Upgraded to RTX 4080. Now crashing.

[Spin-Around Boy]

I upgraded from a GTX 1080 to a RTX 4080 and now I'm getting crashes to the desktop quite regularly. Roughly after 30-mins to an hour.

I get an Unreal Engine report to send after the crash. Wondering if anyone else is experiencing this and possibly notifying of an issue, as I was totally fine with the GTX 1080.

I did a clean driver install. Other games seem to be working OK, so far, so I don't think it's the card... I hope... I even disabled the slight under-volt to my CPU as the 4080 will be demanding a bit more effort from it.

Ryzen 7 5800X (Stock settings)
32GB RAM
Windows 10

[Spin-Around Boy]

Originally posted by simon:

Originally posted by Spin-Around Boy:

I did the virtual memory set exactly as you said. Fixed size. I've tried everything you've suggested above, except rolling back drivers. I've removed under volts and I'm now going to troubleshoot potential memory incompatibilities.

BUT...... Other games have started crashing, now, too... :( So, it's not a Pacific Drive issue, it seems. I have a lot of troubleshooting to do. Hope I don't have to RMA.

I should probably delete this post, soon, as it's nothing to do with Pacific Drive. Thanks for all the suggestions/help, so far, though.

How did you uninstall are install the new GPU drivers when you put the new card in?

I had a GTX 1080 with the same numbered drivers, installed, so just swapped the cards. I've since uninstalled and reinstalled just using Windows Programs and Features uninstaller.

I'll need to give DDU a try.

[byo13]

Sorry if you've already checked this but what is the GPU temperature while running the game?

[simon]

Originally posted by Spin-Around Boy:

Originally posted by simon:

How did you uninstall are install the new GPU drivers when you put the new card in?

I had a GTX 1080 with the same numbered drivers, installed, so just swapped the cards. I've since uninstalled and reinstalled just using Windows Programs and Features uninstaller.

I'll need to give DDU a try.

You should always remove the device from windows device manager before fitting a new device whatever it is. Look in device manager and make sure you dont have two display adaptors listed. Also you should always run DDU if you have an issue with GFX after you install new drivers and or a new card. Make sure you follow the instructions including the part about safe booting your PC as this will enable DDU to completely remove the drivers as windows will not have loaded any of them.

[ʍʀ.ɮɛʋɛʀʟʏ]

Originally posted by Spin-Around Boy:

I upgraded from a GTX 1080 to a RTX 4080 and now I'm getting crashes to the desktop quite regularly. Roughly after 30-mins to an hour.

I get an Unreal Engine report to send after the crash. Wondering if anyone else is experiencing this and possibly notifying of an issue, as I was totally fine with the GTX 1080.

I did a clean driver install. Other games seem to be working OK, so far, so I don't think it's the card... I hope... I even disabled the slight under-volt to my CPU as the 4080 will be demanding a bit more effort from it.

Ryzen 7 5800X (Stock settings)
32GB RAM
Windows 10

Have you already tried uninstalling and reinstalling the game since changing out the card?

Since you say the card appears to be working on other UE5 games the game may do some first run config stuff and doesn't expect you to ever change your graphics card, and now the game is crashing because it expects a different graphics card. Pure speculation on my part

[Spin-Around Boy]

Originally posted by byo13:

Sorry if you've already checked this but what is the GPU temperature while running the game?

My temperatures for GPU max at about 70 degrees and my CPU maxes at about 80 at most, which is within the boundaries for a 5800X. But to make sure I put the CPU into Eco Mode and even with temps only reaching 60 degrees it was still crashing.

Originally posted by simon:

Originally posted by Spin-Around Boy:

I had a GTX 1080 with the same numbered drivers, installed, so just swapped the cards. I've since uninstalled and reinstalled just using Windows Programs and Features uninstaller.

I'll need to give DDU a try.

