I tought these were the frequency steps but they seem to be a different thing: Under linux it seems they are constantly around 2000 RPM.Ĭould the FanControl program be of any use for you developers? The fans's RPM's adjust dynamically non-stop between 20 when not doing anything heavy. I've tried the FanControl (free and open source) program for OSX. # Set ondemand cpu frequency scaling governorĮcho ondemand > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor # Maybe test 10 seconds and compare against 5?Įcho 500 > /proc/sys/vm/dirty_writeback_centisecsĮcho max_performance > /sys/class/scsi_host/host0/link_power_management_policyĮcho max_performance > /sys/class/scsi_host/host1/link_power_management_policy # Wakeup every 5 seconds to check if dirty pages writing necessary # Set kernel dirty page value back to defaultĮcho 5 > /proc/sys/vm/dirty_background_ratio # Set powersave cpu frequency scaling governorĮcho powersave > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor # Wakeup every 15 seconds to check if dirty pages writing necessaryĮcho 1500 > /proc/sys/vm/dirty_writeback_centisecs # Set kernel dirty page value back to default Reduce disk activityĮcho 10 > /proc/sys/vm/dirty_background_ratioĮcho min_power > /sys/class/scsi_host/host0/link_power_management_policyĮcho min_power > /sys/class/scsi_host/host1/link_power_management_policy # Set "aggressive" power savings which then was installed in the directories /etc/pm/power.d and /etc/pm/sleep.d as suggested in this thread: Hopefully someone has the experience so we can set some "default" settings for the MBP 5,1.īut I am really curious to know why there is this huge difference? Is power management on Ubuntu (linux in general) so bad? Probably there are more things to play with. The 5 hours reported in OSX make the "less than 2 hours" in Ubuntu look really bad. * On Ubuntu Intrepid Ibex working only with battery and with some power-saving settings :
* 100% charged on OSX reports 5:30 time remaining MBP 5,1 - Battery duration on Ubuntu Intrepid Ibex What should one do to protect the battery from being damaged? Today the laptop didn't make it over 1 and a half hour :-( The above mentioned "changes" are not permanent/not saved. Am I right thinking that the "temp5" is the GPU's sensor indication and not the CPU as I read in some post (don't remember which post was that)? The fan(s) started to run like crazy (is it the GPU cooler or the processor's cooler) and the temp went after some minutes below 80. The maximum I got yesterday was ~82 (a bit more/a bit less) while compiling several stuff and running lots of applications intentionally at the same time. Maybe the problem occurs when moving between OSX and Ubuntu via a reboot, which does not clear the NVRAM registers, but this is just a guess.
COREDUOTEMP URL ACCESS SOFTWARE
With no software control of the fans, which is the case with the macbooks (fancontrol and pwmconfig does not work), the fan speed seems to be controlled automatically by the SMC without issues - but only if fan1_manual is zero. So, for those that have reported high temperatures due to low fan speeds: check the value ofĬat /sys/devices/platform/applesmc.768/fan1_manual
The fan immediately runs up to 6200 rpms, and then slowly returns as the CPU gets cooler. I watch the temperature reach 85 degC, then I switch fan1_manual back to zero, and abort the experiment. The temperature reaches 80 degC, but the fan speed stays at 2500 rpms. Now I switch the fan1_manual to one and repeat the process. After a couple of minutes, the fan speed slowly drops back to its idling value of 2500 rpms, leaving the CPU at 50 degC. The CPU temp very quickly drops to 60 degC. Now, I stop the compilation, turn the CPU governor to power save, and watch the sensors. At this point, the fan speed starts to increase, and after a while reaches the peak 6200 rpms. Checking with sensors, the CPU temp quite quickly reaches 80 degC.
COREDUOTEMP URL ACCESS FULL
With fan1_manual set to zero, I put the CPU governor to full throttle, then start a kernel compilation. I just ran a little experiment on my MBA.
0 kommentar(er)
