Computer shutdowns only while gaming, not stress testing

[Rikku]

Allo,

I'm having a problem, and it's making me nuts. When playing Baldur's Gate 3, and Once Human, (The only two that have caused it thus far) my computer will randomly shut down. Not crash, just turn off. I can play other games just fine so far, Kingdom Hearts 3 (On Epic), Rabi-Ribi, Sea of Stars, Palia, etc. without issue.

I've run all manner of stress tests (OCCT, all tests, some repeatedly; Prime95, Furmark, MemTest64) on the entire system, as a whole and individual components, and nothing happens. It just passes the tests.

I've been monitoring the temps, and while they're warmer than I'd like, there's nothing alarming (Not much I can do about it, it's a physical problem with the space the computer is in and the layout of the apartment, there's a bit of recirculation, but it's still well within limits)

When nothing was showing any defects, I sort of defaulted to the PSU. I got a new one, and the issue persists.

I've reloaded Windows as well, and that did nothing. My next step is to clear the CMOS, but I just keep forgetting to do it when it's a good time.

Specs:

Ryzen 7800x3D
Gigabyte X670 Aourus Elite AX
BeQuiet Straight Power 1200W PSU (Single rail) (Old one was BeQuiet Dark Power Pro 11 1000W)
64GB Corsair Vengeance DDR5
RTX 4090
Various NVMe drives and SSDs

I'm at my wit's end at this point. I can just start buying new components, but I'd rather just find out what's happening and fix it. Has anyone had this issue before? Was there a resolution?

[[-iD-]]

oh an my bad its manuel, "Manuel@NVIDIA"
he is the one who handles the drivers.
google wont search the forum so tomorrow.
dis guy.
https://x.com/manuelguzman

[Freqsync]

Attempt, to run the games having issues with,
WIthout the GPU slotted
WIth only 1 ram slotted
Without WiFi activated
OS storage device with Storage with steam game
Keyboard mouse
No headphones, controllers, speakers, printers, usb, sd card reader, or the cat water food in the room,smoking/vaping,swordplay,banging of cymbals,own drinking eating,will add more for mere amusement.

After a safeboot(s) anti virus and OS /sfc and the others not sure of windows 11 exacts since I still play in dos... If happened to not update manually go through the channels to windows store and settings. You can switch off tiles on start and configure how many cosmetic mouse trails embossed fonts and all that to off whole list of things to do.
After .....something you concluded needs to be included whether switched on for prevention of a shutdown there are debug modes that create those crash files maybe in this post can describe where they might be control panel program and files advanced settings, windows features, core isolation etc etc

and INCREASE PAGEFILE if you have it off to times four your ram for the time being still uncertain how many slots are being used and if 1 slotted would be of the amount of ???

[Bad 💀 Motha]

引用自 -iD-：

引用自 Bad 💀 Motha：

Said what?
That NVIDIA sneaks into our PCs and tweaks NVIDIA CP? Do you have a clue what you're even talking about there. Cause that's basically what you said in a nutshell

they don't sneak they push a package that updates the configs for all your profiles and fixes incompatibility.
you agreed to this in the eula.
one of these is it and its prob the 300ms ones.
https://i.imgur.com/ozO9Hb7.png

No such thing ever happens; you're referring to game ready driver updates; to which happens automatically if you have GFE installed, which I never do. Just stop.

Your screenshot showing the Nvidia connections is just telemetry data. I disable all of that as well; I'm not a newb.

[[-iD-]]

引用自 Bad 💀 Motha：

引用自 -iD-：

they don't sneak they push a package that updates the configs for all your profiles and fixes incompatibility.
you agreed to this in the eula.
one of these is it and its prob the 300ms ones.
https://i.imgur.com/ozO9Hb7.png

No such thing ever happens; you're referring to game ready driver updates; to which happens automatically if you have GFE installed, which I never do. Just stop.

Your screenshot showing the Nvidia connections is just telemetry data. I disable all of that as well; I'm not a newb.

no i removed all telemetry from nvidia drives except the WHQL check.
yes you are.
"Hello everyone. We will be pushing a fix for the following bug to users via an over the air update in the coming days.

[Cyberpunk 2077] In game map may display corruption [3829994]

If you do not receive the update or if you wish to apply the fix sooner, you may do so by following the instructions in the KBA below: https://nvidia.custhelp.com/app/answers/detail/a_id/5400/"

-ManuelG

[_I_]

引用自 -iD-：

引用自 Bad 💀 Motha：

No such thing ever happens; you're referring to game ready driver updates; to which happens automatically if you have GFE installed, which I never do. Just stop.

Your screenshot showing the Nvidia connections is just telemetry data. I disable all of that as well; I'm not a newb.

no i removed all telemetry from nvidia drives except the WHQL check.
yes you are.
"Hello everyone. We will be pushing a fix for the following bug to users via an over the air update in the coming days.

