Jump to content

DSDT for Asus P8P67-M PRO

Time2Retire

By Time2Retire,
in Patches

 Share
834 posts in this topic

Recommended Posts

flAked
flAked
Posted
Posted
Also note this:

AppleIntelCPUPowerManagement: Turbo Ratios 1BBBA

Think that's not correct...

The new version of MSRDumper shows you the current multiplier in kernel.log.

 

TurboRatios should be 1234 on stock settings, see DHP's post a while back on how to reset it correctly.

 

p.s. Lights went out here in a burst. Every single light bulb in the house that was on is now broken. Using a LED tube. Sucks. Can't work in the dark.

Wow, that sucks. I hope you got no fried hardware!

 

Did you find something new about the on-demand clock modulation?

Link to comment
Share on other sites
mrmojorisin17
mrmojorisin17
Posted
Posted
The new version of MSRDumper shows you the current multiplier in kernel.log.

Yep ;)

May 24 01:10:19 iHack122-di-Mr-Mojo-Risin kernel[0]: MSRDumper: multiplier: 33
May 24 01:10:24 iHack122-di-Mr-Mojo-Risin kernel[0]: MSRDumper: multiplier: 16

TurboRatios should be 1234 on stock settings, see DHP's post a while back on how to reset it correctly.

It should be 1234 also with Turbo Mode enabled?

What I have to look?

 

BTW iMac12_2.plist doesn't work here, always get blue screen after verbose...

Any suggestion?

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
Wow, that sucks. I hope you got no fried hardware!

Yeah. And I can't use that – sold my hardware already. That would be a blow in the face. For both parties.

 

Did you find something new about the on-demand clock modulation?

No. No idea yet. I also noticed that it jumps from 16 to 34 (and back). Turbo works of course.

 

Note that I disabled EIST/C1E. Will now try to add:

gPEClockFrequencyInfo.cpu_frequency_min_hz = 1600000000;
gPEClockFrequencyInfo.cpu_frequency_max_hz = 3400000000;

Doesn't seem to do anything. Fixed sysctl alright but that's about it.

 

@mrmojorisin17: read this (HOWTO reset the Turbo Ratios).

 

Ah. There are two different kinds of MSR's. First there are global, or package, MSR and then there's the per-core MSR. Some MSR's have to be set on a per-core basis. As in on all cores. Food for thought - for later today. Getting tired. Was a long day. Good night all.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted

Didn't find anything clear about how to read/write msr for multiple cpus.

 

Coreboot seems to just iterate over cpu_num doing rdmsr. Does that mean that if I do 4 rdmsr it cycles through the cpus itself?

 

Linux kernel is using smp_call_function_single in smp.c and saving some irqs before and after the func.

 

 

First step would be to read the multipliers for all cpus.

Link to comment
Share on other sites
mrmojorisin17
mrmojorisin17
Posted
Posted
@mrmojorisin17: read this (HOWTO reset the Turbo Ratios).

I suggested that post...

Now Turbo Ratios is correct:

AppleIntelCPUPowerManagement: Turbo Ratios 1234

Does your sleep and deep-sleep works correctly there?

Here:

  • sleep -> sleep works, but I don't have a correct wake, in fact I have to shutdown the PC with power button after waking;
  • deep-sleep -> all appears to be ok, sleep-image is written and then PC shutdown, but at next boot sleep-image is not correctly loaded and PC start normally.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
I suggested that post...

Now Turbo Ratios is correct:

I think you mean something different? :( Glad it worked, stupid ASUS bug.

 

Does your sleep and deep-sleep works correctly there?

Sleep works, but I do get the GPU glitches I mentioned (flickering) upon wake.

 

And RevoBoot doesn't support booting from a deep-sleep image at the moment.

Link to comment
Share on other sites
mrmojorisin17
mrmojorisin17
Posted
Posted
I think you mean something different? ;) Glad it worked, stupid ASUS bug.

;)

Sleep works, but I do get the GPU glitches I mentioned (flickering) upon wake.

Not here :(

And RevoBoot doesn't support booting from a deep-sleep image at the moment.

Like I presuppose...

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
Ouh, some action over at hardforum, finally some response from Asus!

Yeah. They needed a kick in the...

 

BTW. Reading MSR's from different cores is easy. First you let it sleep, which triggers it to switch to a different core, and then you read it. It' all in CPUi.cpp Look at loopTimerEvent and specifically this part of it: 

	// State Readout
IOSimpleLockLock(SimpleLock);
mp_rendezvous_no_intrs(IntelState, &magic);
IOSimpleLockUnlock(SimpleLock);

Now something else. Please verify this:

 

1.) Disable EIST/C1E in UEFI settings.

