It seems that powerd does very little in terms of reducing heat, and sacrifices performance while doing so. Am i wrong to assume that lowering the cpu's frequency should reduce consumed power, and therefore reduce the amount of heat produced? I have tested with mbmon and i see no difference between an idle system running with powerd at 75mhz, and at full rate without. Also, while testing the speed of a php script, i found myself refreshing it quickly in the browser to see the results of the timers. I was getting sporadic results, on an otherwise idle system. I noticed that if anything cpu intensive ran in the background, my script would execute quicker. I disabled powerd and restored the frequency, at this point i got more consistent results, and the script would execute over 100msec faster. It seems like its not adjusting the clock fast enough. Are these problems with powerd or just my hardware? It is an old athlon system, running on the via133 chipset . --- CPU: AMD Athlon(tm) Processor (1210.79-MHz 686-class CPU) Origin = "AuthenticAMD" Id = 0x642 Stepping = 2 Features=0x183f9ff<FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,MMX,FXSR> AMD Features=0xc0440800<SYSCALL,<b18>,MMX+,3DNow+,3DNow> real memory = 536788992 (511 MB) avail memory = 520237056 (496 MB) npx0: [FAST] npx0: <math processor> on motherboard npx0: INT 16 interface acpi0: <ASUS A7V> on motherboard acpi0: Power Button (fixed) Timecounter "ACPI-fast" frequency 3579545 Hz quality 1000 acpi_timer0: <24-bit timer at 3.579545MHz> port 0xe408-0xe40b on acpi0 cpu0: <ACPI CPU> on acpi0 acpi_throttle0: <ACPI CPU Throttling> on cpu0 acpi_button0: <Power Button> on acpi0 pcib0: <ACPI Host-PCI bridge> port 0xcf8-0xcff on acpi0 pci0: <ACPI PCI bus> on pcib0 pcib1: <PCI-PCI bridge> at device 1.0 on pci0 pci1: <PCI bus> on pcib1 isab0: <PCI-ISA bridge> at device 4.0 on pci0 isa0: <ISA bus> on isab0 atapci0: <VIA 82C686A UDMA66 controller> port 0x1f0-0x1f7,0x3f6,0x170-0x177,0x376,0xd800-0xd80f at device 4.1 on pc
On Thu, 12 Jan 2006 10:57, Mike Jakubik wrote:> It seems that powerd does very little in terms of reducing heat, and > sacrifices performance while doing so. Am i wrong to assume that > lowering the cpu's frequency should reduce consumed power, and therefore > reduce the amount of heat produced? I have tested with mbmon and i see > no difference between an idle system running with powerd at 75mhz, and > at full rate without. Also, while testing the speed of a php script, iIf you want to test how much heat your system draws for a given clock speed you should dispense with powerd and just set the frequency by hand, ie.. sysctl dev.cpu.0.freq=XXX powerd won't run your CPU at a specific clock frequency - it varies the CPU frequency based on current load conditions. Do you have thermally controlled fans? If so I wouldn't expect the temperature to vary with clock speed very much at all.> found myself refreshing it quickly in the browser to see the results of > the timers. I was getting sporadic results, on an otherwise idle system. > I noticed that if anything cpu intensive ran in the background, my > script would execute quicker. I disabled powerd and restored the > frequency, at this point i got more consistent results, and the script > would execute over 100msec faster. It seems like its not adjusting theThe powerd defaults do not change frequency that quickly - every 500ms by default. I run it with '-p 200' and it seems fine although you do notice it 'stick' sometimes (where the CPU change doesn't happen quickly enough). You could try what I do but there are some systems which are very slow to change clock speed so this could be an impediment.> clock fast enough. Are these problems with powerd or just my hardware? > It is an old athlon system, running on the via133 chipset .I'm suprised a system this old even supports a clock speed as low as 75Mhz. -- Daniel O'Connor software and network engineer for Genesis Software - http://www.gsoft.com.au "The nice thing about standards is that there are so many of them to choose from." -- Andrew Tanenbaum GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 187 bytes Desc: not available Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20060112/63861f7b/attachment-0001.bin
Hi all!
This issue was written over and over,
and a lot of people, including me, have
more questions. Don't want to crosspost,
cause I think that it is related.
On Celeron M laptop, with cpufreq and
powerd, also with this in rc.conf:
performance_cx_lowest="LOW"
economy_cx_lowest="LOW"
I have C2 and 1400 -> 174 MHz. Should it
be set "LOW" for performance_cpu_freq
and economy_cpu_freq? Battery lasts
for almost 3 hours with no X and exten-
sive cd/modem/hdd work. Temperature is
34-38.
On my amd64 box (754, 1800 MHz), with
Coll'n'Quiet, acpi2 and the same confi-
guration, I have just one C (C1) and
two states (1800 and 800). Usb is com-
piled into kernel. (One poster removed
it and got more steps.) Anyway, the box
is really cool, so it is just a matter
of curiosity. Is there anything, beside
using usb as module, to get more c's
and put frequency down dinamicaly?
And no temp measurement on this nforce3
mobo.
Best regards
Zoran
On Wed, Jan 11, 2006 at 07:27:13PM -0500, Mike Jakubik wrote:> It seems that powerd does very little in terms of reducing heat, and > sacrifices performance while doing so. Am i wrong to assume that > lowering the cpu's frequency should reduce consumed power, and therefore > reduce the amount of heat produced? I have tested with mbmon and i see > no difference between an idle system running with powerd at 75mhz, and > at full rate without. Also, while testing the speed of a php script, i > found myself refreshing it quickly in the browser to see the results of > the timers. I was getting sporadic results, on an otherwise idle system. > I noticed that if anything cpu intensive ran in the background, my > script would execute quicker. I disabled powerd and restored the > frequency, at this point i got more consistent results, and the script > would execute over 100msec faster. It seems like its not adjusting the > clock fast enough. Are these problems with powerd or just my hardware? > It is an old athlon system, running on the via133 chipset .I know that this is not what you're asking for, but anyway: You should try sysutils/fvcool, it will reduce de heat a lot and the performance will not suffer. -- La prueba mas fehaciente de que existe vida inteligente en otros planetas, es que no han intentado contactar con nosotros.
Mike Jakubik wrote:> It seems that powerd does very little in terms of reducing heat, and > sacrifices performance while doing so. Am i wrong to assume that> CPU: AMD Athlon(tm) Processor (1210.79-MHz 686-class CPU)It is very unlikely this processor supports any kind of frequency modification by software. Basically there are two cases with modern processors: - Very modern processors, from Pentium M class onwards, support true frequency modification, which can and does offer significant savings. - Somewhat older processors, and the whole Celeron M line support only "CPU throttling", which is something like forcing idle cycles (like the "HLT" instruction) only on hardware level. I think this can be distinguished in FreeBSD by the second number in freq_levels being -1, like here: dev.cpu.0.freq_levels: 1403/-1 1315/-1 1227/-1 1139/-1 1052/-1 964/-1 876/-1 789/-1 701/-1 613/-1 526/-1 438/-1 350/-1 263/-1 175/-1 87/-1 (this second type of power management management support doesn't alter the physical frequency). I think the infrastructure used by powerd supports both cases, but won't get you much savings if the CPU doesn't support the first case. This information was gathered because I have a Celeron-based laptop and wanted to squeeze as much autonomy as possible - it may not be authoritative :) What I would like for FreeSBD to support is turning off of devices like WinXP does. Not only hard drives, but it seems that WinXP can somehow turn off network cards, USB controllers and/or devices and similar peripherals when running on batteries and those are not used (it seems it's not like disabling them completely but something else).