Does the AMD Core WDT fail to activate if the operating system becomes unresponsive?
Does the AMD Core WDT fail to activate if the operating system becomes unresponsive?
I need my PC to restart automatically if Windows freezes. In BIOS I've turned Core Watchdog on, set the interval to about 10 seconds and severity to fatal. However when I force a crash using a Rust program with many threads and realtime priority, nothing happens. I think Core Watchdog might differ from System Watchdog and only activates when the CPU isn't responding, which doesn't seem to be the case here? Could you confirm if gaming motherboards have system watchdogs?
Core Watchdog functions as intended. Similar mechanisms exist in GNU/Linux distributions. A practical illustration: when an application becomes unresponsive on Windows, a watchdog timer activates after a set period, displaying a warning that the program isn't responding or is taking too long. It then presents two choices—terminate the process or remain and wait for it to resolve itself.
Watchdogs operate even at the kernel level. For instance, encountering a BSoD typically reveals error details, after which the operating system initiates a countdown (around 15 seconds) before automatically restarting the entire system. Alternatively, it may prompt the user to manually reboot without an automatic timer.
While I lack deep technical knowledge of software-specific watchdogs, it is clear that each program should have its own dedicated watchdog and timing mechanism, tailored to its needs. Some applications require more time to respond than others.
When the operating system freezes, it's rarely the OS itself that becomes unresponsive, but rather a particular application. If that app holds the highest priority, it can interrupt other processes, rendering the remaining system unusable. This creates the illusion of a full system freeze.
There are kernel-level commands available to adjust program priorities and resolve such issues. A common solution is pressing Ctrl + Alt + Del, which opens Task Manager to terminate frozen programs or restart the OS.
Regarding hardware-based watchdogs on consumer devices, they aren't standard in regular consumer hardware. They are typically found on server systems where a single task is executed, allowing the watchdog to trigger a hardware restart when needed.
On consumer machines, these watchdog chips can cause significant issues because PCs are used for diverse tasks, not just one specific function. Setting an inappropriate timeout makes it difficult to determine when the chip should intervene.
For example, if a watchdog is set to time out after 30 seconds and triggers a reboot upon detecting inactivity, it might restart the system even if the application isn't actually frozen. Conversely, if the program appears to be stuck but remains active, the watchdog could mistakenly initiate a reboot.
Determining whether the freeze was caused by the OS or a specific program requires careful analysis.