black, it's frozen on the game and there is repeating audio playing over the headphones.
Well I haven't changed my PSU or where the PCIE cables are plugged into it and it ran fine for literally months with zero crashes. So I doubt it's multi rail vs single rail. That said, I am only using two PCIe power cables to power the video card not three. So I could try routing a third one in (though I don't see why it working perfectly for months beforehand). And of course, maybe the PSU has started to fail since January, though I'm not sure how I could test that.
Looking at total power draw it was less than 300W on a 760W PSU at the last crash. Which seems like even with two monitors and an external HDD on the same power board wouldn't be enough to overload it let alone when the PC is running on the powerboard by itself. Although, maybe the power board itself is wearing out too... At least that is an easy test. I'll move the PC to the new powerboard and put the monitors and HDD on the old one...
I might also try disconnecting the MSI monitor and just using the Dell for awhile and see if that makes a difference. Maybe it's something to do with the high refresh rate?
I just wish this wasn't a sporadic crash. If I could reliably produce it within a set amount of time or doing one specific thing it'd make it a heck of a lot easier to diagnose. Being able to go several days at a time sometimes without a crash make diagnosis super slow.
I understand the frustration of inconsistent errors or issues that happen especially when you have plenty of test hardware around and try to get a consistent repeat to narrow down the problem.
You said you were running just 2 power connections from the PSU to the GPU. Are you using a pigtail from one of the PCIE power cables? If so definitely change that. I tried that and found that my GPU got very erratic power draw issues when overclocking and increasing power output to the GPU. I disconnected the pigtail and used 3 separate power cables for my RX 6900 XT. My overclocking and power issues nearly stopped completely. I noticed that for me if I don't push my power slider above 10% in Adrenaline custom tuning it really cuts a lot of crash and freeze ups out. They still happen but instead of everyday to several times a day its like once a month and usually follow windows updates.
My current MOBO also restricts some of my performance capabilities. Running a MSI X370 M7 ACK which only has PCIEx16 3.0. I'm watching the market and plan on upgrading to a MSI X570 Carbon Pro MOBO and a Ryzen 9 5800X or 5900X CPU. Probably drop in a couple of Samsung 970 EVO 2 TB NVME M2 SSD's.
Sorry about the gibber jabberish just enjoy working on computers. They are like adult legos to me.
The PCIE cable has multiple connectors built into it, it's not an aftermarket extension that splits the cable. But yeah, if that's what you mean by pigtail it is. Like I said I'd be surprised if that worked since it ran happily for around six months with that exact same setup with practically zero crashes. But I'm certainly willing to try it. Just need to wait for the next crash now lol
I haven't adjusted power tuning in adrenaline. It's currently disabled. I adjusted the fan control so it never turns off and the speed ramps up rapidly to 100% once you get past about 50c. Which seems fine, GPU sits about 35-40c while at the desktop. And most of the time while gaming it's probably 60ish. Though it obviously spikes up and down depending on the game. The only other setting that I've applied that is "new" is making sure VRR is on so I can take advantage of Freesync.
I went from an x470 board to the x570 thinking it'd be more stable, maybe slightly better performance and while it was fine for six months I'm regretting it now. If I had the choice over I'd have stuck with my Ryzen 2700x and Crosshair VII Hero and waited out the release of the next gen. Any performance increases have been so minimal as to be undetectable and now I'm having these crashes (though to be fair I don't know where the blame for that lies yet). The MSI Carbon looks like a decent choice. I'm actually wondering whether having a fan in the heatsink for the southbridge might be a good thing (the Dark Hero doesn't have one).
Funny you mention the Samsung SSDs, I've posted this same question on the Asus forums and they're questioning the SSD (can I try it in a different slot, do i have another one to test...) I personally think the Samsung drives are amazing and if you're running a smaller/older SSD they're well worth the money. The one thing I've really noticed is they run hotter than the standard SSDs (on both mobos I've used). While I understand that an M.2 SSD running at 60c should be fine it's a bit offputting compared to a standard SSD running at 40c (or less). I would say if you're going for performance and you can afford it go for the 980 PRO or Kingston do a really nice PCIE4 NVME drive.
"The PCIE cable has multiple connectors built into it"
Like ThunderBeaver said, use three cables (two cables if the card has two PCIe connectors), and don't use the two connectors from one PCIe connector. Use three distinct PCIe connectors, one connector from each cable. Don't use two connectors from one cable, and while you are at it, look up what the specifications for your PSU on the AMPs delivery per specific PCIe cable, sometime you have to choose specific ones to get a certain amount of AMPs, yes it is true PSUs are made like that nowadays.