2.) Boot.

3.) Run Geekbench.

4.) do a tail/var/log/kernel.log

 

Repeat the steps but now with EIST/C1E enabled. Please attach the output because I think to have found a way to boot quicker (will update RevoBoot sources when confirmed).

Link to comment
Share on other sites
flAked
flAked
Posted
Posted

Will test it after I make some coffee :(

 

Did you see my post on Page #139 at hardforum? How can we test this?

 

4.) do a tail/var/log/kernel.log

This will usually output vmnet info on my system, which info do you need? Current multiplier from MSRDumper?

Link to comment
Share on other sites
mrmojorisin17
mrmojorisin17
Posted
Posted
Now something else. Please verify this:

 

1.) Disable EIST/C1E in UEFI settings.

2.) Boot.

3.) Run Geekbench.

4.) do a tail/var/log/kernel.log

 

Repeat the steps but now with EIST/C1E enabled. Please attach the output because I think to have found a way to boot quicker (will update RevoBoot sources when confirmed).

Here we go ->Archivio.zip

Any suggestion regarding this?

Also don't solve issue regarding iMac12,2 mac model identifier. Always get blue screen after boot using iMac12_2.plist in /System/Library/Extensions/IOPlatformPluginFamily.kext/Contents/PlugIns/ACPI_SMC_PlatformPlugin.kext/Contents/Resources/.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
Also don't solve issue regarding iMac12,2 mac model identifier. Always get blue screen after boot using iMac12_2.plist in /System/Library/Extensions/IOPlatformPluginFamily.kext/Contents/PlugIns/ACPI_SMC_PlatformPlugin.kext/Contents/Resources/.

Sorry, without more data you'll have to do some debugging on your own. A blue screen suggests some sort of gpu issue, though.

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
Will test it after I make some coffee :)

LOL Coffee? What about food?

 

Did you see my post on Page #139 at hardforum? How can we test this?

You mean if turbo works without EIST? Look at the dump I asked you for :)

 

This will usually output vmnet info on my system, which info do you need? Current multiplier from MSRDumper?

What I am after here is the ratio scale with and without EIST enabled in UEFI.

 

Here turbo ratio only works (the full scale) when I disable EIST. Then I see MSRDumper: multiplier: 16, 35, 36, 37 and 38 but as soon as I enable EIST I only see 16 and 34 – hence the lower Geekbench score.

 

I found a way to boost it one stage, by setting an MSR, but AICPUPM handles turbo states in a far more relaxed way, compared to what the hardware/UEFI does, and thus I want to use the kext instead.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
I'm checking AGPM.kext Info.plist...

See that in DHP's AGPM.kext Info.plist there's a part of code regarding iMac12,2 that's not present in mine AGPM.kext Info.plist.

Search the thread please, DHP posted a diff. You also need to change vendor/device in that plist.

I currently don't use the mod, because I want to understand the QSC package first. I do have the iMac12,2 plist in there, but it shouldn't match on the stock device ids present there.

 

Ok so EIST/C1 disabled, Geekbench: 9730

May 24 12:58:41 slave kernel[0]: MSRDumper: multiplier: 36
May 24 12:58:46 slave kernel[0]: MSRDumper: multiplier: 16
May 24 12:58:56: --- last message repeated 1 time ---
May 24 12:58:56 slave kernel[0]: MSRDumper: multiplier: 35
May 24 12:59:06: --- last message repeated 1 time ---
May 24 12:59:06 slave kernel[0]: MSRDumper: multiplier: 34
May 24 12:59:11 slave kernel[0]: MSRDumper: multiplier: 35
May 24 12:59:16 slave kernel[0]: MSRDumper: multiplier: 34
May 24 12:59:21 slave kernel[0]: MSRDumper: multiplier: 35
May 24 12:59:26 slave kernel[0]: MSRDumper: multiplier: 16
May 24 12:59:31 slave kernel[0]: MSRDumper: multiplier: 22
May 24 12:59:36 slave kernel[0]: MSRDumper: multiplier: 16

 

EIST/C1 enabled, Geekbench: 9600, 9700, 9750 (wth?)

