On Sun, 12 Oct 2025, Van Ly wrote:
At the moment btop reads 46degC and the following script reads 42.000, 43.000
Both envstat and btop read the same data, but, some variation is normal because these on-die thermocouples are a) pretty accurate and b) can instantly sample the current core temp. So you might see diff. values each time--even with envstat(1) alone. Try this cmd. a few times to see this: ``` envstat -ns coretemp0:'cpu0 temperature',coretemp1:'cpu1 temperature' ```
The first thing I asked of the code assistant was for at tool to flowchart the logic in envstat and btop but the friction and hoops to jump thru lead to a timebomb. I think it should be possible to infer what's going with the kernel and the cpu temperature reading using an initial schematic like the citric acid cycle https://sdf.org/~van.ly/img/citric_acid_cycle.jpg
I can't figure out how the Krebs cycle relates to CPU temp. variance :), but, the way NetBSD gets this data is pretty simple: https://github.com/NetBSD/src/blob/trunk/sys/arch/x86/x86/coretemp.c#L359 It just reads some MSRs on x86. The same values are passed to _all_ programs. -RVP
