On 5/19/21 2:03 AM, Douglas McGarrett wrote:
On 5/16/21 9:36 PM, Felix Miata wrote:
Found _sensors_ routine finds and outputs cpu temps. Can someone suggest a simple
closed-form equation that will keep all the cores of the cpu busy? I will try and write some
kind of routine that will output temps vs. time elapsed and see what happens.
--doug
Write something in Pascal??? Why must you always make everything harder? Sensors takes readings from the heat/voltage sensors that are on the mobo and/or on the cpu. There needs to be a configuration clause in its config file specific to the sensor devices. The sensors app can set that up for you, or at least the bare structure which you can edit, add labels to, etc. Some boards do not have such devices, on others the sensors app may not be able to find the chips. #man sensors #man sensors-detect #man sensors-conf Do not confuse monitoring readings with stress testing. Stress testing is a technique used in overclocking and to burn in new hardware, etc. My preference is mprime (prime95 on W$), but there are others . . . https://wiki.archlinux.org/title/Stress_testing https://linuxhint.com/useful_linux_stress_test_benchmark_cpu_perf/ https://www.mersenne.org/download/ https://www.unixmen.com/how-to-test-system-stability-using-mprime/ I presume you are doing this because you've had several crashes and you think the problem may be hardware, e.g., overheating. Monitoring the temps can easily be done in an app like KsysGuard (which also allows the user to graph the sensor devices). 'Keeping all cores busy' essentially means nothing; busy at what, how busy, what thresholds, etc.? Do you even understand power throttling? It's one thing to monitor the system, another to use stress-test tools, yet another to know how to interpret results. So another safe assumption is that you don't really know what you're doing (watching the fan speed? calculate time elapsed? you gotta be kidding). You are venturing into dangerous waters, pal. --dg