Can video memory be managed manually?

[AWT]

I am experiencing random crashes on certain titles (Just Cause 2, Far Cry 5/6, The Division 2, Red Dead Online; sometimes they run for 4h with no issues, but other times they crash 10 minutes into the game. Most other games run fine).
When I play Wolfenstein II, sooner or later I get a warning that the video driver has run out of memory. Usually, when I choose "accept and ignore further warnings", the game continues to run fine, but I do very rarely get crashes on that game too.
Given that my card has 10GB VRAM (RTX 3080, full PC specs in my profile), how can this be happening (most users use cards with less memory, according to Steam tech survey).
Any ideas how this could be addressed manually? I guess this could be what's causing the crashes.
According to Windows Reliability Report, the game that crashes most frequently (The Division 2) stopped communicating with Windows.
(Bonus question: According to CPUID HWMonitor, my Ryzen 1700 clocked up to 4.7GHz while playing last night - but I have disabled core performance boost and set the target CPU clock speed to 3.7GHz manually. Is this a reading error, or what else could be causing this?)

[Bad 💀 Motha]

Those crashes wouldn't bave anything to do with vram which for your specs is more then enough.

If one moment a game (pick one) runs fine for 4 hours and then randomly crashing 10-15 minutes in the next go around. Then it's something else. Could be settings for cpu and ram in the bios. Corrupted game files, corrupted OS, faulty ram or drive. Bad thermals... etc. Could be gpu driver as well.

[Illusion of Progress]

I am not aware of ways to manage VRAM, but that's not to say they don't exist.

Can you clarify what it is that showing these messages? Windows doesn't give you a message when VRAM is full. I've never seen a game or application or driver do this either, and I'm fairly sure they wouldn't even be aware of it. When VRAM is exhausted, the system can use RAM as shared VRAM. Maybe when THAT gets full it does, but if so, I'd find it confusing how it gets to that point and you aren't aware of it. Performance will typically drop quite a bit if RAM needs to be used in place of VRAM.

So what is showing this message (or messages) and what is (are) the exact message(s)?

The Ryzen 7 1700X boosts "up to 3.7 GHz" so I'd presume 4.7 GHz isn't accurate. If you adjusted default behavior, it may occurring as some result of that.

[Wichtelman]

you could run some benchmarks like userbenchmark or passmark etc to check if some part of your pc operates in a strange way or below expectation while hwmonitor runs in the background

[Overseer]

No VRAM is locked away and hard to deal with. I just hope you got a legit 3080 from a proper store and not one of these firmware scams. Make sure performance lines up and everything is on default settings to avoid any confusion. A baseline is important to find the problem.

[AWT]

Thermals are fine (CPU max 65°C, GPU about 80°C in heavy gaming), the RTX is a legit one (purchased/sold from Amazon directly, no Marketplace offer or similar questionable source). I switched from a GTX 1070, and the performance change is about as expected.

The message about video RAM running out only comes up in Wolfenstein II if I play it for a while, it's in the game's usual font etc for settings etc.

The new Ubisoft games (which are mostly affected, besides JC2 and RDR2 which are also both known for weird crashes/issues) usually crash without any notification/error message (on rare occasions, the Ubisoft launcher asks me to send an error report), I have to go to Control Panel->Security & Maintenance->Reliability Monitor to get any details about the crash.

In Windows reliability monitor, I have the following entries, the first is rare, the second multiple times each time I play the game:

Hardware Error
Description
A problem with your hardware caused Windows to stop working correctly.
Problem signature
Problem Event Name: LiveKernelEvent
Code: 141
Parameter 1: ffff878b04c6b010
Parameter 2: fffff802266a16d0
Parameter 3: 0
Parameter 4: 472c
OS version: 10_0_19045
Service Pack: 0_0
Product: 256_1
OS Version: 10.0.19045.2.0.0.256.48
Locale ID: 2057

Problem: Stopped Responding
Description
The program TheDivision2.exe version 1.0.0.0 stopped interacting with Windows and was closed.
Process ID: 3dc8
Start Time: 01d955020a8834e7
Termination Time: 11
Application Path: E:\Program Files\Ubisoft\Tom Clancy's The Division 2\TheDivision2.exe
Report Id: 2b6815ce-304d-4037-ad08-2c5e6e9666e9
Faulting package full name:
Faulting package-relative application ID:
Hang type: Unknown

Thanks for the hints, will run a benchmark app and report back when I did it.
Most games run fine (I have 300+ in my collection), it's mostly new Ubisoft titles (and a few specific other ones) that cause problems. To mention some relatively new games: Isonzo, Mafia (Remastered), AC Odyssey, and various others never cause any issues or crashes.

[_I_]

what are you using to see temps?

alt+tab takes long enough for temps to drop 20+c

try with hwmonitor, it logs min/max

[MoonC A T]

I had a card that did this. Had nothing to do with temps, etc. It was just a "bad" card.

I solved it by underclocking the card in MSI Afterburner.

[AWT]

I'm using CPUID's HWMonitor to monitor temps (if I want to check temps, I leave the program running in the background whilst gaming, and then check the results when I finish playing for the day).
I also have Core Temp running as an alert programme (which gives an alarm and/or shuts down the system at a set temp, as I don't want the CPU to exceed AMD's prescribed maximum). It also allows me to verify whether temp readings in HWmonitor are accurate, as it has a CPU min/max temp display too (and always runs in background as it's in the autostart list).
What's really strange is that all cores report a maximum clock speed of 4.7 GHz in HWmonitor, even though I manually set the CPU to 3.7 GHz in BIOS, and disabled core performance boost (as even at default base clock settings, having Core Performance Boost active results in clock speeds of 4GHz or even higher, which draws too much voltage for the CPU's own good - I want it to last until an upgrade to a 3DFX model becomes necessary).

[Bad 💀 Motha]

Don't have it shutdown the system just because the system hits a certain temp. That can do more damage then good. Your Motherboard BIOS as well as Firmware onboard the GPU card already has that scenario covered as they will thermal throttle the cpu and gpu to prevent overheating and thus damages from that.

[Obsessive Power]

Don't know about The Division 2, but wolfenstein 2 will throw up a message about VRAM. It shouldn't cause the game to crash though.

[AWT]

Wolfenstein 2 crashes very, very rarely, but I do frequently get the games VRAM message. That's why I guessed the VRAM issue could be what's causing the issues in Ubisoft games. I'll report back when I did the benchmark test.

[Bad 💀 Motha]

If you are getting in-game messages about VRAM running low then you either running the game too long (good to exit and restart game every few hours) or you have the graphics set too high for your hardware.

Even sometimes a game exit does not fully flush the VRAM. But you can do that via OS User Log-Off. Then log back in again.