Jump to content

SpeedStepper (now supports Mountain Lion 10.8.3)

flAked

By flAked,
in Developers Corner

 Share
574 posts in this topic

Recommended Posts

geticus
geticus
Posted
Posted
Writing DSDT=/Extra/DSDT.aml is different of writing DSDT=/Extra/DsDt.aml

 

Right, I'm always using DSDT.aml. So, is not here the problem.

 

Maybe I was wrong when I compiled ssdt_i5_2400.dsl from your first page with DSDTSE ?

Is this the good way ?

Link to comment
Share on other sites
peebee01
peebee01
Posted
Posted

After doing a clean install of LION GM I used speed stepper to patch AppleIntelCPUPowerManagement, added my <vendor-id><device-id> to the list of ids in AppleLPC.kext info.plist and removed NullCPUPM from /E/E before rebooting.

 

After reboot I see only two power states:

7/17/11 4:58:17.000 PM kernel: MSRDumper PStatesReached: 16 43

Not the eight states I used to see on SL with the same over clock settings.

 

I did not make any changes to /S/L/E/IOPlatformPluginFamily.kext/C/P/ACPI_SMC_PP/C/R as imac12_2.plist is already embedded, correct?

 

Replaing my current imac12_2 definition in smbios.plist with the Macbook 8.3 SMBIOS.plist posted by oSxFr33k I get 5 power states. So some improvement but Flaked indicated we're better sticking with imac12_2 definition, right?

 

So what's best to try to get all power states? Trying different smbios.plist? Or would we need a new version of speedstepper?

Link to comment
Share on other sites
joshua7
joshua7
Posted
Posted
Hi all and thank you for what you're doing

 

Joshua7 I don't think your speedstep is working, mine is not and CPU is stuck at 1.6 Ghz.

I'm running on MSI P67A-GD65 and I have patched AppleIntelCPUPowerManagement.kext with Flaked Speedsteper and now is loading and sleep is working without SleepEnabler.kext.

I've tried also to ad to Extra an SSDT.aml for my i5 2400 but nothing changed.

 

@buoo How to Boot dropping patched acpi tables ?

 

So, here we go...

 

geticus, you were right... SpeedStep was not working, I misinterpreted the blinking LEDs, MSRDumper made it clear. My i7 was running at 1.6 Ghz :)

 

I tested "DropSSDT=Yes" and "DropSSDT=No", no effect. When I renamed SSDT.aml to SSDT-1.aml, my system did not boot up anymore ("SSDT.aml not found")

 

Then I tried to use another smbios.plist. My choice was MacPro3,1 of tonymacx86's [url=&quot;http://www.insanelymac.com/forum/topic/279450-why-insanelymac-does-not-support-tonymacx86/&quot;]#####[/url] 3.8.

 

Here the results... ;)

 

post-677453-1310920969_thumb.jpg

 

So five PStates are reached, Geekbench result is pretty OK for a stock i7 2600K. I'm happy with that ;)

 

When I was using iMac12,2.plist I did not add it in the ACPI_SMC_PlatformPlugIn folder... no idea if this would have changed something.

 

Best regards,

joshua

Link to comment
Share on other sites
joshua7
joshua7
Posted
Posted

actually, I don't care if it's nonsense... it works.

 

seems like it would not work, if I copied the smbios to plugins folder - as you can see in geticus' case.

 

...but maybe I will try it, just to see if you're right and the i7 would reach more PStates.

 

EDIT:

 

I tested the iMac12_2.plist... not successful. My System is not properly recognized, it says imac 12,1 after rebuildcaches and reboot.

 

post-677453-1310988643_thumb.png

 

I added the plist in the specific Plugins folder, but when rebuilding caches, I get a message that the "IOPlatform-whatever.kext" seems to be damaged and cannot be used.

 

Speedstep is not working, CPU is stuck at 3.4 GHz, no Turbo.

 

post-677453-1310988666_thumb.png

 

So right now, MacPro3,1 is the better choice for me... and actually my G5-Casemod looks more like a MacPro :thumbsdown_anim:

Link to comment
Share on other sites
geticus
geticus
Posted
Posted
So, here we go...

