Question Mysterious crashes of 5900X & 3090. PC starts only through bios override.

Mar 9, 2024
4
0
10
Well then.
I have a very interesting issue. After buying the CPU and installing it, for some time, it worked really fine, reaching its highest frequency, without any issues at all.
Some time later, PC started rebooting by itself for no reason. The bios of the motherboard allows to boot override with Windows boot manager(somewhy the boot manager located on completely different drive, 24UET0, but choosing the bios override to boot from 24UET0 redirect to 00PXH0 and boot from it, so the task manager says that the 00PXH0 is a system drive when bios override is set to 24UET0. Arch Linux contains memtest86+ in it, so there was an attempt to start it three times, booting directly without override and with it. Booting directly caused 593 errors, with override test passed. Third test caused the PC to reboot(direct booted from USB). Also noticed that override way limits the CPU speed to 3.7GHz.) so that's how the PC works till now. Now another problem appeared. During some gaming or other GPU-intense tasks, the display turns off with 0x00000116 (0xffffd20e14b68010, 0xfffff8017fbb6730, 0xffffffffc000009a, 0x0000000000000004), and there is a chance that or the GPU RGB would stuck on single color, or the RGB would keep working. Tried to start the GPU benchmark on arch linux, encountered only the GPU RGB stuck there. Windows, on the other hand, also turns the display off, but the RGB still works (and there is no BSOD somewhy). Noticed that setting a manual fan control helps in some way(that way, both gpu fans are set to 100% so the furmark passes 20 mins, no errors). Also passed the VRAM vulkan test, not sure if 5 minutes is enough, but no errors encountered.
The last crash with RGB keep changing colors was i playing a game for some time, exiting and restarting it, then the crash happened almost instantly after loading.
Help.
CPU Ryzen 9 5900X
GPU Palit RTX 3090 24576Mb GAMINGPRO OC (NED3090S19SB-132BA)
RAM Kingston HyperX FURY Black HX426C16FB3K2/16
Motherboard asrock x570 extreme4
BIOS v3.90
win10 64bit 22h2
Drive: WDC WDS100T2B0C-00PXH0,M.2 & WDC WD10JPCX-24UET0,SATA
PSU thermaltake BX1 750w
Latest GPU drivers(551.76)
 
Well then.
I have a very interesting issue. After buying the CPU and installing it, for some time, it worked really fine, reaching its highest frequency, without any issues at all.
Some time later, PC started rebooting by itself for no reason. The bios of the motherboard allows to boot override with Windows boot manager(somewhy the boot manager located on completely different drive, 24UET0, but choosing the bios override to boot from 24UET0 redirect to 00PXH0 and boot from it, so the task manager says that the 00PXH0 is a system drive when bios override is set to 24UET0. Arch Linux contains memtest86+ in it, so there was an attempt to start it three times, booting directly without override and with it. Booting directly caused 593 errors, with override test passed. Third test caused the PC to reboot(direct booted from USB). Also noticed that override way limits the CPU speed to 3.7GHz.) so that's how the PC works till now. Now another problem appeared. During some gaming or other GPU-intense tasks, the display turns off with 0x00000116 (0xffffd20e14b68010, 0xfffff8017fbb6730, 0xffffffffc000009a, 0x0000000000000004), and there is a chance that or the GPU RGB would stuck on single color, or the RGB would keep working. Tried to start the GPU benchmark on arch linux, encountered only the GPU RGB stuck there. Windows, on the other hand, also turns the display off, but the RGB still works (and there is no BSOD somewhy). Noticed that setting a manual fan control helps in some way(that way, both gpu fans are set to 100% so the furmark passes 20 mins, no errors). Also passed the VRAM vulkan test, not sure if 5 minutes is enough, but no errors encountered.
The last crash with RGB keep changing colors was i playing a game for some time, exiting and restarting it, then the crash happened almost instantly after loading.
Help.
CPU Ryzen 9 5900X
GPU Palit RTX 3090 24576Mb GAMINGPRO OC (NED3090S19SB-132BA)
RAM Kingston HyperX FURY Black HX426C16FB3K2/16
Motherboard asrock x570 extreme4
BIOS v3.90
win10 64bit 22h2
Drive: WDC WDS100T2B0C-00PXH0,M.2 & WDC WD10JPCX-24UET0,SATA
PSU thermaltake BX1 750w
Latest GPU drivers(551.76)
Shop for a quality 1KW psu.
 
Can you please advice me one? And are you sure that it would fix anything? i dont think that GPU that can't reach 84C and just disconnects on 71C can comply to the PSU.
AIDA 64 test somewhy passes.
Recco?.....no.
Shop for a unit with at least a 7 yr warranty and then read the reviews.

The 3090 can have some big power spikes so you need a psu that can handle that.
 
Recco?.....no.
Shop for a unit with at least a 7 yr warranty and then read the reviews.

The 3090 can have some big power spikes so you need a psu that can handle that.
but the entire build worked as intended a year ago... you mean, PSUs degrade? even the reboots in memtest86+ happen just because of the PSU when gpu isnt used? Even if so. CPU working at 3.7ghz and the same CPU working at 4.7GHz is not THAT much of a difference, but at 4.7 it fails or reboots, on 3.7 it passes fine. 750W should handle it either.
 
but the entire build worked as intended a year ago... you mean, PSUs degrade? even the reboots in memtest86+ happen just because of the PSU when gpu isnt used? Even if so. CPU working at 3.7ghz and the same CPU working at 4.7GHz is not THAT much of a difference, but at 4.7 it fails or reboots, on 3.7 it passes fine. 750W should handle it either.
A proper psu just gets one thing out of the mix.
It might not be the answer to your problem.