You should always remove the device from windows device manager before fitting a new device whatever it is. Look in device manager and make sure you dont have two display adaptors listed. Also you should always run DDU if you have an issue with GFX after you install new drivers and or a new card. Make sure you follow the instructions including the part about safe booting your PC as this will enable DDU to completely remove the drivers as windows will not have loaded any of them.

I've usually been fine with things like this with past cards. I checked Device Manager to make sure and I just have the one display adapter.

Originally posted by ʍʀ.ɮɛʋɛʀʟʏ:

Originally posted by Spin-Around Boy:

I upgraded from a GTX 1080 to a RTX 4080 and now I'm getting crashes to the desktop quite regularly. Roughly after 30-mins to an hour.

I get an Unreal Engine report to send after the crash. Wondering if anyone else is experiencing this and possibly notifying of an issue, as I was totally fine with the GTX 1080.

I did a clean driver install. Other games seem to be working OK, so far, so I don't think it's the card... I hope... I even disabled the slight under-volt to my CPU as the 4080 will be demanding a bit more effort from it.

Ryzen 7 5800X (Stock settings)
32GB RAM
Windows 10

Have you already tried uninstalling and reinstalling the game since changing out the card?

Since you say the card appears to be working on other UE5 games the game may do some first run config stuff and doesn't expect you to ever change your graphics card, and now the game is crashing because it expects a different graphics card. Pure speculation on my part

I first verified the game files and then I did a total reinstall, even deleting left behind folders and shader cache files.

** I THINK I may have found the culprit, however. I noticed my RAM was set to 1.35v. I don't remember setting it to that. The default is 1.2v. (edit) It must be the XMP profile that set it to 1.35v **

My RAM is Corsair Vengeance RGB PRO Black 32GB (2x16GB) 3600MHz AMD Ryzen Tuned DDR4 Memory Dual Kit.

Anyway, I just loaded all the defaults into my BIOS/UEFI and had an 80 minute session of the game without a crash. So MAYBE that was the issue. It's hard to say for sure as sometimes it was crashing quickly and sometimes took a bit longer.

Going to be one of those things to keep an eye on.

[simon]

Originally posted by Spin-Around Boy:

** I THINK I may have found the culprit, however. I noticed my RAM was set to 1.35v. I don't remember setting it to that. The default is 1.2v. (edit) It must be the XMP profile that set it to 1.35v **

My RAM is Corsair Vengeance RGB PRO Black 32GB (2x16GB) 3600MHz AMD Ryzen Tuned DDR4 Memory Dual Kit.

Anyway, I just loaded all the defaults into my BIOS/UEFI and had an 80 minute session of the game without a crash. So MAYBE that was the issue. It's hard to say for sure as sometimes it was crashing quickly and sometimes took a bit longer.

Going to be one of those things to keep an eye on.

Well thats progress any way. Hope thats the issue and its fixed :)

[Buzzzz]

Originally posted by Spin-Around Boy:

Originally posted by byo13:

Sorry if you've already checked this but what is the GPU temperature while running the game?

My temperatures for GPU max at about 70 degrees and my CPU maxes at about 80 at most, which is within the boundaries for a 5800X. But to make sure I put the CPU into Eco Mode and even with temps only reaching 60 degrees it was still crashing.

Originally posted by simon:

Anyway, I just loaded all the defaults into my BIOS/UEFI and had an 80 minute session of the game without a crash. So MAYBE that was the issue. It's hard to say for sure as sometimes it was crashing quickly and sometimes took a bit longer.

Going to be one of those things to keep an eye on.

Thats good to hear.


Please let us know if you are from this point forward crash free.

[Spin-Around Boy]

@Buzz313TH

Well, so far I've had two 80-odd minute sessions of Pacific Drive without a crash and a 90-odd minute game of Starfield without a crash.