It could be your power supply has aged a bit and some changes to the AMD drivers and now things are changed and it can't deliver the load. I hope that fixes it for you.
I couldn't find anything in the specs or manual about AMPs per cable, but the settings for the PSU show max Amperage of 40A per cable (that you can dial up and down). 40A is the highest they can be set to.
I'm actually concerned about the external power (ie: the quality of the power from the electricity grid) so I've literally just hooked the PC (and monitor) up to a new 1500VA (900W) UPS. I'll update in a week.
Just for the record I'm trying a single change at a time, then running the PC basically waiting for a crash. Since they're anywhere from several times a day to several days between them the testing is taking awhile.
I'm currently testing the three PCIe cables solution. If it gets to more than a week (maybe two) without a crash I'll be thinking that's the cause and solution. Will update later.
The next thing to try (if this fails) is trying to source a different NVME drive to test. *ugh*
Haven't had a crash yet, but I have noticed audio problems. I'm 90% convinced that the crashes when they do happen are related to audio. I can't say 100% of them, but most of them would be when running audio through my Dell Monitor's speakers (while gaming) which is of couse powered by the Radeon. I am using the other MSI monitor for graphics though. Could that be an issue?
I did notice today a couple times where audio hitched and crackled during games but didn't actually crash. I wondered initially if maybe the load on the video card when it was working on both graphics and audio was causing problems. However, I even noticed problems a couple times just playing MP3s in AIMP while browsing the internet and typing in Notepad++. At least once it distorted a good two seconds of a song. Which is disturbing because there should be next to no load on the PC doing that. Windows (and games) run off a Samsung 970 EVO 2TB. The mp3s run off a 2TB Samsung 860 QVO. So it's not like a hard disk in their choking throughput. Or trying to decode MP3s on an old 486...
Please, if You should find solution for this, keep us posted. I suspect I have the same issue. At the end of January I've started experiencing freezes (nothing responding, just looking at frozen screen, not even power button on the case reacting, had to shut down the pc via power switch on the back) at random times - no obvious correlation with activity on pc, sometimes i was even idle, with nothing opened (not even web browser), usually during work. First day it happened once, second day six time (various intervals inbetween), third day I updated UEFI and it SEEMINGLY stopped. Before that on the second day, I did: disk health check, system files check, few passes (16 hours cca) of memtest (free version), SSD health scan, antimalware tests, registry cleanups, driver updates, searched through event log for some correlation and had my eyes on temperatures and voltage. Nothing showed obvious culprit (at least not to me, not exactly computer savy person). I was recommended to update the bios and coincidently, new version was just released at the time. Did so and the lockups did not occured anymore.
Until yesterday. Six times since, I'm even writing this for second (edit: third) time. Coincidentally, there has been another uefi update week back. Updated again, this time didn't help. There's difference though, only once yesterday the whole thing froze with screen still on (as in January). Other five times, monitors went black.
What makes me think, we might have same source, is the audio thing, I've been experiencing for last weeks. Once in a while (an hour or so) for 1-2 seconds the sound turns sluggish and rough and crude (dont know how to describe it better, english is not my native language), as if if "stumbled" upon something. I cannot atest to the sound countinuing after the freeze, as You have described, since it always happened, when I wasn't listening to anything.
Finally, apart from my PC configuration, i shall add, that during both times, I haven't done any change to the hardware, I haven't even installed any SW at those times, apart from Adobe indesign in January. I do have two monitors, both on 60HZ. I do have my headphones connected trough the monitor.
MB: ROG CROSSHAIR VIII DARK HERO, CPU: AMD Ryzen 9 5950X 16-Core Processor, 3401 Mhz, Video Card: GIGABYTE GTX 1650 Super WindForce2 OC, Memory: 128GB, DIMM PATRIOT Viper 4 Blackout DDR4, Drives: HD SAMSUNG SSD 970 EVO Plus and HD SAMSUNG SSD 980 PRO, Power source: CORSAIR ATX 750W RM750x (2018). 9 months old.
The problem sounds a little different because mine only crashes when gaming. It's rock solid while in Windows whether I'm working with Visual Studio, browsing with Firefox, using Office etc. The audio has occasional lagouts (which it shouldn't) while in Windows but no crashes.
