From: James Cook <falsifian@falsifian.org>
Subject: Re: [new/wip] sysutils/apmtop
To: Benjamin Stürz <benni+openbsd@stuerz.xyz>
Cc: ports@openbsd.org
Date: Wed, 10 Jan 2024 21:29:20 +0000

>>Thanks, I tried it out on my laptop.
>Thanks for trying.
>>
>>The BAT and PWR displays always show 0% and 0mW respectively as far 
>>as I can tell. Note "apm -l" can show my batter percentage so I 
>>guess that at least should be possible to get right. dmesg below in 
>>case it's useful.
>Now it's fixed.
>I forgot to unveil("/dev/apm", "r").
>See attached port.

Now BAT works (matches apm -l). Power still shows 0mW 
(min: 0/max: 0) but I don't even know if my laptop can report power 
consumption.

>>CPU percentage is lower than what "top -1" shows. E.g. with
>>"++$x while 1" running in a perl repl, top -1 shows 19% user but 
>>your tool shows about 10%. I don't know which is more accurate, so 
>>this isn't necessarily a complaint.
>Try with a different delay, maybe the same one as top uses.
>Use `apmtop -d DELAY` with DELAY being measured in tenths of seconds.
>e.g. `apmtop -d 10` refreshes every second.

Comparing "apmtop -d 10" with "top -1s1" it's still a bit different, 
but I don't have any evidence top's estimate is better.

I'm running "find /" in an xterm to use the CPU a bit.

Two side-by-side "top -1s1" instances agree within a couple percentage 
points. "apmtop -d 10" shows a different number, e.g. 21% when top 
shows 32%. Usually lower but sometimes higher.

>>CPU frequency looks plausible. Temperature is at least changing 
>>(went from 32 C to 33 C); not sure if it's accurate.
>I'll look into that.

I'm not saying 32C sounds wrong, just that I don't have 
a way to verify it :)

-- 
James