[Cyberpunk 2077] In game map may display corruption [3829994]

If you do not receive the update or if you wish to apply the fix sooner, you may do so by following the instructions in the KBA below: https://nvidia.custhelp.com/app/answers/detail/a_id/5400/"

-ManuelG

nice, so you re-wrote your own driver to do something that orig driver never does in the first place?

it would be much easier to just uninstall gfe
or not use the beta nvidia driver, which is never a good idea anyway unless it has a critical fix that is not yet pushed to public driver ver

btw, your link is broken, or manualg does not exist

[[-iD-]]

引用自 _I_：

引用自 -iD-：

no i removed all telemetry from nvidia drives except the WHQL check.
yes you are.
"Hello everyone. We will be pushing a fix for the following bug to users via an over the air update in the coming days.

[Cyberpunk 2077] In game map may display corruption [3829994]

If you do not receive the update or if you wish to apply the fix sooner, you may do so by following the instructions in the KBA below: https://nvidia.custhelp.com/app/answers/detail/a_id/5400/"

-ManuelG

nice, so you re-wrote your own driver to do something that orig driver never does in the first place?

it would be much easier to just uninstall gfe
or not use the beta nvidia driver, which is never a good idea anyway unless it has a critical fix that is not yet pushed to public driver ver

no i didn't rewite anything LOOOOL
i put 2 entries into install config
and i removed everything in the folder that has to do with telemetry.

[[-iD-]]

everything you see in an nvidia driver folder is telemetry but these items.
https://i.imgur.com/kwgRCAW.png

[_I_]

引用自 -iD-：

everything you see in an nvidia driver folder is telemetry but these items.
https://i.imgur.com/kwgRCAW.png

lmao, all of that can be deleted, its not the files windows is using, thats the temp setup for the drivers

you think you know so much, but truly are so gullible

on the root of the drive, you probably have an intel folder, thats the same thing you can delete that too

if you take the nvidia driver installer you download from nvidia, and extract it, thats exactly all you deleted after the install was complete and it was too late for your changes to do anything

[[-iD-]]

引用自 _I_：

引用自 -iD-：

everything you see in an nvidia driver folder is telemetry but these items.
https://i.imgur.com/kwgRCAW.png

lmao, all of that can be deleted, its not the files windows is using, thats the temp setup for the drivers

you think you know so much, but truly are so gullible

on the root of the drive, you probably have an intel folder, thats the same thing you can delete that too

if you take the nvidia driver installer you download from nvidia, and extract it, thats exactly all you deleted after the install was complete and it was too late for your changes to do anything

no you can't if you delete anything else, the install with throw an error and kick you out.
i used zip to extract the install exe.

[_I_]

sure you did, you didnt even know the installers were zipped before i told you

[[-iD-]]

引用自 _I_：

sure you did, you didnt even know the installers were zipped before i told you

you can unzip any exe file that is an archive.
this is basic knowledge.
and if you know your stuff you can access any exe without a zip.

[[-iD-]]

引用自 _I_：

sure you did, you didnt even know the installers were zipped before i told you

btw i don't dl drivers from nvidia. and all nvidia drivers unzip to a file of your choice before install.
you can cancel the install at any time

[C1REX]

引用自 Rikku：

I've run all manner of stress tests (OCCT, all tests, some repeatedly; Prime95, Furmark, MemTest64) on the entire system, as a whole and individual components, and nothing happens. It just passes the tests.

Have you done extensive single core stress tests? Gaming is a lot about pushing single cores.
Maybe try Cinebench single core for 30min or more?

Single core stress tests boost clocks higher than multi core tests. These single core tests were also crashing some affected Intel CPUs.

Other than that check GPU hot spot and junction temp. Not just GPU temp that is the lowest.
Additionally try to not use daisy chain connection for your GPU if you are using a 12vhpwr converter. You may need 4 separate 8pin cables.

[[-iD-]]

引用自 C1REX：

引用自 Rikku：

I've run all manner of stress tests (OCCT, all tests, some repeatedly; Prime95, Furmark, MemTest64) on the entire system, as a whole and individual components, and nothing happens. It just passes the tests.

Have you done extensive single core stress tests? Gaming is a lot about pushing single cores.
Maybe try Cinebench single core for 30min or more?

Single core stress tests boost clocks higher than multi core tests. These single core tests were also crashing some affected Intel CPUs.

Other than that check GPU hot spot and junction temp. Not just GPU temp that is the lowest.

cinibench is no good.

prime95 using avx2.0
and for gpu
furmark 2.0

[C1REX]

引用自 -iD-：

cinibench is no good.

Why it's not good.
Do other stress tests stress the hardware more?