 

geticus, you were right... SpeedStep was not working, I misinterpreted the blinking LEDs, MSRDumper made it clear. My i7 was running at 1.6 Ghz :P

 

I tested "DropSSDT=Yes" and "DropSSDT=No", no effect. When I renamed SSDT.aml to SSDT-1.aml, my system did not boot up anymore ("SSDT.aml not found")

 

Then I tried to use another smbios.plist. My choice was MacPro3,1 of tonymacx86's [url=&quot;http://www.insanelymac.com/forum/topic/279450-why-insanelymac-does-not-support-tonymacx86/&quot;]#####[/url] 3.8.

 

Here the results... :(

 

post-677453-1310920969_thumb.jpg

 

So five PStates are reached, Geekbench result is pretty OK for a stock i7 2600K. I'm happy with that :P

 

 

Best regards,

joshua

 

joshua, glad to hear that is working for you, this give some hope for my rig to get speedstep working.

I tried with MacPro 3.1 system definition but is not working for me. Maybe I was wrong when I compiled ssdt_i5_2400.dsl with DSDTSE to get SSDT.aml ? Is this the good way ? In boot I can read SSDT.aml not found if I use SSDT-1.aml and vice versa.

Link to comment
Share on other sites
joshua7
joshua7
Posted
Posted
joshua, glad to hear that is working for you, this give some hope for my rig to get speedstep working.

I tried with MacPro 3.1 system definition but is not working for me. Maybe I was wrong when I compiled ssdt_i5_2400.dsl with DSDTSE to get SSDT.aml ? Is this the good way ? In boot I can read SSDT.aml not found if I use SSDT-1.aml and vice versa.

 

Are you using Chimera or Chameleon as bootloader? I'm on Chimera, works flawless...

 

Chimera and Chameleon (in recent builds) are searching for "SSDT.aml" in your extras folder. So don't rename it to SSDT-1.aml - just let it "SSDT.aml".

 

I don't know how to extract oder modify a SSDT, so if yours is not alright, you might research for a solution...

 

EDIT: I read you are using the Legacy-Kernel? This might be the problem... you should be using the Darwin-Kernel. With the newest Chimera build, this is possible on a MSI board.

Link to comment
Share on other sites
geticus
geticus
Posted
Posted

Joshua,

I am using vanilla kernel coming with 10.6.8 of course, and tried with Chimera latest or newest Chimera build, but is still not working for me. MacPro 3.1 doesn't change enything either.

I'm still asking myself about how to obtain SSDT.aml from SSDT.dsl for my i5 2400 ?

Link to comment
Share on other sites
downlord
downlord
Posted
Posted
P.S. for lenovo laptop owners: ThinkPad BIOS also blocks write operations to 0xe2, confirmed with latest BIOS BITS 438, you have to use the patch to get native PM.

 

Tht's interesting. Could be the reason why I'm getting unpredictable behaviour with a t60p, sometimes Steedstep is working, sometimes not.

Thanks for the input. Will make some tests...

Link to comment
Share on other sites
flAked
flAked
Posted
  • Author
Posted
I did not make any changes to /S/L/E/IOPlatformPluginFamily.kext/C/P/ACPI_SMC_PP/C/R as imac12_2.plist is already embedded, correct?

 

Replaing my current imac12_2 definition in smbios.plist with the Macbook 8.3 SMBIOS.plist posted by oSxFr33k I get 5 power states. So some improvement but Flaked indicated we're better sticking with imac12_2 definition, right?

 

So what's best to try to get all power states? Trying different smbios.plist? Or would we need a new version of speedstepper?

The iMac12_2.plist is included in 10.6.8 and beyond, yes.

 

The main component is the stepper-program defined in that plist. I've tested several other stepper-programs of different models and got a slight deviation every time. The stepper-program seems to be stripped down in 10.7 for iMac12_2.

 

The only solution is to use the old stepper-program from 10.6.8.

 

SpeedStepper is not a solution to get speedstep working fully, it only prevents a KP of AICPUPM.

 

 

 

23.07.11 12:07:40,000 kernel: MSRDumper PStatesReached: 8 11 14 17 18 20 
23.07.11 12:07:40,000 kernel: MSRDumper CoreMulti(8)

 

on i7-2630QM. Is that okay?