Most times when I've seen crashes on PCs in Windows it's been heat (usually CPU) related. I know you said temps look ok but what is your normal CPU idle temp (ie: in Windows doing nothing), and what is the temp when it's working on something? Next most common culprit is often PSU, though usually that manifests when doing something strenuous (gaming, 3d Rendering, Video encoding etc).
Which bios are you using on the Dark Hero? I saw the 4004 got pulled from Asus website with no explanation so you might want to try a different one if that's what you're using. I haven't seen any difference in terms of stability between 3801 and 3904 (what I'm currently using). Haven't tried the new 4006 bios yet.
I also rolled back to the 6.0.8960.1 realtek drivers from the Asus site. They put up 6.0.9244.1 realtek drivers briefly and again pulled them from the site with no explanation. My system had been stable until I installed the 9244.1 drivers. Since then it's been problematic - even after rolling back drivers and even clean installing Windows 10 fresh. You can always try the drivers from the Mokichu thread on the Rog forums: https://rog.asus.com/forum/showthread.php?117596-DRIVERS-Realtek-Audio-(MB-Intel-2xx-3xx-4xx-5xx-6xx...
The drivers you'll want are the "- Realtek Audio Drivers (UAD - ASUS ROG SS3-DTS MB) :" Just a warning though, I haven't tried these ones yet. So no idea how well they work.
Just some other suggestions (from an Asus person) - try swapping in a different PSU. Assuming your boot SSD is NVMe try moving it to a different slot and if possible test a completely different boot drive. I'm waiting on a fresh drive to arrive and I'll try that once it does.
I'm also guessing you're running 4 sticks of RAM to get 128GB. I have seen a couple reports about problems when the RAM slots are completely full (even if the RAM itself is fine). So it might be worth trying it with just two sticks in place. And (again Asus recommendation) make sure the RAM is running default settings (no D.O.C.P or manual tweaking).
I think I'm ready to call this problem solved. I've managed to go a full week with my current setup without a crash. I made two real changes;
1. Turned off enhanced sync
2. Turned off the TPM module in the UEFI.
One thing I've noticed with those changes is less audio stuttering and I think LatencyMon bears that out. Previously I was seeing spikes where the highest latency was 80,441us. And highest ISR of 779 us which doesn't sound much (considering that's micro seconds) but it's now showing 27,500 on it's highest spikes (roughly a third) with a highest ISR of 162 (less than a quarter).
I could be wrong and the fix is unrelated but when I'm no longer hearing the audio lagouts and seeing lower latency and more importantly not seeing hard lockups that seems like a pretty good correlation to me.
I'm just debating whether to reverse the changes and see if I can get it to crash again...
And another update for anyone having similar problems. It's now more than a month since I made those changes and still no crashes. So I'm 100% convinced that the TPM is the culprit. Whether it's because of the physical device or a Microsoft driver, Asus Bios, AMD driver or what I don't know. But that seems to be the root of the problem.
Looking at Asus site there is a new Beta bios (4201) available for the Dark Hero with the new 1207 agesa that is rumoured to address issues with TPM stutters and freezes. I'm going to give that a whirl and see if my PC remains stable once I turn TPM back on in the bios. I'll update immediately if there are crashes, otherwise it'll be a couple weeks before I'm convinced of stability.
Bios is available here (it's Beta so use at your own risk): https://rog.asus.com/au/motherboards/rog-crosshair/rog-crosshair-viii-dark-hero-model/helpdesk_bios
So no crashes yet and the 4201 bios has also gone official (ie: no longer beta). Checking CRC for the BIOS it's literally the same file as the beta, so no need to install if you've already tried the beta.
I ran LatencyMon and it's giving even better results than the previous fix (disabling TPM). Results below in μs (microseconds);
1. Original bios, with TPM enabled
2. Original bios, with TPM Disabled
3. New Bios, with TPM enabled
Highest Measured interrupt to process latency: 54775.8, 27508.9, 256.10
Highest reported ISR routine execution time: 237.8, 161.92, 243.17
Highest reported DPC routine execution time: 27675.12, 27541.3, 357.84
Disabling TPM on the old bios roughly halved the highest measured interrupt to process latency, which was apparently enough to stop the lockups. The new bios reduced it to nearly 1/200th of the latency. The DPC exec time also showed a massive improvement. LatencyMon now shows that my setup is suitable for real time audio tasks without dropouts (as it should always have been).
