You are not logged in.
In spite of my scepsis regarding software scaling of the CPU¹ I decided to hook my laptop up with a watt meter and load the acpi_cpufreq module. It seems to be quite popular.
Although it instantly throttled the clock speed down from maximum (2.5-something MHz) to minimum (1.1-something MHz) it had no effect on the watt usage. I was on 14.9 watts both before and after².
Am I doing something wrong or is there some reason behind my scepsis?
The laptop is a Thinkpad X201.
1) I find it hard to believe that manufactures of mobile processors don't already do something to cut the power usage when the CPU is idle. Will frequency scaling really top this?
2) In comparison enabling power management on wlan0 gave me somewhere between 1 and 1.5 watts. (I was impressed, really.)
Last edited by Lars Stokholm (2011-07-07 14:31:49)
Offline
It's not just laptops, desktops also don't read a difference in my experience. My X3360 hops from 3.40 GHz down to 2.40 GHz with no measurable reduction in power per my Kill-A-Watt meter.
CPU-optimized Linux-ck packages @ Repo-ck • AUR packages • Zsh and other configs
Offline
What reduction occurs when you shut off HDD?
Prediction...This year will be a very odd year!
Hard work does not kill people but why risk it: Charlie Mccarthy
A man is not complete until he is married..then..he is finished.
When ALL is lost, what can be found? Even bytes get lonely for a little bit! X-ray confirms Iam spineless!
Offline
Maybe removing the battery will give different results.
Offline
What reduction occurs when you shut off HDD?
I don't know, but I can find out later today. Why?
Maybe removing the battery will give different results.
No, I've already tried with and without a battery. But thanks anyway.
Offline
That should reduce the load by several watts..
Prediction...This year will be a very odd year!
Hard work does not kill people but why risk it: Charlie Mccarthy
A man is not complete until he is married..then..he is finished.
When ALL is lost, what can be found? Even bytes get lonely for a little bit! X-ray confirms Iam spineless!
Offline
That should reduce the load by several watts..
Oh. Yeah, it probably would, but I can't get that to work: https://bbs.archlinux.org/viewtopic.php?id=113516&p=2.
Offline
Did you load only acpi_cpufreq or did you also load some governors like cpufreq_{ondemand,conservative,powersave}? Did you set the governor with cpufreq-set?
R00KIE
Tm90aGluZyB0byBzZWUgaGVyZSwgbW92ZSBhbG9uZy4K
Offline
Well, how about your CPU temperatures. From my experience the temperatures and fan speed are lower with ondemand governor, also battery time is much longer. Of course acpi claims that power usage hasn't went down. How do you measure your watt usage? Without simple external wattmeter or other capable meter it's impossible to obtain true value.
It's not just laptops, desktops also don't read a difference in my experience. My X3360 hops from 3.40 GHz down to 2.40 GHz with no measurable reduction in power per my Kill-A-Watt meter.
Did you overclock your processor? I thought that X3360 has 2,83GHz on stock. Reduction in frequency won't give to much (especially on desktop where the whole power usage is rather big and efficiency of PSU on lower loads is poor) but with it goes voltage reduction which should give good results (with overclocked processor the voltage reduction usually doesn't happen).
Last edited by einhard (2011-07-07 10:42:09)
Offline
Did you load only acpi_cpufreq or did you also load some governors like cpufreq_{ondemand,conservative,powersave}? Did you set the governor with cpufreq-set?
As I said, I loaded the acpi_cpufreq module, yes. Aside from the performance governor, I've tried ondemand. Same thing.
Well, how about your CPU temperatures. From my experience the temperatures and fan speed are lower with ondemand governor, also battery time is much longer. Of course acpi claims that power usage hasn't went down. How do you measure your watt usage? Without simple external wattmeter or other capable meter it's impossible to obtain true value.
CPU temperature unchanged, which doesn't surprise me. The watts have to go somewhere. I use an external watt meter, yes.
Offline
You should be able to notice a difference between the performance and powersave governors, however I guess modern cpus rely not only on frequency and voltage scaling but also on clock gating to help reduce power usage so even at maximum frequency and voltage there is some power saving going on.
Frequency scaling helps to reduce power because power usage (in the active units) is directly proportional to the frequency, how much it can reduce when compared with turning off inactive units depends on the number of active units, power saving is a team game every little bit helps.
R00KIE
Tm90aGluZyB0byBzZWUgaGVyZSwgbW92ZSBhbG9uZy4K
Offline
R00KIE is right
I didn't notice but you have Core i5 540M 2,53 processor in your laptop. Am I right? It has Turbo feature which purpose is not only overclocking one core (in your case) on increased one thread load but also undervolting and turning off unused cores. Turbo is hardware solution (microchip), so yes, with enabled Turbo (it can be disabled in Bios/Uefi but in many notebooks this option is hidden) you probably won't notice any difference with software powersaving on.
Last edited by einhard (2011-07-07 14:15:30)
Offline
Cool. Thanks a lot guys, then I can put my mind to rest. I was wondering why laptop-mode didn't do much for me, so I decided to disable it and try each of its tricks individually. Now I can cross one more item off my list.
Marking as solved.
Offline