Update 17 - frequent crashes and sometimes BSOD
- Thread starter rootsrat
- Start date
It could well be the game. I haven't done any Thargoid content recently, so my lack of issues isn't very telling.
Still, should be pretty easy to fire up y-cruncher or TM5 for a few hours to rule out obvious memory controller and RAM issues.
I don't get VSOD/crashes, but transitioning between on-foot (or SRV) and ship and back takes many times longer than before the update. 40-50 seconds instead of 5.
This is the only thread that mentions update 17 so i hope you don't mind me using it?
I have an issue with Discovery scanner. I enter the frequency scan mode (whatever that was called) and it puts me into somekind of "how to use" mode and i can't find any keys that actualy do something and mouse doesn't work. Is this something new or is it something related to the fact that i just had to replace the hard drive and re install the game. Lots of the key bindings were still as i remember but i had to reconfigure a lot.
I have an issue with Discovery scanner. I enter the frequency scan mode (whatever that was called) and it puts me into somekind of "how to use" mode and i can't find any keys that actualy do something and mouse doesn't work. Is this something new or is it something related to the fact that i just had to replace the hard drive and re install the game. Lots of the key bindings were still as i remember but i had to reconfigure a lot.
Not had any issues or changes in behaviour with the FSS so I would suspect you missed something in the huge number of bindings.
Edited to stay on topic, I have not had any changes in the type or frequency of crashes since U17 and honestly can’t remember when I last had a BSOD.
Last edited:
Any luck?
rootsrat
Volunteer Moderator
No, I lost my other PC due to PSU giving up, which I used to use for some side work while wife works on the main one during the day. So the side stuff I was usually doing on it now moved to evenings to the main PC, which means I'm out of time (and luck, it seems
)Memory related BSODs aren't always indicative of issues with the memory modules themselves (though that is a distinct possibility); memory controller and CPU cache/ring are also distinct possibilities. Board and power delivery are less likely, but cannot be ruled out without testing.
Crashes while streaming while using NVENC could be an issue with the card itself. NVENC, in addition to using the GPU's encoder block, is also an extra load on the card's memory controller and PCI-E interface.
Run y-cruncher's FFT, VST, and N63 tests. If there are obvious issues with the CPU or memory those should reveal them within a few hours. If you do encounter errors, try disabling XMP on the memory, or manually setting a lower speed, and check again.
I also suggest running memtest_vulkan while keeping an eye on GDDR6X junction temps to see if there are any clear problems with the GPU's memory controller or VRAM.
If neither test produces any errors independently, you may need to load the GPU while testing CPU/system memory. Most 3080s will dump a lot of heated air right on to system memory, while significantly increasing load on the PSU, so combination testing that simulates worst case holistic loads may be the only way to isolate problems. I usually use FurMark as a GPU load when trying to isolate such main memory issues because it can run GPUs right at their power limit while having very little overhead (that might reduce CPU/memory load).
Crashes while streaming while using NVENC could be an issue with the card itself. NVENC, in addition to using the GPU's encoder block, is also an extra load on the card's memory controller and PCI-E interface.
Run y-cruncher's FFT, VST, and N63 tests. If there are obvious issues with the CPU or memory those should reveal them within a few hours. If you do encounter errors, try disabling XMP on the memory, or manually setting a lower speed, and check again.
I also suggest running memtest_vulkan while keeping an eye on GDDR6X junction temps to see if there are any clear problems with the GPU's memory controller or VRAM.
If neither test produces any errors independently, you may need to load the GPU while testing CPU/system memory. Most 3080s will dump a lot of heated air right on to system memory, while significantly increasing load on the PSU, so combination testing that simulates worst case holistic loads may be the only way to isolate problems. I usually use FurMark as a GPU load when trying to isolate such main memory issues because it can run GPUs right at their power limit while having very little overhead (that might reduce CPU/memory load).
Last edited:
rootsrat
Volunteer Moderator
Thanks, I'll run all those test as soon as I find some time.
So far I have also found out that there was a newer version of my BIOS, so updated that.
I'll admit I've not cleaned my PC for a good while and it's long overdue now, so might do that first and resit everything in their slots.
One thing - heat should not be an issue i think. I have replaced thermal pads on the GPU some time ago with the best available ones on terms of thermal efficiency and I think I have a rather good airflow inside with a spacious case.
Normally but there are ways to get windows to BSOD I done it many times back in the day.
Definitely run a hardware test on the PC as it happening in another program screams imminent failure, the reason for livestream as has been suggested due to hardware stress test. Maybe running a intensive benchmark tool as well to see if they can get a BSOD
rootsrat
Volunteer Moderator
X4 foundations, and that was during a keybinds setup
@rootsrat
I'm no expert in such things but i do know you were creating and using mods in Starfield and tomshardware suggests that if you (or others) don't use the official bethesda api, it can cause BSOD.
I don't know how much that might relate to your current situation but maybe worth a thought, unless you can rule it out.
I'm no expert in such things but i do know you were creating and using mods in Starfield and tomshardware suggests that if you (or others) don't use the official bethesda api, it can cause BSOD.
I don't know how much that might relate to your current situation but maybe worth a thought, unless you can rule it out.
Dont just clean it, unseat the memory modules are plug them back in as well. A little isoprop on a cotton bud to clean the contacts
And as people above said, use a memtest type program to exercise the hardware.
But it could be the gpu, it could be a failing motherboard component, it may be the install. Its very hard to isolate.
And as people above said, use a memtest type program to exercise the hardware.
But it could be the gpu, it could be a failing motherboard component, it may be the install. Its very hard to isolate.
rootsrat
Volunteer Moderator
All I had time today was the basic, 5-min + about 15 mins of extended test from memtest_vulkan, where the junction temp maxed at 84*C and 4 iterations of y-cruncher's FFT, VST, and N63, which did not return any errors.
I should be able to find a few hours to run these properly during the week, so we'll see.
By the way, thanks for all your suggestions everyone, I really appreciate them
I used to repair PCs, one quick trick back in the day was to lift the desktop 5 cm over the desk and let it fall down (while powered off). The jolt would juggle the connections and sometimes it started working again. But observed more professionally, I was surprised at how many times intermittent problems could actually be fixed by reseating cards and components. I'd start with cleaning out any dust, reseat the RAM and possibly the GPU. Reseating the CPU would be a bigger operation.. Also worth trying in a situation like this is another power supply.
It would also be handy to have an easy way to recreate the problem, like a specific point in a game or a combination of stress tests.
It would also be handy to have an easy way to recreate the problem, like a specific point in a game or a combination of stress tests.
Ah, the old Miggy drop.