I been noticing my pc has been shutting down at random since I got the 6900 XT reference model at Microcenter in December.
Long story short, 6900 XT recommended PSU is 1000w due of the power spike, a Microcenter employee test three different 850w PSU with my 6900 XT PC shut off at random, switching to 1000w PSU fix the problem at the store and for me at home, hope this help if you having a random shut down with a 6900 XT with a 850w PSU.
I have a 5800x and 6900xt and want to run on my 750w Corsair SF750. I have the 5800x undervolted and boost limited via bios settings. I currently have a 3080 that is also undervolted and they have run just fine. I'm scared that with the 6900xt I may boost above what the PSU is capable of. Should I suck up and buy a new 850w psu from Coolermaster or am I okay with the Corsair?
And after making sure the video card is powered by three separate PCIe connectors it's just crashed again. So that's not the solution.
My current test is now turning the refresh rate of both monitors to the same (60Hz) and turning off Variable Refresh Rate. That's not a long term solution because I didn't spend $1500 on a monitor and $2200 on a video card to be stuck at 60Hz. But maybe it'll help diagnose the issue.
After that, I'll be disabling onboard sound on the mobo and see if that makes a difference.
This is super frustrating.
edit: actually maybe it's time to pull the radeon and see if my old geforce card has the same crashes.
Unless you really need the latest version of the drivers, you could try some a few months old.
Off topic but I am running 22.2.1 and I now have audio stuttering with vsync disabled or with enhanced sync on on my 6700 XT, no problems with older drivers. I suspect it has something to do with the driver release notes saying that some configurations have a black screen with enhanced sync, or some other newly introduced fix broke something. I have just been leaving vsync on and hoping they fix it. I hope you get your problem fixed.
I was using VRR aka Freesync and enhanced sync before getting the crashes. The biggest difference I can see in the last couple months is the change to a new higher refresh rate monitor (144Hz vs 60Hz). Since I changed the windows refresh rate back to 60Hz I haven't had a crash. But to be fair, I've also had two reboots because I installed drivers/apps and restarted Windows. So my maximum uptime since that change is only about 48 hours. Once I reach a full week I think I'll feel safe to say that the high refresh rate is the trigger. After that I'll bump it back up and see if I can get the PC to crash within a few days again.
Of course, the cause may not be the monitor as such. It could be the extra strain it's putting on the PSU/RAM/Video Card/something else. But if I can pin down that trigger it's a good start.
If that doesn't pan out I might try rolling back to the December Adrenaline and see how that goes. Because the audio stuttering is an interesting similarity.
@Skrybe I think you are on the right track, you could try limiting the FPS to 144 to see if maybe it going over you monitor's refresh has anything to do with it. You are correct, it will put more stain on the system to push more frames. I have seen other posts where on high sync monitors it seems to cause more problems if it spikes above the monitor's max refresh rate. I think Windows and AMD now are having a few problems with bugs when it tries to go over 60 Hz. Let's hope they get it sorted out soon.
Not sure about frame rates spiking above the refresh as a cause. The previous monitor was freesync too but only 48-60Hz. I had framerates capped at 120Hz (for the multiple of 60Hz) and don't think I ever had a crash in six months of that set up.
As for the new monitor, I had previously set it to 120 fps max in game with both 120Hz and 144Hz in Windows. Same random crashes, only while gaming. I tried both with and without the second (60Hz) monitor attached and that didn't make a difference either. But so far just the one monitor set to 60Hz with the game capped at 120fps seems to be crash free.
Higher refresh definitely puts more strain on the video card if nothing else. I have seen spikes in WoW where the card maxes out 100% GPU and 100% fan because the temps rocket. The weird thing is the crashes aren't limited to that scenario. I've seen crashes where I'm in low impact areas with around 60% GPU utilization and pretty moderate temps.
You mention Windows and AMD getting drivers sorted out - I did wonder whether it's going to be a bug that will disappear if I switched to Windows 11. But I'm reluctant to do that at the moment. Might see if I can scrounge up another SSD and install Win 11 on that to test.
If you test with Win 11 let us know how it goes, another thread mentioned a fix had been made for Win 11 but so far not incorporated into 10 yet? Just speculation. I'd like to know how 11 runs as I will probably go to Win 11 at some point, probably a year or two or sooner if Win 10 is not getting many fixes.
