> Hi!
> I've built a new machine with an ASUS P5N32-E SLI PLUS motherboard,
> equipped with an Intel(R) Core(TM)2 Quad CPU. The system seems stable
> and well-performing. However, I'm curious about the following:
> In the BIOS settings, there is an "Enhanced Intel SpeedStep (TM)" item,
> which can be disabled or enabled. I've tried to hook a cpufreq module on it
> but it seems to be unsuccessful. Neither speedstep nor acpi-cpufreq modules
> are recognizing the hardware, so the feature seems unsupported.
> However, when the BIOS SpeedStep setting is enabled, strange things happen.
> In /proc/cpuinfo, the CPU frequency is always reported as 900 MHz (while
> normally it's 2.6GHz), but the bogomips value is at 5220, which corelates
> with the standard 2.6G clock. Also the system performance is not so low as
> for 900M clock - I can see a small slowdown, but it may be about 5%, not more.
> Even with an artificial load of all cores (4 cpu-eating tasks run concurrently)
> the frequency reported by /proc/cpuinfo is at 900 MHz and doesn't never
> increase.
> So what does it mean ? Is the SpeedStep feature usable just by setting the
> BIOS ? Isn't it limiting the performance too much ? Why the clock always seems
> to be low ? I'm keeping it off until I know all the answers :-).
> I'm attaching my DSDT for reference.
> With regards,
> Pavel Troller
Hi!
I'm here again with a new info, which may be important.
When the "Intel Enhanced SpeedStep" feature is active, there are 5 additional
SSDT tables, which are not there, when the feature is disabled. Linux states
that the tables are there, but I can't see any other difference than that
with exception of another ordering of CPU cores and some minor timing issues.
Can Linux profit from those tables ? Their dissassembly is attached.
With regards, Pavel Troller
Attachment:
ssdt.tar.bz2
Description: Binary data
