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

Apparently AICPUPM loads fine on current P67 Gigabyte mainboards with just a HPET fix, so I requested a ACPI dump over at tonymac and got this file. Maybe it will give us some insight as to how Gigabyte is doing the magic.

The dump is in a strange format, but you can search it alright.

 

I didn't find any mind-bending differences yet.

 

AcpiTbls.txt.zip (P67A-UD3-B3 by ghorwith)

Link to comment
Share on other sites
davidm71
davidm71
Posted
Posted

So I can cut this segment out of the DSDT file?

 

 Method (_PIC, 1, NotSerialized)
{
	If (Arg0)
	{
		Store (0xAA, DBG8)
	}
	Else
	{
		Store (0xAC, DBG8)
	}

	Store (Arg0, PICM)
}

 

Also if I remove the DEB0 part I get errors...

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
Apparently AICPUPM loads fine on current P67 Gigabyte mainboards with just a HPET fix, so I requested a ACPI dump over at tonymac and got this file. Maybe it will give us some insight as to how Gigabyte is doing the magic.

The dump is in a strange format, but you can search it alright.

 

I didn't find any mind-bending differences yet.

 

AcpiTbls.txt.zip (P67A-UD3-B3 by ghorwith)

Ah not again. LOL I mean it's the GB boys again that beats Asus, again. Just my luck. Let's see what I can find. Probably nothing but still. I have to be sure. Fingers crossed.

 

Hey it is late alright but we have holidays and since I can't sleep anyway :D

 

 

So I can cut this segment out of the DSDT file?

 

 Method (_PIC, 1, NotSerialized)
{
	If (Arg0)
	{
		Store (0xAA, DBG8)
	}
	Else
	{
		Store (0xAC, DBG8)
	}

	Store (Arg0, PICM)
}

 

Also if I remove the DEB0 part I get errors...

Yes you can. That is if you remove all callee's as well. Just search for them.

 

Update: Hmm. I see a lot of 64-bit addresses. Something I don't see in our table data! We should get real GB factory ACPI tables to see what is going on.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
Ah not again. LOL I mean it's the GB boys again that beats Asus, again. Just my luck. Let's see what I can find. Probably nothing but still. I have to be sure. Fingers crossed.

They must truly enjoy their hackintoshes :)

 

Also, only AICPUPM and -Client seem to be necessary for SpeedStep to work with the MBP8x kernel, not the other kext mentioned (LPC for example).

 

 

EDIT: I wrote tony about a ACPI dump, he is currently editing a lot of those Gigabyte boards *fingers crossed*

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted

Oh dear. We don't have this do we?

New 100% Hardware VRD 12 Compliant CPU Power Design

 

GIGABYTE 6 series motherboards incorporate an Intel® approved Intersil PWM controller that is VRD 12 (Voltage Regulator Down) compliant. This means that it offers new features that include SerialVID (SVID) which transfers power management information between the processor and voltage regulator controller, allowing more robust and efficient signaling control between the CPU and PWM controller – hence, delivering a more energy efficient platform.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
Oh dear. We don't have this do we?

 

Not to worry! Thats marketing, Intel made those requirements for the new chipset platform.

 

Here is ASUS:

 

Advantages of ASUS DIGI+ VRM Digital Power Design

Unlike previous VRD versions, Intel® VRD12 uses digital signals (SVID). To ensure perfect power delivery, ASUS specially designed DIGI+ VRM to sync completely with this new technology.

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
Not to worry! Thats marketing, Intel made those requirements for the new chipset platform.

Phew. Great.

 

What model identifier do they use? MacBookPro8,n?

I also see that the base address in the MCFG.dsl is different:

0x00000000E0000000 (Asus)

0x00000000F4000000 (GB)

 

Note: Made some changes. Going to perform a/some reboot(s) to see what, if anything, changed.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
What model identifier do they use? MacBookPro8,n?

[url="http://www.insanelymac.com/forum/topic/279450-why-insanelymac-does-not-support-tonymacx86/"]#####[/url] opted for MacBookPro8,1

Link to comment
Share on other sites
davidm71
davidm71
Posted
Posted
Yes you can. That is if you remove all callee's as well. Just search for them.

 

Update: Hmm. I see a lot of 64-bit addresses. Something I don't see in our table data! We should get real GB factory ACPI tables to see what is going on.

 

 

Sorry what does 'callee' refer to? Still learning the basic lingo...

