On Thu, 26 Mar 2015 11:37:51 +1000, Da Rock wrote:
 > On 03/25/15 09:52, Kevin Oberman wrote:
 > > On Tue, Mar 24, 2015 at 3:23 AM, Da Rock
 > > <freebsd-a...@herveybayaustralia.com.au
 > > <mailto:freebsd-a...@herveybayaustralia.com.au>> wrote:
 > > 
 > >     I have 2 laptops as mentioned, 3 all amd athlon based. The 3rd is
 > >     an asus which I'm relatively happy with.
 > > 
 > >     What I have is when I pull the AC out of it, the sysctl for cpu
 > >     speed goes from 2200 to 100 or 400. Basically the system becomes
 > >     rather unusable.
 > > 
 > >     I tried the acpi_hp module, and it now switches to 800. This is
 > >     better, but barely usable still.
 > > 
 > >     I'd like to see a response similar to the asus if its possible;
 > >     this effectively stays the same, but drops speed if nothing is
 > >     happening.
 > > 
 > >     Ideally, I'd think that it would be better if the system adjusted
 > >     speed to use requirements during operation, but neither does that.
 > >     I suspect that the asus should (in theory) as it does do it on
 > >     battery only; but unless I'm really hammering all the time, it
 > >     just doesn't seem to happen when I'm looking at it.
 > > 
 > >     The settings used on all for powerd is hiadaptive for AC, adaptive
 > >     for battery.
 > > 
 > >     If I'm doing something wrong let me know, if more data is required
 > >     I'm happy to help the cause :)
 > > 
 > >     TIA
 > > 
 > > First, let's get a bit more information. Please provide:
 > > sysctl dev.cpu.0 (on AC and then on battery)
 > AC:
 > 
 > dev.cpu.0.%desc: ACPI CPU
 > dev.cpu.0.%driver: cpu
 > dev.cpu.0.%location: handle=\_PR_.C000
 > dev.cpu.0.%pnpinfo: _HID=none _UID=0
 > dev.cpu.0.%parent: acpi0
 > dev.cpu.0.temperature: 85.3C
 > dev.cpu.0.freq: 200
 > dev.cpu.0.freq_levels: 2200/7543 1925/6600 1900/6150 1662/5381 1600/4777
 > 1400/4179 1300/3750 1137/3281 975/2812 800/2091 700/1829 600/1568 500/1306
 > 400/1045 300/784 200/522 100/261
 > dev.cpu.0.cx_supported: C1/1/0
 > dev.cpu.0.cx_lowest: C1
 > dev.cpu.0.cx_usage: 100.00% last 682us

Kevin appears to be psychic :) as for the second time in 3 days has 
replied while I was composing mine, but this time I heard the beep and 
saw 'new message from Kevin ..' and of course he's well covered it, 
except possibly regarding one aspect which I'll leave in:

The clue is all those freq_levels. For powerd to increase frequeny under 
load requires stepping through all of them, possibly one per polling 
interval, which as you've observed takes too long.  If you were running 
Intel I'd say you need to add to /boot/loader.conf these, and you should 
add them anyway:
hint.p4tcc.0.disabled=1
hint.acpi_throttle.0.disabled=1

But from your one (1) clue that this is an AMD system, I'm not so sure:
 > amdtemp0: <AMD CPU On-Die Thermal Sensors> on hostb4

so you should check dmesg to check if you're running powernow(none!) 
instead of est(4) plus any other drivers providing 'thermal control', 
and if so, disable it (but not powernow or amdtemp) similarly.  It may 
already be using acpi_throttle?, or perhaps p4tcc also covers AMD CPUs
these days, I'm not sure?  If that's unclear, post /var/run/dmesg.boot

Yes, set both *_cx_lowest=Cmax and take delight in dev.cpu.0.cx_usage

Playing with powerd_flags can make quite a lot of difference to powerd's 
behaviour, so I echo Kevin's request to see those, and his advice to run 
powerd -v in a terminal while you play, especially with -i and -r flags, 
though reducing -p interval may be beneficial to responsiveness, at a 
slight increase in powerd's CPU usage.

I don't know if any of this changed between your 10.0 and 10.1 ..

cheers, Ian
_______________________________________________
freebsd-acpi@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-acpi
To unsubscribe, send any mail to "freebsd-acpi-unsubscr...@freebsd.org"

Reply via email to