despite my Dell Inspiron 531 being certified for SLED/SLES This one is certified? Then we should have one somewhere around, I try to find
From your initial bug description: But the following command shoud output 1800000: cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq 800000 It looks like we have more than a sampling or say cpufreq algorithm problem. If scaling_max_freq is lowered to the lowest the dynamic freq algorithm can't do anything and cur_freq will always stay at lowest. It looks like you only run into this sometimes (maybe this got fixed with the latest BIOS?)? If not set through a BIOS option (you might want to check your BIOS configs),
https://bugzilla.novell.com/show_bug.cgi?id=436717
User trenn@novell.com added comment
https://bugzilla.novell.com/show_bug.cgi?id=436717#c12
Thomas Renninger
while (true); do cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_cur_freq | grep -v 1000000; done you might want to use: watch -n1 cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_{cur_freq,max_freq} to monitor current and maximum frequency. watch -n1 refreshes the output every second.
For the algorithm: I want to look closer at this, but did not have the time yet. The load threshold when to switch up the frequency and a patch series for improved sampling on no_hz kernels I'd still like to test for SLE11 (which would get integrated/backported into 11.1 then also). Benchmarking power vs performance is time consuming, but we have some nice tools now and I hope to be able to work on that soon. For this bug report I'd only like to examine the situation when max_freq gets lowered. Do you use libsensors? This might falsify ACPI temperature readings and could cause the kernel to lower max_freq, due to wrong, high temperature values. If you could reproduce when/why max_freq gets lowered (or if at all with the new BIOS), this would be great. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.