 

Also I found out if I change my smbios.plist file with [url="http://www.insanelymac.com/forum/topic/279450-why-insanelymac-does-not-support-tonymacx86/"]#####[/url] to reflect a Mac Pro 5.1 the system fails to boot. Any ideas for us unenlightened as far as why?

 

Also I hope you get some sleep! Try warm milk or Camomile tea ...

 

Edit: Google is my friend and I thing I understand..

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
Sorry what does 'callee' refer to? Still learning the basic lingo...

Without going too much into programming, it basically means all references of the method or the variable involved.

Look at _PIC, the only thing it does is set a value to DBG8. As you want to delete _PIC you search for references to the method itself and also DBG8.

 

Also I found out if I change my smbios.plist file with [url="http://www.insanelymac.com/forum/topic/279450-why-insanelymac-does-not-support-tonymacx86/"]#####[/url] to reflect a Mac Pro 5.1 the system fails to boot. Any ideas for us unenlightened as far as why?

The definitions have several dependencies, I believe the MacPro5,1 actually tries to load the AppleTyMCEDriver, so you could delete/move that. But it is preferred for now to leave it at MacBookPro8,3.

 

 

EDIT

 

1) The new iMac's are out and it seems they are (mostly) using the S-versions for the i5's:

i5-2400S, i5-2500S, i5-2600S and i5-2400, i7-2600

 

2) Another custom build Install DVD for the new iMac's and no supplementary update as of yet

 

3) Sam, how do you reference the APSS table in the CPU0 scope? For now, I put the whole table under CPU0, as "Alias(APSS,APSS)" wont work.

 

4) Using MacBookPro8,3 in RevoBoot panics on me, 8,1 works fine however. And I always get MBP8,3 in SystemProfiler, but don't know where it's coming from.

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted

Sweet. The cat new iMac's are out of the bag. Should get my hands dirty on one tomorrow afternoon. Just a simple i5 but still. Better that than nothing. Let's see what the kexts tell us :rolleyes:

 

1) The new iMac's are out and it seems they are (mostly) using the S-versions for the i5's:

i5-2400S, i5-2500S, i5-2600S and i5-2400, i7-2600

The S editions are low voltage CPU's aren't they?

 

2) Another custom build Install DVD for the new iMac's and no supplementary update as of yet

I use iMac12,1 as model identifier and got an update today, but is was a stupid font thing only.

 

3) Sam, how do you reference the APSS table in the CPU0 scope? For now, I put the whole table under CPU0, as "Alias(APSS,APSS)" wont work.

I only have Name (APSS... in my CPU0 scope. No need for any aliases anymore (in the other CPU scopes).

 

4) Using MacBookPro8,3 in RevoBoot panics on me, 8,1 works fine however. And I always get MBP8,3 in SystemProfiler, but don't know where it's coming from.

Look in: i386/libsaio/smbios/model_data.h – is hard coded to the latest model.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
The S editions are low voltage CPU's aren't they?

Yup, for example the i5-2500S runs at 2,7GHz with a max turbo of 3,7GHz and a TDP of 65W.

 

I use iMac12,1 as model identifier and got an update today, but is was a stupid font thing only.

Look in: i386/libsaio/smbios/model_data.h – is hard coded to the latest model.

I see, will test that out as well. Do you have smc-huronriver defined in FakeSMC?

 

 

Here is a new minimal SSDT_PR for the i5-2500K:

ssdt_pr_minimal.dsl.zip

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
Yup, for example the i5-2500S runs at 2,7GHz with a max turbo of 3,7GHz and a TDP of 65W.

Ok. Thanks. So this means that the RAPL data will be different.

 

I see, will test that out as well. Do you have smc-huronriver defined in FakeSMC?

Yes. Is part of the FakeSMC Info.plist dgsga attached here earlier.

 

Here is a new minimal SSDT_PR for the i5-2500K:

ssdt_pr_minimal.dsl.zip

Thanks. Will have a look at it.

 

Update: I checked the Chimera source code and noticed that the CPU detection code in smbios_patcher.c is broken for Sandy Bridge CPU's:

case CPU_MODEL_SANDY_BRIDGE: // Intel Core i3, i5, i7 LGA1155 (32nm)
if (strstr(Platform.CPU.BrandString, "Core(tm) i3"))
	return 0x901;		// Core i3
if (strstr(Platform.CPU.BrandString, "Core(tm) i5"))
	return 0x601;		// Core i5
return 0x0701;			// Core i7

Maybe this explains why (haven't check it myself) people can boot with AppleIntelCPUPowerManagement.kext and ditto Client.kext loaded (cpu-type is initialized with the wrong value).

