What can cause the system to freeze in a way where even the reset button takes a long time to react?

What can be the reason for system freezes that are so "hard" that even the hardware reset button takes about 3 seconds until it actually resets the system (and then it actually powers down and up again instead of doing a "clean" hard reset like when pressing it during a normally running system).

Since it initially happened mainly while playing videos from YouTube I suspected the graphics card - however, I replaced it recently and it did not change it. It still happens from time to time (and sometimes more often, like a few times times in the last few hours).

The system is running Windows 7 - but I don't think this matters since I don't think any software, not even the OS, can actually affect the reset button's behaviour.

The PC is not overheated and the freezes happen randomly. There is also no malware on the system. The CPU is an Intel Core i7-920 on a Gigabyte EX58-UD5 mainboard.

What could be the cause for this problem? Faulty RAM? I did not run a full memtest86 check yet, but I wonder if there is a more likely issue than faulty RAM - checking 12G of ram does take some time after all!

There are no entries in the event log - but that's what I expected since the system freezes so hard that I doubt it has time to write anything to any log.

Solution 1:

(generic answer)

I'm not sure, but the RESET button does not reload the BIOS bootstrap code into RAM, it just jumps to it. So if it has been changed by a defective driver, the RESET button dies. It was with that suspicion that I actually started a few searches, the last one being "pc freeze reset button not working"

A defective driver or a custom kernel might be able to replace that since it doesn't overwrite actual BIOS code. The power-down (4 seconds power button) makes the next boot load the initial program from the CMOS, thus "reloading the BIOS". That's why THAT works.

I have seen in a text-mode Linux kernel that whenever I pressed the Reset button there was an entire frame of new text (dunno, 10, 15 new lines triggered by the reset?) before actually resetting. One may note that RESET might not go to the actual BIOS location triggered by the normal boot...

You may look up the specifications for the Intel CPU... On Wikipedia the IBM PC does not clear RAM nor does it do the IPL, and that might have been inherited by our usual PCs, but I can't be sure until I actually experiment with it (maybe some code that indeed hooks the reset button?)

Any problems with this button thus might be software-related, though I can't know for sure (If the OS detects something like that from a driver it most likely BSODs, maybe since the respective pages might not be directly available and the driver is "caught" modifying the page table) But Linux has likely "been there, done that"...

Random stuff: The reset button on my PC is something like 4-5 inches away from the power button, and smaller and somewhat more difficult to press, though not impossible.

Solution 2:

A CPU (and any integrated circuit with a reset pin) should always react to its reset signal.

Possible causes:

• Reset signal is not reaching CPU. Not sure how the reset button pins are routed on your board but it may be that it actually goes to the chipset, and perhaps the chipset is buggy or faulty.

• CPU is actually resetting but not getting very far in its BIOS due to an error very early in its initialization. A faulty CPU that fails its initial BIST (built-in self test), a failing BIOS flash chip, or a fault in the motherboard hardware (usually chipset) that routes interrupt signals.

• Some thermal protection either on the CPU or chipset is misbehaving. Perhaps your fans are not correctly or completely connected or there is an issue there.

• Power supply fault.

Very strange. All I can think of, really.