Puppy Linux Discussion Forum

[jemimah]

Jakfish, have you turned on cpu frequency scaling?

Can you run powertop on both 4.3 and 4.1 and tell me what the wattage difference is? Or post schreenshots? The fsb underclocking hack solved the inconsistencies I was seeing on my machine.

Battery life is why I waited to buy a netbook until the 1005ha came out, the advertised 10 hour battery life translates into about 6 actual hours. Not too bad.

[jakfish]

01micko--

I'd be very interested to hear your results. I also, belatedly, see that BK relented and released a 4.3 with the older kernel.

I just don't know what's making it burn hotter--the kernel or the OS, or maybe my machine itself.

But at any rate, I'd be grateful for any insight from users more learned than me.

Jake

[jakfish]

Hi, jemimah,

I have tried frequency scaling which will not work w/ eee/4.3 (there are other threads/confirmations that show the machine stuck on "performance" despite the elected setting.).

Jake

[magerlab]

http://murga-linux.com/puppy/viewtopic.php?t=47267
needed modules for eee keys and fan-control

[jemimah]

Is there a link to source code for the eee module somewhere?

-----------


No wonder you're so running so hot, Jakfish! I can cook eggs on mine in performance mode. Can you link me to the relevant discussion?

Have you tried setting it manually?

[aragon]

See here for a workaround by ARAN: http://www.murga-linux.com/puppy/viewtopic.php?p=344082#344082

maybe a better but larger workaround would be to use the larger iso and to merge the content of the zp*.sfs to the main pup-file.

aragon

[jakfish]

Hi, jemimah,

I'm using EEEControl, so am able to drop down to power-saving speed of 630mAH. So I don't believe I'm stuck on "Performance."

Anecdotally, I can hear the CPU speed drop when I switch to lower rate--it's a whine that audibly decreases. Can you hear such a sound when you're scaling?

Jake

[jemimah]

I've heard of the cpu whining thing but either my netbook is quiet or my hearing is not that good.

On my netbook, eee-control does not modify the cpu frequency, strange... Are you seeing P-states and C-states in powertop? I wonder if acpi is working for you at all....

I've attached a screenshot of what powertop output looks like on my machine. How many C-states do you see when you're unplugged?

[jakfish]

jemimah--

I'm no expert, but when using eeecontrol, I think the command to bring up the bona fide speed is this:

# cat /proc/eee/fsb
70 24 0

At least, this is what I use in my conky. Couple that with the increase/decrease whine of cpu speed makes me think that I'm getting genuine change.

As for powertop, I'm not getting your groovy readout:

Powertop

Cn Avg residency P-states (frequencies)
C0 (cpu running) (14.2%)
polling 0.0ms ( 0.0%)
C1 1.7ms ( 1.1%)
C2 2.0ms (15.7%)
C3 3.3ms (69.0%)

Wakeups-from-idle per second : 293.8 interval: 10.0s
no ACPI power usage estimate available

No detailed statistics available; please enable the CONFIG_TIMER_STATS kernel op
This option is located in the Kernel Debugging section of menuconfig
(which is CONFIG_DEBUG_KERNEL=y in the config file)
Note: this is only available in 2.6.21 and later kernels

[jemimah]

what happens if you try

modprobe p4-clockmod

Is that module there? Supposedly it adds speedstep for celeron M.


----

Humm, further research suggests p4-clockmod is probably no good. I wonder what could possibly be different between the old and new kernels....

[jakfish]

Isn't p4-clockmod found by the cpu scaling gui? I remember seeing that, I think (so much testing, so little progress). As you probably saw in the other threads, in 4.3, scaling/eee appears to produce only powersaving and performance setting, so there's no real scaling. Or at least I didn't get any.

I did get scaling to function in earlier puppies, 4.0/1, I think. And found only a lag in response with no accompanying battery/temperature savings.

Over in the forum.eeeuser.com, there was a general consensus that at least the 700/900 models didn't have a processor that played nice with scaling. It throttles rather than truly drops the CPU speed.

dawnsboy is a real expert at this. I'm just the guy who can't get it to work

[jemimah]

You have a working script for powersaving for the old kernel? Have you tried the old script with the new kernel?

I was thinking of buying one of the Celeron Eees just to figure this stuff out. I imagine you can get them cheap by now. I can probably pawn it off on my husband once I'm done with it.

[jakfish]

The script is the same. As for your lucky husband, he can get a 900 on amazon very cheap. Last I looked, they were $229. I paid $550 16 months ago. Even cheaper on ebay, but buyer beware and all that.

[jakfish]

Here's one for the books, and it connects to the cpu speed problems within the eee.

Background: I had been complaining that 4.3/new kernel was causing temperature spikes. So I went to 4.3/old kernel, but changed one other thing: my startup speed in BIOS. I had been starting at 900mz/Performance to shorten boot time, then w/ eeecontrol, dumbing down to 630 once puppy was up and running.

I decided to change to 630/power saving in BIOS and just keep it there. Lo and behold, my temperature's gone down 5 degrees Celsius. Before, (with 900mh BIOS speed/change in puppy to 600mh), after an hour or so, (no wifi, no fan, medium brightness, cpu between 10-12%), I was idling at about 56C. Same specs now, but I'm at 51C. That's a big drop.

Whiile a 900mh startup/eeecontrol-switch-to-630 may make fsb show the speed at 630, I'm almost convinced that the machine is throttling rather than reducing speed. Even with the cpu whining down, etc.

How else can I account for 5 degree drop? Everything else is the same except for BIOS speed change.

At any rate, I may be able to return to newer kernel and see what's what.

Jake

[magerlab]

i haven't yet tried 4.3 on eee pc( maybe on weekend or would wait for a new NOP , cause i like xfce4 behavior on eee)

but i suggest to use fan control module as i do in 4.1.2
i have fan blowing 60% from the start and tempetrature never gets higher than 50( in fact 47-4