Unrecognized BSOD error - No log file saved, system shut down right away.
Unrecognized BSOD error - No log file saved, system shut down right away.
I've been trying to find a stable overclock on an i5 2500k ASRock P67 Extreme4 for some time and have managed to bring it down from 4.7 to 4.5. Under light or no load, I experience random crashes that flash a BSOD screen before the system stops abruptly. The unusual aspects are:
1) The computer doesn’t shut down after a BSOD.
2) The BSOD fails to properly dump data, even when using a small dump option in System Failures.
3) Suppress Restart is turned on in the System Failure settings.
This reminds me of others I know. I’ve been using it for about 10 hours with moderate use and some Prime95 testing, nothing out of the ordinary. But suddenly it happened again while idling. I’ve also seen a few unrelated BSODs during 4.6–4.7GHz runs, but this one is different now.
Any suggestions? What might be causing these issues? How can I test for them?
What operating system are you running? Have you switched to Windows 10? Did you perform a fresh installation of the system? It might be related to software, and a clean install could resolve the issue. Another possibility is a low-quality power supply unit or a failing PSU. The problem could stem from the PSU not delivering adequate voltage to the CPU—a significant drop causing the BSOD and subsequent shutdown.
What operating system are you running? Have you switched to Windows 10? Did you perform a fresh installation of the OS? It might be related to software, and a clean install could resolve the issue. Another possibility is a low-quality power supply or a failing PSU, which might not be delivering the correct voltage to the CPU—leading to the BSOD and subsequent shutdown.
I’m currently using Windows 10 after upgrading from version 7. I ran Memtest86 overnight while maintaining the original settings, and everything worked fine without any crashes. It’s been a while since I started with a clean setup, so this could be my next step. I also increased the overclock slightly (to 4.5-4.7) right after the upgrade to 10, which makes it hard to rule out other factors.
The power supply might be the problem, but I’ll explore cheaper alternatives first. There are reports of sudden shutdowns in Windows 10, often linked to NVIDIA drivers. Still, I plan to perform a full system refresh.
It might be connected to software. Many users experienced different computer issues after upgrading from Windows 7/8 to Windows 10 without performing a clean installation. The fragments in files and drives can follow along and lead to various strange problems. I would begin with a clean install of Windows 10.
Returned to a Win7 environment (fresh installation) still experiencing crashes. Made several tweaks to the overclock parameters and discovered an unexpected approach.
Here’s my current configuration for reference:
ASRock P67 Extreme4
Intel i5-2500k
Corsair Vengeance 2x4gb DDR3 1600 cas8 @ 1.5v
Corsair CMPSU-550VX 550W
Xigamtek Gaia CPU Cooler (Air)
I followed a guide on offset voltage overclocking (http://www.overclock.net/t/1198504/compl...ck-edition). What stood out during testing was that stress tests generally succeeded, but idle or light loads would fail. This pattern appeared consistently across most overclock adjustments, even when lowering the multiplier slightly.
The voltages were unusually high and temperatures were similarly elevated. The notion that more voltage was necessary to prevent crashes seemed unrealistic and risky without damaging the processor.
I had some uncertainty about LLC settings and their importance. I began with 50% LLC and adjusted it over time. Eventually, I settled on a moderate to high LLC setting. This helped balance load voltages while keeping idle voltage low.
After reaching a stable 4.4-4.5 GHz overclock at the lowest possible turbo and offset increments (+.004, +.005), my voltages stabilized around 1.4V under load (80-85°C). This was unsatisfactory and still caused crashes during idle.
Rather than trusting the guide, I evaluated LLC more objectively. A moderate to high LLC value seemed to push load voltages higher while maintaining a low idle voltage. I decided to disable LLC entirely and let the system handle vDroop naturally. This proved to be the correct decision.
Now I’m achieving a stable 4.5 GHz overclock with no crashes. Idle voltage peaks at about 1.32V, but only briefly before speed adjustments take effect. My load voltage sits around 1.29V (roughly the 65th percentile) and performs well in IBT and Prime95. Further improvements may be possible, but upgrading cooling would be necessary to manage the temperatures. Currently, I’m experiencing low to mid-70°C during intensive loads.
Looking back, this makes sense: higher LLC values increase load voltages while keeping idle stable. The smooth transitions during light loads were what prevented crashes during full-load tests. With LLC off, everything operates more reliably. I’ll continue testing and using it daily, but I’m much more confident now that the results align with expectations.
Thanks for your support, Redneck5439.