Closed metalshark closed 10 months ago
Hm. Can you attach the log file? I'm not sure I'm going to look into this, as per the link you posted the setting is only meant to be set for debugging reasons, so it may have undesired consequences (such as this). I'd still be interested in what the log says though in this case.
I wonder, why did you disable the dynamic ticks? Some pages suggest that it fixes stuttering or frame drops, while other pages state the opposite.
Most systems leaving it on is perfectly fine. For high sample rate audio though the ASIO drivers had issues with small buffer sizes and lower latencies which disabling dynamic ticks resolved. Here's an example: CoreCycler_2023-08-30_14-48-34_YCRUNCHER_22-ZN4 ~ KIZUNA.log Y-Cruncher running manually and/or setting processor affinity manually is fine. Also I can always enable dynamic ticks for running CoreCycler then disable thereafter so it isn't the end of the world if this gets a WONTFIX.
It can't read the CPU usage with the Windows Performance Counters resp. it returns 0%.
I've seen that happen on rare occasions before, so it might be connected to Dynamic Links (which the other users had forgotten about in their case).
If you don't want to change the setting just for CoreCycler, you could try to download the latest alpha version, which can parse the output of y-Cruncher to check for errors instead of solely relying on the CPU usage. And additionally you could try to disable the CPU usage check altogether with disableCpuUtilizationCheck
in the [Debug]
section of the config.
Or maybe setting restartTestProgramForEachCore
to 1 would already be enough, although in your log even after y-Cruncher was restarted by CoreCycler the issue remained.
Perfect, thank you for offering an alternative method to solve the issue. If it can't read CPU usage then case closed (how're you meant to do a CPU utilisation check when it gives you 0% every time).
If disabledynamictick is set to yes bcdedit Additional Settings then the first core always completes, the other cores will always fail. Not an overclock issue (happens on stock speeds). It completes the work for the core then reports the core failed.
Disabled the setting in bcdedit (disabledynamictick = no) to test in isolation.