I have a nagging problem that is driving me batty.
I have a Dell Precision M4500:
Linux m4500 2.6.36-gentoo-r6 #1 SMP Wed Dec 29 07:57:47 PST 2010 x86_64 Intel(R) Core(TM) i7 CPU M 620 @ 2.67GHz GenuineIntel GNU/Linux
model name***** : Intel(R) Core(TM) i7 CPU****** M 620* @ 2.67GHz
and it even has these fancy capabilities:
however, I cannot get the thing to change speeds. It is adamantly stuck at its pokey molasses slow 1.2GHz:
blongman@m4500 /sys/devices/system/cpu/cpu0/cpufreq $ cat scaling_*
2667000 2666000 2533000 2399000 2266000 2133000 1999000 1866000 1733000 1599000 1466000 1333000 1199000
A strangeness I have noted is that /proc/cpuinfo has this for its power capabilities:
power management:
Nothing.
I don't recall what it was on the other i7, but my AMDs have this sort of thing:
power management: ts ttp tm stc 100mhzsteps hwpstate
Why this is driving me batty is that I just this summer set up a similar M4500, but with a fancier i7 in it, *specifically* to enable and manage its power capabilities and I was completely successful. If I swap my disk and boot Windows 7, it behaves like a champ, so I don't think it's a BIOS issue. What else could I look at since I've been through all versions of kernels from 2.6.32 to today's 2.6.36?
--
Bill Longman
12-29-2010, 04:40 PM
Paul Hartman
Core i7 M620 power management problem
On Wed, Dec 29, 2010 at 11:13 AM, Bill Longman <bill.longman@gmail.com> wrote:
> A strangeness I have noted is that /proc/cpuinfo has this for its power
> capabilities:
> power management:
> Nothing.
FWIW I have Core i7 920, and it also has nothing in the power
management in cpuinfo, but CPU frequency scaling does work and speeds
change (using ondemand governor, in a desktop machine).
So it seems similar to yours except that your max_freq and min_freq
are the same! Which matches what you say about it never going faster
than the minimum speed.
In kernel docs Documentation/cpu-freq/cpu-drivers.txt there is some
information about how the min and max speed are set by the policy that
is in use (by whatever driver is controlling the scaling). So I don't
know if there is a userspace program (like KDE laptop stuff) that
might be overriding with its own faulty settings?
Seems like your kernel settings are probably okay since you can see
all of that so far. If you run powertop can it see all of the C-states
and P-states without any problems?