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
Update: I checked the Chimera source code and noticed that the CPU detection code in smbios_patcher.c is broken for Sandy Bridge CPU's

A user at tonymac reported working speedstep _before_ Chimera was actually released. He stated that only MSRTools at 32bit would show a working stepping. I don't trust this entirely to be honest. If SpeedStep was working properly sysctl should show it accordingly (or that tool is buggy, too).

 

As Chimera is only a merge of two Chameleon trunks it could well be that this code is in current Chameleon. But as I was using the latest Chameleon (the one tree from Chimera with SB support) in the last week I think it is still the Gigabyte engineers who crafted some nice ACPI tables.

 

Reference: http://www.tonymacx86.com/viewtopic.php?f=...26&start=20

 

This is in the Chameleon trunk (touched 8 months ago):

case CPU_MODEL_DALES_32NM: // Intel Core i3, i5, i7 LGA1156 (32nm) (Clarkdale, Arrandale)
if (strstr(Platform.CPU.BrandString, "Core(tm) i3"))
return 0x901; // Core i3
if (strstr(Platform.CPU.BrandString, "Core(tm) i5"))
return 0x601; // Core i5
return 0x0701; // Core i7

Copy and paste anyone?

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
A user at tonymac reported working speedstep _before_ Chimera was actually released. He stated that only MSRTools at 32bit would show a working stepping. I don't trust this entirely to be honest. If SpeedStep was working properly sysctl should show it accordingly (or that tool is buggy, too).

Neither do I.

 

As Chimera is only a merge of two Chameleon trunks it could well be that this code is in current Chameleon. But as I was using the latest Chameleon in the last week I think it is still the Gigabyte engineers who crafted some nice ACPI tables.

I'm actually quite disappointed in Chimera. Not much of an improvement. Hoped for more but that is, apparently, a delusion on my side. Must be the GB engineers. Still waiting for the ACPI tables...

 

Thanks. Oh I do love this kind of comments:

"CPU: Attempting to enable IDA: Failed!" --> bdmesg
"kxld[com.apple.kpi.unsupported]: This symbol set has the following unresolved symbols: _lapic start" --> Kernel log

Seems this person is using the 10.7.3 kernel but not the system.kext for it. So is that the trick here?

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
I'm actually quite disappointed in Chimera. Not much of an improvement. Hoped for more but that is, apparently, a delusion on my side. Must be the GB engineers. Still waiting for the ACPI tables...

Maybe I should find somebody here with the knowledge to do this... I asked Tony directly, but he just said that he doesn't think a DSDT will do the job. I'm starting to think this whole venture is a hackish one-shot, pardon my french.

 

Seems this person is using the 10.7.3 kernel but not the system.kext for it. So is that the trick here?

Tonymacs BridgeHelper only copies the mach_kernel and as of the recent update the two AppleIntelCPU kext. So yes, he is using 10.7.0 system.kext instead of 10.7.3

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted

There is one other thing that is bothering me, and that is that all boot loaders report the normal speed at boot time. Not the max turbo speed. I wonder if this means that we can boot faster. That is if we can figure out how to trigger this max turbo mode at boot time (since there's only one CPU core active). Meaning an extra 400 MHz boost for my 2600K. That should help.

 

Oh and the new iMac's come with OS X 10.6.6 (Build 10J4026).

Link to comment
Share on other sites
Ed_Saxman
Ed_Saxman
Posted
Posted

Turbo Boost worked differently over old i5/i7 S CPUs.

The i5-750S compared to the i5-750; Even with the same nominal Turbo clockspeed, the S variant will not Turbo as often because it activates with only 1 or 2 cores enabled compared to 1/2/3/4 on the normal 750.

 

It's reasonable to expect this to carry over to SB.

 

Please read: http://www.tomshardware.com/reviews/core-i...wer,2557-2.html

Link to comment
Share on other sites
cyp75
cyp75
Posted
Posted

Just a quick question as I will buy a sandy bridge system in the next couple of days.

 

It seems to me you implied that Gigabyte boards are easier to work with and more compatible with OsX and generating good DSDT tables. Is that true ?

I have read it all, but it gets pretty technical in the end. What is working and what is to be done at this point ?

 

Thanks ! :(

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
It's reasonable to expect this to carry over to SB.

Yeah but it should be no problem to read the registers and apply Turbo accordingly. Besides, I don't think many hackintoshers will have a S-version :(

 

 

User ghorwith agreed to to a full ACPI dump in the next 2 days, stay tuned.

