Overclock the N900? - Page 329 - maemo.org

Active Topics

Maemo repository (8)
to Community by Maemish - 21 hrs, 32 mins ago
Porting apps to Leste (44)
to Maemo 7 / Leste by norayr - 1 day, 2 hrs ago
List of well performing apps on Leste (35)
to Maemo 7 / Leste by smatkovi - 3 days, 23 hrs ago
Sailfish OS for the Motorola Moto G7 Power (XT1955-5) - (ocean) (13)
to SailfishOS by edp17 - 4 days, 21 hrs ago
Sailfish OS for the Samsung Galaxy Note 4 (SM-N910C) - (treltexx) (79)
to SailfishOS by edp17 - 4 days, 21 hrs ago
U-Boot for Nokia RX-51 with BootMenu (updated version 2012.10-rc3-1) (843)
to Alternatives by Arno_11 - 5 days, 2 hrs ago
more...

Page 329 of 536

329

Thread Tools

	Gusse	2010-04-23 , 19:19
	Posts: 168 \| Thanked: 206 times \| Joined on Apr 2010 @ Finland	#3281

Notes and warnings added to my original post. And there is no need to do any modifications to sudoers to run those scripts.

Thanks You 白い熊 once again!

Quote & Reply |

The Following User Says Thank You to Gusse For This Useful Post:
casper27

	Gusse	2010-04-23 , 19:33
	Posts: 168 \| Thanked: 206 times \| Joined on Apr 2010 @ Finland	#3282

Originally Posted by egoshin

CPU in idle has (at least in theory) clocks switched OFF. However, it has VDD1/2 and it has power leakage too. If we set (somehow) the idle VDD1/2 to a minimum for any frequency we may not need lowfreq anymore (because it looks from my measurement that CPU in high freq may consume energy less per CPU cycle or same. However, SmartReflex should be switched on for that).

OK, when system is in sleep or deep sleep modes, then clock is not probably running. But from time_in_state -file you can check that when system is running what frequencies you have used and how long times. In my case, idle (i.e. nighttime) CPU is at 125MHz 95% of the time (when system is awake), so from that perspective lowers CPU voltage is also best for battery life. Or is it?

850mV @ 125MHz (VDD1 = 20)
vs
975mV @ 500MHz (VDD1 = 30)
vs
912mV @ 250MHz (VDD1 = 25)

Last edited by Gusse; 2010-04-23 at 19:53.

Quote & Reply |

suphiceto	2010-04-23 , 19:34
Posts: 9 \| Thanked: 0 times \| Joined on Apr 2010	#3283

Originally Posted by AlMehdi

Maybe you missed to install the Enhanced kernel?

Try in the terminal as root:
Code:
apt-get install kernel-power-flasher kernel-power-settings

Thanks. I think, it is working now. But, device reboots after couple minutes automatically. I am assuming this is a sign of instability? Is there any other options that I could try? Or shall I go back to Lethos 125-900 kernel?

Quote & Reply |

egoshin	2010-04-23 , 19:53
Posts: 992 \| Thanked: 995 times \| Joined on Dec 2009 @ California	#3284

Originally Posted by Gusse

OK, when system is in sleep or deep sleep modes, then clock is not probably running. But from time_in_state -file you can check that when system is running what frequencies you have used and how long times. In my case, idle (i.e. nighttime) CPU is at 125MHz 95% of the time, so from that perspective lowers CPU voltage is also best for battery life. Or is it?

I can't say - if energy consumption in your N900 at 500MHz is less than 4 times of energy consumption at 125MHz then it may be advantage to stick to 500MHz only. 4times - because at 500MHz CPU cycle is 4 times shorter and any energy dissipation less than 4 times of 125MHz can be a battery saver.

Why in your case - I don't know your energy consumption. You can load bq27x00_battery.ko driver ("modprobe bq27x00_battery.ko"), open a second X-Term window, run a simple set of commands:

while true
do
true
done

and look into values back in first X_Term window by

cat /sys/class/power_supply/bq27200-0/current_now
cat /sys/class/power_supply/bq27200-0/voltage_now
cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
cat /sys/power/vdd1_opp
cat /sys/power/vdd2_opp

EDIT: switch phone offline for accurate measurements.
EDIT2: try to put this commands into shell script file and run it with 12sec delay while lock a phone in that 12secs:

sleep 12; sh ./script-file

Try it with 500MHz and limit your N900 by 125MHz only (into scaling_max_freq) and see the values. It would be very interesting (that voltage do you use for each frequency too?)

Last edited by egoshin; 2010-04-23 at 20:27.

Quote & Reply |

The Following 2 Users Say Thank You to egoshin For This Useful Post:
Gusse, S0urcerr0r

egoshin	2010-04-23 , 20:19
Posts: 992 \| Thanked: 995 times \| Joined on Dec 2009 @ California	#3285

Originally Posted by Gusse

850mV @ 125MHz (VDD1 = 20)
vs
975mV @ 500MHz (VDD1 = 30)
vs
912mV @ 250MHz (VDD1 = 25)

The most interesting is the current values from /sys/class/power_supply/bq27200-0/current_now

Quote & Reply |

The Following User Says Thank You to egoshin For This Useful Post:
S0urcerr0r

	Gusse	2010-04-23 , 20:41
	Posts: 168 \| Thanked: 206 times \| Joined on Apr 2010 @ Finland	#3286

Originally Posted by egoshin

I can't say - if energy consumption in your N900 at 500MHz is less than 4 times of energy consumption at 125MHz then it may be advantage to stick to 500MHz only. 4times - because at 500MHz CPU cycle is 4 times shorter and any energy dissipation less than 4 times of 125MHz can be a battery saver.

