System failure under stress #2 along with ongoing problems that have appeared recently or persist over time.
System failure under stress #2 along with ongoing problems that have appeared recently or persist over time.
specs: cpu model i7 3770 (with a stock cooler from i5 3570 that has one leg damaged, leading to idle at 40c and load at 80c) motherboard: Asus p8b75-m (featuring a broken PCIe X16 lever and several bent CPU pins) gpu options – Sapphire RX 570 nitro+ (second hand) or RX 5700 (also second hand) ram: 2x4 DDR3 1600MHz silicon power + 1x8 DDR3 1600MHz SK Hynix psu: Thermaltake TR2 S 500W case: Aerocool CS-107 details about the problems: 1: Mice freeze or stop responding suddenly – this is the latest occurrence, affecting Logitech M90 and R80 models. It tends to happen during gameplay, possibly due to timing, and occurs more often when connected via USB 2.0/3.0 ports on both the motherboard and case. 2: BIOS displays incorrect CPU voltages (~0.2v) – likely caused by bent CPU pins on the motherboard. Voltages match correctly during boot into OS. 3: Post-shutdown behavior – the system ignores the power button until all power is cut (PSU shut down or case unplugged), and capacitors lose charge. This also impacts restarts, causing the PC to power off before restarting. The problem persists across three different PSUs (400W Gamemax Gm-400-8, Thermaltake, Chieftec Smart GPS 650A8 from another PC). 4: After roughly five minutes of GPU operation at around 40°C, the PC shuts down instantly. This happens with both GPUs, running at minimum voltage/clock speeds. The issue occurs between Linux and Windows environments. Stress tests on CPU and memory showed no problems. Running games without GPUs hasn’t caused crashes. GPUs have adequate cooling. Previously, using a Chieftec Smart GPS 650A8 PSU confirmed the PSU wasn’t the cause, but the problem remained. Two years prior, the 570 GPU could handle ~60°C before crashing, but later it began failing at lower temperatures, requiring underclocking/undervolting. Now it needs 800MHz, 750mV, -50% power limit, and maximum fan speeds just to run demanding games – sometimes it still crashes. The 5700 was purchased about two years after the 570 from a different seller in another country, and it exhibited identical issues.
Mouse performance drops when under stress. Possible causes include electrical issues or weak USB signals, possibly due to a dirty connector. If turning on the power only works after fully discharging, the problem may lie with the MOBO power circuit. Your PC appears to be deteriorating gradually. Multiple hardware components are affected, especially the motherboard—issues like bent CPU pins, damaged PCIe slots, or power delivery faults are likely. GPU instability under load often stems from unstable PCIe power or VRM failure. Swapping the motherboard seems like the best immediate solution to address these widespread problems.
I mean... the cooler is damaged, the CPU socket is bent, and the pins on the socket might be misaligned. It’s possible the machine has been running well for years but is now in its toughest phase. You can find a cheap, low-quality motherboard for around 30 dollars on AliExpress that fits that old CPU—perfect for that kind of setup. Here are two options: one new cooler that still works but isn’t great (MACHNIST H61, 25€ with shipping), and another used one that’s still functional but worn out. Both are available on AliExpress at under 60 dollars if you're willing to search a bit.
The cooler is damaged, so this computer will perform poorly. Consider swapping it out or upgrading the system. Keeping it running with a broken part is risky and damaging.
Also not entirely accurate—ideally you should always pair correctly matched RAM sticks. For instance, choose either two 8GB modules or four 4GB modules to ensure proper dual-channel performance. Stick with the same type of RAM in every slot for better consistency. Different manufacturers may have varying timing specifications, which can cause your system to operate at its slowest speed if mismatched. For example, a CL20 1600MHz chipset and a CL30 1600MHz chipset would force all sticks to run at the slower 1600MHz standard, not the actual clock speeds.
I reduced the fan speed on my 5700 from 100 to 50% and let it run idle to check for crashes. After about 5 to 7 minutes, the temperature gradually increased to around 47°C, and the PC crashed. Prior to lowering the fan speed, it had been running for roughly 3 hours at full GPU fan speed with a GPU temperature of about 39°C. I’m still considering replacing the motherboard or upgrading to an AM4 platform, but given my current budget limits, it seems unlikely this month.
Edited July 21, 2025 by sa1672ndo
your configuration is ideal for light gaming at 720/1080p, especially given your limited budget. The two components I mentioned in my earlier post match what you have now, and everything functions perfectly. It’s exactly what I need, and the noise level is manageable—especially since this is my spare PC.