So, it is looking like the XMP 2.0 RAM profile might have been messing with things. The RTX 4080 Super is understandably demanding more from my CPU and RAM than the GTX 1080 was. It means I'm running my RAM at 2666MHz at 1.2v instead of the full 3600MHz at 1.35v but slower RAM is better than crashing RAM, right? :) Not noticed any significant performance loss, to be honest.

I'm probably going to update my BIOS and motherboard chipset drivers to see if I can get the extra speed back but right now, everything SEEMS stable. Finger's crossed.

[Buzzzz]

Originally posted by Spin-Around Boy:

@Buzz313TH

Well, so far I've had two 80-odd minute sessions of Pacific Drive without a crash and a 90-odd minute game of Starfield without a crash.

So, it is looking like the XMP 2.0 RAM profile might have been messing with things. The RTX 4080 Super is understandably demanding more from my CPU and RAM than the GTX 1080 was. It means I'm running my RAM at 2666MHz at 1.2v instead of the full 3600MHz at 1.35v but slower RAM is better than crashing RAM, right? :) Not noticed any significant performance loss, to be honest.

I'm probably going to update my BIOS and motherboard chipset drivers to see if I can get the extra speed back but right now, everything SEEMS stable. Finger's crossed.

Good find and I am happy for you... Here is an interesting tidbit I found on my end when building this specific computer for DCS... My system is a 12900k, asus 4080 and 64 gigs of ram. When tweaking the system for focused performance specifically for DCS VR, I found that a slower more stable system on paper resulted in a much better performing system when it came to overall smoothness, frame rate stability, input latency and VR latency. An overclock might run a faster framerate in certain unloaded circumstances, but as soon as the system got loaded up, data bottlenecks started occurring and frame rates would become inconsistent and I would start seeing micro-stutters or staccato. My current system runs best and smoothest with a slight underclock of memory, almost stock voltages and stock turbo cpu speeds. On paper, I can get better performance via diagnostic graphs with an aggressive overclock, but the overall gaming experience drops. I think most people get hung up on the numbers and overclock their systems to a stuttering mess.

Cheers and glad you sorted this out.

[Spin-Around Boy]

Originally posted by Buzz313TH:

Originally posted by Spin-Around Boy:

@Buzz313TH

Well, so far I've had two 80-odd minute sessions of Pacific Drive without a crash and a 90-odd minute game of Starfield without a crash.

So, it is looking like the XMP 2.0 RAM profile might have been messing with things. The RTX 4080 Super is understandably demanding more from my CPU and RAM than the GTX 1080 was. It means I'm running my RAM at 2666MHz at 1.2v instead of the full 3600MHz at 1.35v but slower RAM is better than crashing RAM, right? :) Not noticed any significant performance loss, to be honest.

I'm probably going to update my BIOS and motherboard chipset drivers to see if I can get the extra speed back but right now, everything SEEMS stable. Finger's crossed.

Good find and I am happy for you... Here is an interesting tidbit I found on my end when building this specific computer for DCS... My system is a 12900k, asus 4080 and 64 gigs of ram. When tweaking the system for focused performance specifically for DCS VR, I found that a slower more stable system on paper resulted in a much better performing system when it came to overall smoothness, frame rate stability, input latency and VR latency. An overclock might run a faster framerate in certain unloaded circumstances, but as soon as the system got loaded up, data bottlenecks started occurring and frame rates would become inconsistent and I would start seeing micro-stutters or staccato. My current system runs best and smoothest with a slight underclock of memory, almost stock voltages and stock turbo cpu speeds. On paper, I can get better performance via diagnostic graphs with an aggressive overclock, but the overall gaming experience drops. I think most people get hung up on the numbers and overclock their systems to a stuttering mess.

Cheers and glad you sorted this out.

Appreciate the insight and I think you're absolutely right. I would much rather smoother, better frame pacing than sheer higher, average numbers. Better 1% and 0.1% numbers.

A little bit of me did get the RAM based on "bigger number better" but also that it was on a good deal at the time. I don't even use the RGB lighting on it. Haha.