Constant random driver timeout errors

[HockeyJon]

Hey guys!

I'm trying to decide if I have to RMA my card at this point because I'm out of ideas.

At this point I've tried DDU driver clear-outs, formatting windows, installing the card in a new system, editing things like TdrDelay in the registry, and capping the shader clock in Adrenalin to not exceed the specified defaults.

I'm reading Windows 11 24H2 in particular has been bad with AMD drivers, especially because you can no longer disable MPO in the registry. Not sure if anyone can corroborate that.

Just wanted to know if anyone has a solution to driver timeouts that actually works. Last AMD card I had was an R9 280X and it was rock solid. I never had a single problem with that card, and I was overclocking it quite aggressively.

So yeah, outside of RMAing my card (or just getting rid of it altogether and returning to Nvidia), I'm out of ideas, but I read online that AMD's drivers lead to a lot of false RMAs and I'd rather not have to go through that process if the hardware is not actually an issue. Not sure if anyone can corroborate that?

In any case, if anyone has a good solution to driver timeouts, please let me know! Thanks!

 

[pendragon1]

if you have hdr on try turning it off for a bit, see if it stops

 

[kwikgta]

From your signature it looks like you're running a 7900xtx. I 've fixed dozens of these for customers over the past few months. First update to the latest bios for your board, next turnoff driver updates from windows, next disable Fast Startup in Windows, next download and run AMD Cleanup Utility in safe mode, next run Display Driver Uninstaller in safe mode, reboot and install Adrenaline 25.3.1 (i've tested them all and it's the most stable with an xtx card), install the latest chipset driver. Most likely you are good now but If you have any further timeouts, start bumping your clock speed down in adrenaline until you are stable.

 

[Astral Abyss]

Only time I've had issues with my 7900xtx is when it's been a while since I updated motherboard chipset drivers and Windows decided to update 1 or some of the files with generic "updated" drivers. Breaks everything, sometimes even giving me a black screen when Windows loads. Simply installing the latest motherboard drivers straight from AMD has always fixed this for me.

EDIT: Also, timeouts can be caused by another device on the same bus that is holding onto interrupts, especially sound cards. Might want to run some latency testing software to find out if there's a smoking gun culprit.

 

[HockeyJon]

From your signature it looks like you're running a 7900xtx. I 've fixed dozens of these for customers over the past few months. First update to the latest bios for your board, next turnoff driver updates from windows, next disable Fast Startup in Windows, next download and run AMD Cleanup Utility in safe mode, next run Display Driver Uninstaller in safe mode, reboot and install Adrenaline 25.3.1 (i've tested them all and it's the most stable with an xtx card), install the latest chipset driver. Most likely you are good now but If you have any further timeouts, start bumping your clock speed down in adrenaline until you are stable.

Thanks, I'll give that a go.

When you say fixed, how often was hardware to blame, and how often was it corrected via software?

 

[kwikgta]

For the ones that I've worked on, it was software 100% of the time.

 

[THUMPer]

Yeah, are you overclocking anything? Tweaking ram speeds and timing?

 

[kwikgta]

Yeah, are you overclocking anything? Tweaking ram speeds and timing?

My personal card has a 550w bios and is overclocked. My cutomer's cards are not overclocked.

 

[chameleoneel]

Hey guys!

I'm trying to decide if I have to RMA my card at this point because I'm out of ideas.

At this point I've tried DDU driver clear-outs, formatting windows, installing the card in a new system, editing things like TdrDelay in the registry, and capping the shader clock in Adrenalin to not exceed the specified defaults.

I'm reading Windows 11 24H2 in particular has been bad with AMD drivers, especially because you can no longer disable MPO in the registry. Not sure if anyone can corroborate that.

Just wanted to know if anyone has a solution to driver timeouts that actually works. Last AMD card I had was an R9 280X and it was rock solid. I never had a single problem with that card, and I was overclocking it quite aggressively.

So yeah, outside of RMAing my card (or just getting rid of it altogether and returning to Nvidia), I'm out of ideas, but I read online that AMD's drivers lead to a lot of false RMAs and I'd rather not have to go through that process if the hardware is not actually an issue. Not sure if anyone can corroborate that?

