New: p4-clockmod gets a wrong max clock (1.2GHz insted 1.6GHz)

http://bugzilla.kernel.org/show_bug.cgi?id=10653

           Summary: p4-clockmod gets a wrong max clock (1.2GHz insted
                    1.6GHz)
           Product: Power Management
           Version: 2.5
     KernelVersion: 2.6.25
          Platform: All
        OS/Version: Linux
              Tree: Mainline
            Status: NEW
          Severity: normal
          Priority: P1
         Component: cpufreq
        AssignedTo: [email protected]
        ReportedBy: [email protected]

Latest working kernel version: I don't know
Hardware Environment: P4 1.6 GHz
Problem Description:
My p4 1.6GHz gets a wrong max speed using p4-clockmod. Available frequencies taken from cpufreq-info (same output for cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies).

Using acpi-cpufreq:
1.20 GHz, 1.60GHz
using p4-clockmod:
150 MHz, 300 MHz, 450 MHz, 600 MHz, 750 MHz, 900 MHz, 1.05 GHz, 1.20 GHz

I'm forced to use acpi-cpufreq :).

Useful cpuinfo for model and stepping:
processor : 0
vendor_id : GenuineIntel
cpu family : 15
model : 2
model name : Intel(R) Pentium(R) 4 Mobile CPU 1.60GHz
stepping : 4
cpu MHz : 1200.000
cache size : 512 KB
fdiv_bug : no
hlt_bug : no
f00f_bug : no
coma_bug : no
fpu : yes
fpu_exception : yes
cpuid level : 2
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat
pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm up pebs bts
bogomips : 2395.99
clflush size : 64

Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.kernel.org/show_bug.cgi?id=10653

------- Comment #1 from [email protected] 2008-05-08 18:18 ------- Another notice:
I've just done a benchmark (PI calculation) and the 1.2GHz reported using p4-clockmod is not a fake number, it is _really_ working at 1.2GHz, because with acpi-cpufreq the PI calculation is done in 14 seconds, while with p4-clockmod set at full speed I reach PI calculation in 18 seconds.

Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.kernel.org/show_bug.cgi?id=10653

------- Comment #2 from [email protected] 2008-05-11 09:21 ------- the 'frequencies' that p4-clockmod reports are utterly bogus. they always have been. This driver doesn't actually scale frequency at all. It does thermal throttling, which is usually expressed as a percentage. There's discussion going on right now on cpufreq-list whether or not to actually just remove p4-clockmod entirely, as it's a cause of frequent confusion with very few benefits.

Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.kernel.org/show_bug.cgi?id=10653

------- Comment #3 from [email protected] 2008-05-11 10:21 ------- I've reached to use p4-clockmod with 1.6GHz.

The issue is that p4-clockmod reads the cpu speed when it is loaded (modprobe), sets it as the maximum frequency then generates the lower available ones.

In fact, if I load the module when my laptop is running at 1.2GHz (for example when using built-in speedstep technology of my motherboard) p4-clockmod will give me these results (with 1.2GHz as maximum available), though if I load p4-clockmod when my frequency is 1.6GHz (when my laptop is in AC) then i get the right ones: cat /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies 200000 400000 600000 800000 1000000 1200000 1400000 1600000

To fix it I have disabled speedstep hardware management from the BIOS, so when p4-clockmod is loaded in the bootup the frequency is 1.6GHz (no speedstep) and it gets the right frequencies.

P4-clockmod is fine over acpi-cpufreq (in fact acpi-cpufreq gives me only 1.2 and 1.6 which is quite useless for power-management), but maybe it should be fixed when getting the right CPU info after is first loading.

Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.kernel.org/show_bug.cgi?id=10653

------- Comment #4 from [email protected] 2008-05-11 10:35 ------- actually, odds are that you'll be saving more power with acpi-cpufreq. When that switches to 1.2Ghz, you're also running at a lower voltage.

p4-clockmod runs at the same voltage all the time. It doesn't even change clock frequency. All it does is insert 'wait' states every N clock cycles.

Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.kernel.org/show_bug.cgi?id=10653

------- Comment #5 from [email protected] 2008-05-11 12:30 ------- I haven't say I was saving more power, I just said I was running at lower speed (and I thought p4-clockmod was lowering the voltage)
So with acpi-cpufreq I should save more power even if it runs 1.2Ghz vs 400Mhz? (400Mhz is enough for coding :) )

Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.kernel.org/show_bug.cgi?id=10653

------- Comment #6 from [email protected] 2008-05-15 10:20 ------- You're not actually running at 400Mhz with p4-clockmod. All it's doing is
basically idling every N clocks. See comment #2.

Configure bugmail: http://bugzilla.kernel.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.