May 24 13:03:30 slave kernel[0]: MSRDumper: multiplier: 30
May 24 13:03:35 slave kernel[0]: MSRDumper: multiplier: 16
May 24 13:03:40 slave kernel[0]: MSRDumper: multiplier: 34
May 24 13:03:46 slave kernel[0]: MSRDumper: multiplier: 35
May 24 13:03:50 slave kernel[0]: MSRDumper: multiplier: 22
May 24 13:03:55 slave kernel[0]: MSRDumper: multiplier: 34
May 24 13:04:01 slave kernel[0]: MSRDumper: multiplier: 35
May 24 13:04:05 slave kernel[0]: MSRDumper: multiplier: 16
May 24 13:04:20: --- last message repeated 2 times ---
May 24 13:04:20 slave kernel[0]: MSRDumper: multiplier: 34
May 24 13:04:30: --- last message repeated 1 time ---
May 24 13:04:30 slave kernel[0]: MSRDumper: multiplier: 16

 

By the looks of it, this is not conclusive for the i5-2500K?

Link to comment
Share on other sites
dhy8386
dhy8386
Posted
Posted

hey guys

 

Been following the thread and its amazing stuff. I have one question, and sorry if its a distraction so feel free to pass on it for now...

 

I have the i5 2400 and not sure how i would adjust my ssdt_pr file to reflect the turbo ratios. I was looking at DHPs latest ssdt_pr compared to flAkeds latest for the 2500K and it is not just simply a few lines deleted to remove extra frequencies.

 

Besides this one item, i think i got everything else working with your files in my P8H67-I Deluxe...

 

Much thanks for any help or direction.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
I have the i5 2400 and not sure how i would adjust my ssdt_pr file to reflect the turbo ratios. I was looking at DHPs latest ssdt_pr compared to flAkeds latest for the 2500K and it is not just simply a few lines deleted to remove extra frequencies.

Sorry about that, it's not exactly diff-friendly, it's based on my old revision. But the only differences are in APSS, _PSS and the processor value in PSD. I'll get back to you, once I have some time on my hand.

 

LOL Coffee? What about food?

Hehe, always comes with some toast and jam though :)

 

You mean if turbo works without EIST? Look at the dump I asked you for :)

More like how to send him our results. I found a tool for windows to directly read MSR's. So I could do some screenshots with CPU-Z. But I would need to be sure if EIST is really disabled in windows.

 

BTW, if you need more granularity, set the event timeout to 5.

 

 

Note that I have 0x1A0: 0x850089 for each test run, so the EIST bit didn't change (the UEFI bug).

Link to comment
Share on other sites
flAked
flAked
Posted
Posted

Now, here a some insights. Raja said they tied TurboMode and EIST together. That would mean if one disables EIST and leave TurboMode on, then EIST would still be enabled.

 

But, testing shows:

1) EIST + Turbo enabled: 0x850089
2) EIST disabled + Turbo enabled: 0x850089
3) EIST enabled + Turbo disabled: 0x850089 + 0x40 (edx)
4) EIST + Turbo disabled: 0x850089 + 0x40 (edx)

 

1) is fine

2) is a poor BIOS choice, so that people can enable Turbo separately, but in reality, it just leaves EIST on

3) is fine

4) is totally wrong, EIST should be disabled in this case, but it isn't

 

Even if C1M is off, 4) is still not altered.

 

Shouldn't it be possible to set a fixed ratio and then only enable Turbo? Say I want my CPU at 33x at all time and if thermal allows it, go up to 37x.

Link to comment
Share on other sites
elitee
elitee
Posted
Posted
I'm checking AGPM.kext Info.plist...

See that in DHP's AGPM.kext Info.plist there's a part of code regarding iMac12,2 that's not present in mine AGPM.kext Info.plist.

 

patch plist -i new_plist.diff -o plist.new

Link to comment
Share on other sites
flAked
flAked
Posted
Posted 
patch plist -i new_plist.diff -o plist.new

Until recently, RevoBoot used iMac12,3 internally, which was the reason why AGPMEnabler didn't match. This is fixed. We don't need to edit the plist in order to get SpeedStep working. It is only for GPU throttling.

 

I haven't tackled this part yet. But if those QSC values in fact set the GPU frequency I would be careful.

Link to comment
Share on other sites
mrmojorisin17
mrmojorisin17
Posted
Posted
Until recently, RevoBoot used iMac12,3 internally, which was the reason why AGPMEnabler didn't match. This is fixed. We don't need to edit the plist in order to get SpeedStep working. It is only for GPU throttling.

Confirmed ;)

With "old" version of RevoBoot (e4e61b2) is necessary to patch AGPM.kext Info.plist in order to boot succesfully with iMac12,2 mac model identifier.

With new one (6e99411) all works properly also without patching AGPM.kext Info.plist.

 

Just made some other test but sleep doens't work.

I tried to make a reset instead of a shutdown after trying to wake up the PC; result: I get 3 beeps and the PC shutdown.

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