Troubleshooting 6700XT hard crashes

[Little_Crow]

Bought a 6700XT as my 1080ti bit the dust giving the classic blue sparklies.

Was pretty happy with the purchase, but started getting AMD driver crashes more and more frequently, and now it's giving hard lockups with the screen going completely black.
Following a reboot after a hard crash the Adrenaline software doesn't appear in the systray and the 6700xt shows up in device manager with 'This device is disabled. (Code 22)'

I can re-enable the device and reboot, but another crash when playing a game is never far away

I've tried to induce crashes using furmark and the heaven benchmark, but they run just fine - kick off some Halo and it'll likely crash within a few minutes.

So far I've:
Reinstalled Windows
Unplugged and reseated power to both motherboard and graphics card
Removed the XMP memory overclock profile
Removed 2 of my 4 sticks of RAM
Replaced the 2 sticks remaining with the 2 taken out
Tried different output on the card
Reinstalled drivers with factory reset option

I'm at a bit of a loss at the moment - Just looking for some assurance I've not missed anything completely obvious at this point or another benchmark to try?
I have another machine that I can potentially start switching parts into, just looking for anything I might have missed

Grateful for any help!
Cheers

####Specs####
AMD 5800X CPU
4 x Crucial Ballistix 8GB 3600 DDR4 memory
MSI MAG X570 Tomahawk Wifi, 7C84v19 BIOS Motherboard
Sapphire Pulse RX 6700XT - Drivers 22.8.2
Phanteks Revolt Pro 850W PSU
AW3418DW 3440x1440 monitor

 

[Tetras]

Are you using any other software at all, to monitor the graphics card, or other system stuff?

Are you using separate power cables for the 6 and 8 pins?

Have you tried massively downclocking the graphics card in the drivers? e.g. -50% core clock

Have you tried using much older drivers, maybe even the release drivers?

 

[Little_Crow]

Are you using any other software at all, to monitor the graphics card, or other system stuff?
Using the overlays in Adrenaline, not seeing anything alarming temperature or power draw wise

Are you using separate power cables for the 6 and 8 pins?
Yep

Have you tried massively downclocking the graphics card in the drivers? e.g. -50% core clock
Not yet, will look to try that tomorrow

Have you tried using much older drivers, maybe even the release drivers?
Not yet, will also try that tomorrow

I've now installed 3DMark and tried running both timespy and firestrike benchmarks - they sat for a while before giving an AMD driver crash.
The 2nd time I ran firestrike it induced a hard crash with just a black screen.

Thanks for the suggestions, now I think I've found a benchmark capable of inducing crashes I'm more hopeful of narrowing down the cause

 

[SkeeterUK]

theres a newer bios. 7C84v1B 2022-08-30

MAG X570 TOMAHAWK WIFI | RETURN TO HONOR

Military style with Pre-installed IO shielding, tuned for better performance by Core Boost, DDR4 Boost, M.2 Shield Frozr, Wi-Fi 6, Frozr Heatsink Design, Lightning Gen4

www.msi.com

Btw u tried reseating the card?

If 3d marks ok but halo crashes it could be cpu or mem or both imo. Check cpu temps when gaming not benchmarks, check power plan see if thats ok, check latest amd drivers for ur mb from amd site.

amd drivers for mb

https://www.amd.com/en/support/chipsets/amd-socket-am4/x570

https://www.minitool.com/data-recovery/how-fix-this-device-is-disabled-code-22.html Give that a read maybe too.

 

[Little_Crow]

@SkeeterUK
I did try reseating the card, and installing the latest AMD chipset drivers - forgot to note that in my actions so far
3dMark either fails to start or crashes almost immediately, Halo will run but usually crashes within a few minutes - even seen it happen at a pause screen.
Temperatures look to be under control - hitting 95 degrees when running furmark as you'd expect, but that has never induced a crash

@Tetras
I've switched the card to my HTPC (specs at bottom)
I reran both timespy and firestrike benchmarks and it it has the same behaviour, driver crashes and hardlocks leading to the device showing as disabled in device manager.

Reducing the max clock to 50% allows the benchmarks to run and complete, which is an improvement, but obviously not ideal.
Also tried the 21.12.1 drivers, and same outcome - benchmark completes at 50% max core, but crashes when at the 100% default.

I'm getting more confident that it is the card itself,

####HTPC Spec####
AMD 1600AF CPU
2 x HyperX Predator 8GB 3200 DDR4 Memory
MSI B450i Gaming Plus AC Motherboard
Corsair TX550M 550W PSU
AW3418DW 3440x1440 Monitor

 

[SkeeterUK]

checked for damage on the card? a scratch or a capacitor missing etc?

If its under 14 days maybe u can get a refund or replacement? Even if its after u might wana think about a rma.

 

[Little_Crow]

@SkeeterUK
The back of the card is pretty much entirely shrouded, only a couple of cutouts which I've inspected closely and can see no obvious signs of damage.

It does look like it's an RMA, but dread sending off a card in case the fault isn't reproduced, so I'm trying to be as thorough as I can possibly be

 

[Tetras]

Reducing the max clock to 50% allows the benchmarks to run and complete, which is an improvement, but obviously not ideal.
Also tried the 21.12.1 drivers, and same outcome - benchmark completes at 50% max core, but crashes when at the 100% default.

I'm getting more confident that it is the card itself,

Ouch, that definitely doesn't sound good. You could try bumping it up slowly? Some cards are unstable at factory overclock, but I don't think the Pulse has a factory overclock.

 

[SkeeterUK]

i know u got windows reinstalled but just to be sure of no driver issues have u tried ddu to clear old drivers then tried some drivers, i think im on 22.7.1 which seem ok to me on my 6700xt.

 

[eeii]

You need to RMA this card now. From what you have tried it’s the card. In the the uk you have 14 days for money back or instant replacement.

 

[Tired9]

RMA my dude, if that fails it could be the motherboard.

 

[SkeeterUK]

Its not the pci-e setting in mb bios is it? Like if its set to 3 try 4 or vice versa.

Are you using a riser?

 

[Little_Crow]

I transplanted the card back to my main PC, and it was all fine last night for a couple of hours.
Tried again today and it booted with 640x480 resolution, restarted and it crashed within a few minutes, so an RMA will be raised with a link to this thread

Card was bought at the beginning of August, so well outside of DSR, now
It wasn't until I gave freebee Saints Row a shot that it gave me driver crashes - I initially put it down to just that game, but then other games started experiencing issues and it's only got worse from then.

Thanks all, just wanted to make sure I've got everything covered - will see how the RMA process goes, hopefully they can reproduce what I'm seeing

@SkeeterUK
Restarted to safemode and ran DDU, following a reinstall I get the same issues
No riser cable, and tried on both PCIE4 and PCIE3 in the BIOS now - same outcome

 

[Showboat]

I had similar problem with a 5700xt.
Tried it in second PC and was able to replicate the problem. Tried new PSU in main pc and still no better.
RMA’d and they could find no fault, they offered to send back to manufacturer who then lost it and just gave me a refund… 6 weeks without a card, buy something temporary on MM if you going to RMA and just buy nvidia, had no problems with my 3700.
Good luck!

 

[Little_Crow]

@Showboat
Yeah that's the sort of thing I dread - Played some Halo again today and got a couple of hours of smooth sailing until it started crashing again

My HTPC has my 2 upgrades old GTX 970 in it which I'll probably throw in and content myself with turn based strategies and retro gaming for the duration - hopefully the vendor will be thorough enough to get it to fail.
I am a patient boy, I'll wait