It's hard to be enthused when the updated drivers don't work on any game I am playing. I hope it's fixed because there are notes in the latest 22.2.3 drivers saying black screen problems with enhanced sync, so I hope that fixes my audio problem too.
With now having an audio problem with stutter where the audio is unlistenable on any new or older game, I hope to not be stuck on an older driver, some new games I wanted to play since there are a few new raytracing games like Guardians Of The Galaxy that look fun.
I assume you've tried the steps outlined here (or similar guides):
I'm currently running LatencyMon myself to see if it picks up anything. But random is random so won't know whether it helps til it happens.
edit: Just done a slightly over 3 hour run of LatencyMon and while there were no crashes it is interesting to see the results. While "just" in windows (browsing, doing some text editing) there were zero latency issues. Started WoW and immediately started seeing spikes of high latency. After roughly an hour the results look like this:
Your system appears to be having trouble handling real-time audio and other tasks. You are likely to experience buffer underruns appearing as drop outs, clicks or pops. One or more DPC routines that belong to a driver running in your system appear to be executing for too long. One problem may be related to power management, disable CPU throttling settings in Control Panel and BIOS setup. Check for BIOS updates.
LatencyMon has been analyzing your system for 3:17:49 (h:mm:ss) on all processors.
Which does not sound good for a (fairly) high end setup. Latencies;
MEASURED INTERRUPT TO USER PROCESS LATENCIES
The interrupt to process latency reflects the measured interval that a usermode process needed to respond to a hardware request from the moment the interrupt service routine started execution. This includes the scheduling and execution of a DPC routine, the signaling of an event and the waking up of a usermode thread from an idle wait state in response to that event.
Highest measured interrupt to process latency (µs): 80441.60
Average measured interrupt to process latency (µs): 3.598785
Highest measured interrupt to DPC latency (µs): 80438.80
Average measured interrupt to DPC latency (µs): 1.259217
That's using the HDAudBus.sys driver from windows driving speakers on the Dell monitor (I thought this was supposed to be driven by AMD audio drivers btw). I'm going to try a similar test with the Realtek drivers running to headphones over 3.5mm jack and see what that looks like next.
Ok, I tried a similar test with LatencyMon while using the onboard Realtek sound instead of the AMD audio from the RX6900XT and saw very similar results. HDAudBus.sys is the highest latency reported. It never spiked as high as the previous test but I only gave it a few minutes as I saw spikes almost immediately.
So I guess regardless of whichever audio device is producing the sound it's all going through the same Microsoft High Definition Audio Bus Driver. Which is interesting and frustrating.
Is it freezing in windows, or is it full black screen crashing and self rebooting ever?
This issue is due to transient spikes happening where the power draw can cause the PSU to trip overcurrent protection (OCP). I had the same issue on a Seasonic 1000w with a Vega 64. After swapping to a Corsair HX1200 I no longer have any of these types of issues and have tested with all the high wattage gpus including the 6900XT and the 3090 ti.
The fix is to make sure you are using a PSU that is more tolerant of these transient spikes. In general higher wattage PSUs can cope but it depends on the brand.
It's absolutely not transient power spikes (at least for me) because the problem is resolved with either of these solutions;
1. Disabling fTPM in the bios.
2. The latest Asus Bios with the 1207 agesa.
The AX750i allows me to monitor power usage and the highest draw I've seen has been about 620w and that was absolutely smashing the card with max settings and you could hear the fans on full load. It didn't crash then. It doesn't crash in stress tests or benchmarks which max it out. It's definitely related to the latency spikes caused by the fTPM (or the Windows driver accessing it). And that's pretty much confirmed by the fact that they caused audio stuttering even when the graphics card is basically idle (sitting at desktop). Said stuttering is completely gone with the new bios.
"I had the same issue on a Seasonic 1000w with a Vega 64. After swapping to a Corsair HX1200 I no longer have any of these types of issues and have tested with all the high wattage gpus including the 6900XT and the 3090 ti."
That's unlikely to be the raw wattage that's at fault in that case.
The Vega64 was rated at 240W, even if it's having transient spikes of double it's load that's 480W. The rest of your system would not be drawing 500W+ in order to go over 1000W. More likely the 12V rail had poor voltage control and it was getting spikes/drops and that was what was causing the crashes. The Corsair HX1200 has super tight ripple control so it likely resolved that problem.