Why in your case - I don't know your energy consumption. You can load bq27x00_battery.ko driver ("modprobe bq27x00_battery.ko"), open a second X-Term window, run a simple set of commands:

while true
do
true
done

and look into values back in first X_Term window by

cat /sys/class/power_supply/bq27200-0/current_now
cat /sys/class/power_supply/bq27200-0/voltage_now
cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
cat /sys/power/vdd1_opp
cat /sys/power/vdd2_opp

Try it with 500MHz and limit your N900 by 125MHz only (into scaling_max_freq) and see the values. It would be very interesting (that voltage do you use for each frequency too?)

Here are results in same order as you listed. I did several runs, which all showed very similar values. Here are 2 results per frequency.

- Current (Vbat)
- Voltage (Vbat)
- Frequency
- what vdd1_opp & vdd2_opp tell?

EDIT: Updated results here

Results in Offline-mode with 12s delay.

Code:

@ 125 MHz       |    @ 250MHz
Run #1  Run#2   |    Run #1  Run#2
737     737     |    824     824
4113    4113    |    4094    4094
124800  124800  |    249600  249600
2       2       |    3       3
3       3       |    3       3
----------------------------------
@ 500MHz        |    @ 850 MHz
Run #1  Run#2   |    Run #1  Run#2
1051    1051    |    1623    1623
4105    4105    |    4089    4086
500000  500000  |    849920  849920
4       4       |    10      10
3       3       |    3       3

EDIT: Also 850MHz results added
EDIT2: All result in Offline-mode with 12s delay.
EDIT3: 250MHz added

Last edited by Gusse; 2010-04-24 at 09:46.

Quote & Reply |

The Following 2 Users Say Thank You to Gusse For This Useful Post:
egoshin, titan

AlMehdi	2010-04-23 , 22:45
Posts: 1,751 \| Thanked: 844 times \| Joined on Feb 2010 @ Sweden	#3287

Originally Posted by suphiceto

Thanks. I think, it is working now. But, device reboots after couple minutes automatically. I am assuming this is a sign of instability? Is there any other options that I could try? Or shall I go back to Lethos 125-900 kernel?

No.. try another setting. If the ideal not working maybe the lv or xlv is better for you. When you have found a stable setting you could start tweak the VSEL to match your device better.

It probably restarts cause it doesn't get enough juice..

Quote & Reply |

The Following User Says Thank You to AlMehdi For This Useful Post:
suphiceto

suphiceto	2010-04-23 , 23:22
Posts: 9 \| Thanked: 0 times \| Joined on Apr 2010	#3288

Thank you so much for your help, I have been trying to flash to lv,ulv or xlv settings. But for some reason it doesnt flash to those settings or I am not noticing any differance. I am not not sure if I am doing something wrong but, I enter
/usr/sbin/kernel-load /usr/share/kernel-power-settings/lv
Then I check the Conky and I dont see any changes? it doesnt seem to be any different than stock settings.

Last edited by suphiceto; 2010-04-23 at 23:38.

Quote & Reply |

egoshin	2010-04-23 , 23:26
Posts: 992 \| Thanked: 995 times \| Joined on Dec 2009 @ California	#3289

Originally Posted by Gusse

Here are results in same order as you listed. I did several runs, which all showed very similar values. Here are 2 results per frequency.

I looked into your measurements.

1. Your power consumption is more than my with lesser VDD1/2 besides 500MHz. Your battery provides more voltage (my stays about 3860), so - I guess, the VDD values (20, 25 and 30) are not very accurate and that may be a reason why somebody has a good OC but somebody doesn't - all VDD values are too inaccurate. Note: vdd1_opp & vdd2_opp are the number of element from array of voltages (like XLV is: "30 20 30 33 38 38 45 45 48 48 54 54 60 67 72").

2. Do you sure you did measure with locked N900 (screen!) for low frequencies - your N900 spends significantly more energy than my N900 even with your lower VDD1/2? (824mA * 4094mV versus my 400mA * 3860mV). Or - did you change some 'ondemand' parameters like sampling_rate?

3. The frequency and power consumption ratios clearly points that you have big advantage in 500MHz. After that (850MHz) the frequency increase by 1.7 is accompanied by energy increase by 1.6, so the energy per clock is pretty close (no advantage). But frequency increase from 125MHz to 500MHz (by 4) collects only 1.4 more energy and each CPU clock spends almost 2.8 times LESS energy on 500MHz (1.4 * 2.8 == 4)

So, if you don't run poorly written endless applications you definitely save more energy on frequencies 500MHz and more. Happy you!

Quote & Reply |

The Following 2 Users Say Thank You to egoshin For This Useful Post:
Gusse, S0urcerr0r

egoshin	2010-04-23 , 23:50
Posts: 992 \| Thanked: 995 times \| Joined on Dec 2009 @ California	#3290

I looked briefly in kernel code and it looks like:

if you have vdd2_opp > 2 it means you have some application which grabs some ARM resource - DSP, SGX, audio etc. VDD2 is automatically set by calculation of overall bandwidth of internal CPU bus - it boosts voltage if there is some transfer via internal bus to special ARM CPU resource.

You have VDD2 == 3 even with 125MHz and that is probably some energy leak on low frequencies.