Looks good, which model identifier are you using?

 

 

flAked can we patch bios to avoid patching AICPM by SpeedStepper ?

Yes. Working on it, only for the P8P67 though.

 

Only Sandy bridge.

I'd call it a ASUS bug and I fear they will use the same UEFI component for the next generations of motherboards, so no fun for ASUS owners in the future (possibly!).

Link to comment
Share on other sites
oSxFr33k
oSxFr33k
Posted
Posted

As far as I know my Asus Laptop G51JX also has a UEFI but its an i7 first generation 720QM HM55 chipset. I am pretty sure its a UEFI Bios but its does not seem to be affected? Maybe it is and I just don't know it :wacko:

 

Is there a way to see if this MSR lock at 0x2E is affected in any other Asus MB besides the P8P67?

 

I think the G51JX HM55 chipset should be ok because there is no kernel panic with Lion 10.7 using vanilla AICPUPM. On the other hand the Asus G74SX Sandy Bridge seems to kernel panic or not even boot into the Desktop without the speedstepper patch.

 

Both of these models are Asus Gamers Laptops.

Link to comment
Share on other sites
flAked
flAked
Posted
  • Author
Posted
As far as I know my Asus Laptop G51JX also has a UEFI but its an i7 first generation 720QM HM55 chipset. I am pretty sure its a UEFI Bios but its does not seem to be affected? Maybe it is and I just don't know it :)

 

Is there a way to see if this MSR lock at 0x2E is affected in any other Asus MB besides the P8P67?

If C-State configuration is locked in UEFI, then AICPUPM panics on boot and shows 0x2E in ECX in the verbose panic description.

 

It seems that this lock was introduced with SandyBridge UEFIs and many P8* desktop motherboards are affected as well as some notebooks (can't remember which models).

 

The C-State configuration is part of a generic CPU UEFI component and is used across all P8* models as far as I can tell.

 

 

Anyone tested 10.7.2 for a new AICPUPM?

Link to comment
Share on other sites
buoo
buoo
Posted
Posted

I've already tested with success, but there is something strange in my kernelog.

I'll retry it on another Lion's installation. soon the result...

 

Edit: It works good, I discover the reason of those strange messages in my kernel.log. The speed stepper is innocent :)

 

 

Report:

SpeedStepper v1.2 - © flAked 2011
for AICPUPM v166.0.0 - Lion GM

Searching for wrmsr #0: a4fe
-> found bytes.
-> patched.

Searching for wrmsr #1: a5c6
-> found bytes.
-> patched.

Searching for wrmsr #2: a644
-> found bytes.
-> patched.

Searching for wrmsr #3: a68d
-> found bytes.
-> patched.

Searching for wrmsr #4: af48
-> found bytes.
-> patched.

Searching for wrmsr #5: affa
-> found bytes.
-> patched.

Searching for wrmsr #6: b0b8
-> found bytes.
-> patched.

Searching for wrmsr #7: b6d5
-> found bytes.
-> patched.

Searching for wrmsr #8: 11abe
-> found bytes.
-> patched.

Searching for wrmsr #9: 11b7c
-> found bytes.
-> patched.

All done. Bye 
bash-3.2#

Link to comment
Share on other sites
androdev
androdev
Posted
Posted
Looks good, which model identifier are you using?

Tried both MacBook Pro 8,2 and 8,3. The 8,2 is more close, because it has i7-2635QM (same as mine i7-2630QM).

BUT

Looks like I miss turbo states, because only 8-20 is reached, but maximum turbo is 29. How can I fix missing turbo?

 

Thank you!

Link to comment
Share on other sites
flAked
flAked
Posted
  • Author
Posted

You need to have all P-States in the APSS object of CPU0 in your DSDT/SSDT, but it's much easier to use RevoBoot and let it auto-filled for you.

 

I don't know if GeneratePStates with Chameleon works for SandyBridge.

 

 

Maybe Turbo is disabled in UEFI or set wrongly? What if you search kernel.log for "ratio"?

 

Thats mine:

AppleIntelCPUPowerManagement: Turbo Ratios 1234

Link to comment
Share on other sites
 Share
Go to topic listing
×
×
  • Create New...