In any case, if anyone has a good solution to driver timeouts, please let me know! Thanks!

Another thing, Are any of the GPU's 8 pin power connections daisy chained, or do they all have their own dedicated cable to the power supply?

 

[HockeyJon]

Yeah, are you overclocking anything? Tweaking ram speeds and timing?

Not on this card. I normally overclock everything just because I like overclocking stuff, but when I had my first driver timeout, my assumption was it was overclocking instability, which didn't have me too annoyed because if I choose to run the card out of spec, then that's on me. The problem is that I'm running a cap at the specified boost clock and still getting occasional timeouts.

 

[HockeyJon]

For the ones that I've worked on, it was software 100% of the time.

Interesting. I've been hesitant to go through the RMA process because I've been reading from so many people that this is the case, and I don't even know if it's worth paying for shipping at this point and being without a card for a few weeks only to have MSI come back and be like "looks good, lmao". I would have assumed a hardware problem would manifest itself more commonly and consistently.

Last driver update had it having more timeouts than normal, so I chalked it up to that, but unfortunately I don't know of a way to confirm it's definitely not hardware outside of going through the RMA process. From what I've been reading though, this really seems like a common issue with AMD cards in the current era. Based on the number of threads I see on Reddit and other tech sites, this seems to be a widespread common issue and I have no idea how this doesn't get corrected given that.

Until now, I didn't have much to complain about in terms of AMD's drivers. As I said, my card before my 1070 Ti was an R9 280X that i was overclocking aggressively and never had an issue, so I always assumed people's AMD driver complaints were more from using the product years ago rather than anything new.

How often have you need encountering driver timeouts with your AsRock 7900XTX?

Also, when you said you fixed your customer's cards, I presume you did it with the steps you mentioned earlier? Nothing changed hardware-wise on the card itself?

 

[HockeyJon]

Another thing, Are any of the GPU's 8 pin power connections daisy chained, or do they all have their own dedicated cable to the power supply?

All connectors have their own direct connection to the PSU, so I don't suspect there's a power delivery issue.

 

[chameleoneel]

All connectors have their own direct connection to the PSU, so I don't suspect there's a power delivery issue.

And how old is the PSU?

You've got an X470 mobo. Which means your PSU could be 7+ years old. I've seen plenty of people whom had to get new PSUs around 30 series release. Because they couldn't handle the power and transient spikes of the newer GPUs. 7900 XT and XTX seem similarly sensitive to power. The 3 connector versions, in particular.

All that said, I've also seen a few people RMA their RDNA 3 cards and all was fine after that.

 

[HockeyJon]

And how old is the PSU?

You've got an X470 mobo. Which means your PSU could be 7+ years old. I've seen plenty of people whom had to get new PSUs around 30 series release. Because they couldn't handle the power and transient spikes of the newer GPUs. 7900 XT and XTX seem similarly sensitive to power. The 3 connector versions, in particular.

All that said, I've also seen a few people RMA their RDNA 3 cards and all was fine after that.

I have to check, but it’s not 7 years old. Might be 3. 4 tops. I ended up upgrading it when my mom’s went, so I gave her my older 650W unit and upgraded to accommodate a planned GPU upgrade, which became the 7900XTX.

I might just have to go through with that and call it a day. I’m running out of ideas at this point. I appreciate your input on that. The PSU is an easy one to overlook, but I can’t see it being the case here.

For the folks that were fine after RMA that you know, what was the reported issue with their cards that was solved by RMA?

 

[chameleoneel]

I have to check, but it’s not 7 years old. Might be 3. 4 tops. I ended up upgrading it when my mom’s went, so I gave her my older 650W unit and upgraded to accommodate a planned GPU upgrade, which became the 7900XTX.

I might just have to go through with that and call it a day. I’m running out of ideas at this point. I appreciate your input on that. The PSU is an easy one to overlook, but I can’t see it being the case here.

For the folks that were fine after RMA that you know, what was the reported issue with their cards that was solved by RMA?

