Jump to content

Vanilla SpeedSteep: _LPC/_PSS added but Vcore still constant, CPUPLimit=0x03


1 post in this topic

Recommended Posts

Hi,

Maybe somebody can hint me where issue is:

 

I have GA-P35-DS4 with OC Q6600 (3,6G@1,375V). BIOS is F14, bootloader is Clover r1050, OS is ML 10.8.2.

Basically OC HW/OS seems to be 100% functional - also SpeedSteep seems to be working - but only for freq. Vcore is costant (1,375V)

 

What I want to achieve is FULL SpeedSteep (Freq & Vcore) on OC HW (Q6600@3,6G, FSB=400, Vcore=1.375).

 

Currently I'm able to get FULL SpeedSteep if HW is NOT OC (Q6600@2,4G, FSB=266, Vcore="Normal").

 

I already added _PSS & LPC sections to DSDT.

My _PSS have declared 4 P-states for 1600, 2000, 2800, 3600 frequencies. I disabled P-state/C-state generation in Clover.

With those modifications launching IORegExpl shows LPC in device tree, AICPMVers is present, I see my 4 P-states in PerformaneStateArray, but I'm geting CPUPLimit=0x03.

 

Basing on other forum threads - CPUPLimit=0x03 means native PM isn't working.

 

With above config:

1. I see that load causes CPU freq changing. HWmonitor reports 2,4-3,6G - but never to 1,6G. 2,4G isn't defined in my _PSS - so something wrong is here....

2. Vcore is constant despite Vid entries in _PSS.

3. P-States/C-States in SMC monitor are reported correctly (1600, 2000, 2800, 3600) - but for some reason system behaves like ignoring my _PSS entries in DSDT and is using it's own (lower freaq is always 2,4G and Vcore is always constant).

 

What I investigate:

-I can get dynamic Vcore ONLY if BIOS has Vcore option set to "Normal". Any change Vcore option from "normal" causes SpeedSteep Vcore becomes constant.

-above PM SpeedSteep issue is the same in Chameleon r2016 & clover r1050.

-all this pretends like my system is using ACPI/BIOS PM, while OSX PM runs in parallel but without any effect on HW.

 

May somebody hint me where issue might be?

 

Thx in advance

×
×
  • Create New...