Reported issues being: Driver timeouts and/or game crashing and/or system crashing/rebooting unexpectedly.

1. What clock are you capping at? AMD reference clock or your AIB brand's advertised clock?
The worst ones I have seen, people had to cap at AMD's reference "Game Clock" 2300 MHz, to gain stability. Some of them could move up to the reference "Boost Clock", of 2500 mhz. Generally its somewhere around or a bit over 3GHZ, where some of these cards would easily show problems. But, ultimately, capping way down to reference clocks, is what brought actual/near 24/7 stability.

2. The replacement cards had no weird issues with driver timeouts, game stability, etc.


**I edited a couple of times. refresh to see it all.

 

[HockeyJon]

Reported issues being: Driver timeouts and/or game crashing and/or system crashing/rebooting unexpectedly.

1. What clock are you capping at? AMD reference clock or your AIB brand's advertised clock?
The worst ones I have seen, people had to cap at AMD's reference "Game Clock" 2300 MHz, to gain stability. Some of them could move up to the reference "Boost Clock", of 2500 mhz. Generally its somewhere around or a bit over 3GHZ, where some of these cards would easily show problems. But, ultimately, capping way down to reference clocks, is what brought actual/near 24/7 stability.

2. The replacement cards had no weird issues with driver timeouts, game stability, etc.


**I edited a couple of times. refresh to see it all.

1. I was capping it at the specified boost clock speed for the card itself, not the AMD reference clocks. Perhaps I can try that, although I'd be surprised if that solves it since I have been able to hold clocks stable over the rated boost clock for extended periods of time.

2. Interesting to hear. I'd be curious to see what changes were made to the card to enable the stability.

 

[chameleoneel]

2. Interesting to hear. I'd be curious to see what changes were made to the card to enable the stability.

I think its probably just that validation wasn't tight enough on binning chips for the OC models which boost near/over 3GHZ, by default.

 

[HockeyJon]

From your signature it looks like you're running a 7900xtx. I 've fixed dozens of these for customers over the past few months. First update to the latest bios for your board, next turnoff driver updates from windows, next disable Fast Startup in Windows, next download and run AMD Cleanup Utility in safe mode, next run Display Driver Uninstaller in safe mode, reboot and install Adrenaline 25.3.1 (i've tested them all and it's the most stable with an xtx card), install the latest chipset driver. Most likely you are good now but If you have any further timeouts, start bumping your clock speed down in adrenaline until you are stable.

Just want to say thanks. I tried this out and, so far, so good. I think what I hadn't done was the AMD Cleanup Utility in the past. I had done DDU, but corroborating what you wrote here, I have read as well that Windows installing a driver from a fresh install before you have a chance to get the AMD one in can cause conflicts and driver timeouts, and I think this step might have actually helped address it. I'll be putting the RMA on hold for now, thanks again!

Question for you since you obviously have a lot of experience dealing with this. Have you found features like FMF or FSR were contributing to driver timeouts? I have read from some that they suggest this can also be a cause.

 

[kwikgta]

Just want to say thanks. I tried this out and, so far, so good. I think what I hadn't done was the AMD Cleanup Utility in the past. I had done DDU, but corroborating what you wrote here, I have read as well that Windows installing a driver from a fresh install before you have a chance to get the AMD one in can cause conflicts and driver timeouts, and I think this step might have actually helped address it. I'll be putting the RMA on hold for now, thanks again!

Question for you since you obviously have a lot of experience dealing with this. Have you found features like FMF or FSR were contributing to driver timeouts? I have read from some that they suggest this can also be a cause.

You're welcome bud. FMF and FSR made no difference to me as far as driver timeouts.

 

[HockeyJon]

You're welcome bud. FMF and FSR made no difference to me as far as driver timeouts.

Good to know. I’ll keep playing around with it, but this hopefully has resolved the issue. I assume if I had a hardware problem I’d have already had another timeout by now, but we’ll see!

 

[HockeyJon]

Unfortunately it looks like I spoke too soon. Another driver timeout today. Was fun while it lasted .

 

[Astral Abyss]

You really need to run LatencyMon or something similar to see what's responsible for the driver timeouts.