 

 

It seems to me you implied that Gigabyte boards are easier to work with and more compatible with OsX and generating good DSDT tables. Is that true ?

Yes, they are doing a pretty good job over at Gigabyte. That was the case for the last platform as well. You could easily get the basic functionality up and running with a ASUS board as well, though.

 

Power management, sleep and bluetooth aren't working on the P8P67 at the moment.

 

If you want a more convenient build with complete how-tos and ready made DSDT, go for a Gigabyte.

 

There is one other thing that is bothering me, and that is that all boot loaders report the normal speed at boot time. Not the max turbo speed. I wonder if this means that we can boot faster. That is if we can figure out how to trigger this max turbo mode at boot time (since there's only one CPU core active). Meaning an extra 400 MHz boost for my 2600K. That should help.

If we see P0 with +1 Mhz it could very well be a display discrepancy. TurboMode should be active if CPU0 is in C0 (and it's on in the BIOS of course).

 

See http://download.intel.com/design/processor...nots/320354.pdf

In this whitepaper: a algorithm to get the correct current frequency.

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
User ghorwith agreed to to a full ACPI dump in the next 2 days, stay tuned.

Fabulous news.

 

Power management, sleep and bluetooth aren't working on the P8P67 at the moment.

Sleep is PM related. Works here with SleepEnabler.kext and Bluetooth is working fine with one additional kext (someone used a Linux file to incorporate it in a kext and it had been confirmed to work. Did I miss anything on this?

 

If you want a more convenient build with complete how-tos and ready made DSDT, go for a Gigabyte.

I first want to see some evidence before I believe that PM works on a GB. So far the reports are either incomplete, inconclusive or simply false.

 

If we see P0 with +1 Mhz it could very well be a display discrepancy. TurboMode should be active if CPU0 is in C0 (and it's on in the BIOS of course).

No. The frequency reported by MSR shows me that we are not doing a great job. It's also to note that the same CPU on a iMac runs much faster. This can only be due to the PM activation. I did notice that my memory scores, the defaults, were already better so this is promising.

 

See http://download.intel.com/design/processor...nots/320354.pdf

In this whitepaper: a algorithm to get the correct current frequency.

Thanks. I have this as well. Gave me some pointers as to what is going on. And I now hope to have found a way to switch on the after burner on our Asus boards (still work in progress).

 

p.s. My next board will be a Deluxe edition – with Bluetooth – because I do happen to use a Magic Mouse / Trackpad and keyboard combo. All Bluetooth of course :unsure:

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
Sleep is PM related. Works here with SleepEnabler.kext and Bluetooth is working fine with one additional kext (someone used a Linux file to incorporate it in a kext and it had been confirmed to work. Did I miss anything on this?

I'm only aware of this: http://www.tonymacx86.com/viewtopic.php?f=...410&start=0

Using a LaunchDaemon to upload the firmware. Could you link the kext method?

 

p.s. My next board will be a Deluxe edition – with Bluetooth – because I do happen to use a Magic Mouse / Trackpad and keyboard combo. All Bluetooth of course :unsure:

The Pro and Deluxe versions are supporting Nvidia SLI, other than that all P8P67 are pretty identical, especially regarding Bluetooth. But they do have the nicer Intel Gigabit Ethernet.

 

 

Oh and the new iMac's come with OS X 10.6.6 (Build 10J4026).

Interesting. So either we will get a special made 10.6.7 soon or a unified 10.6.8 a little later, I would guess.

Link to comment
Share on other sites
Time2Retire
Time2Retire
Posted
  • Author
Posted
I'm only aware of this: http://www.tonymacx86.com/viewtopic.php?f=...410&start=0

Using a LaunchDaemon to upload the firmware. Could you link the kext method?

I will when I find it again. It was the reason for me to order a new board. Out of stock for now. Have to wait a little longer.

 

The Pro and Deluxe versions are supporting Nvidia SLI, other than that all P8P67 are pretty identical, especially regarding Bluetooth. But they do have the nicer Intel Gigabit Ethernet.

Yeah but I do not have Bluetooth on my P8P67M-PRO board and that is why I want to upgrade.

 

Interesting. So either we will get a special made 10.6.7 soon or a unified 10.6.8 a little later, I would guess.

Received an e-mail telling that the iMac order is pushed back to Friday. What a bummer. I hate waiting.

 

BTW: GB boards come with SMBIOS 2.4 (Asus with SMBIOS 2.6) and ACPI 1.0b (Asus with 2.0a) so I definitely pick Asus over GB.

Link to comment
Share on other sites
flAked
flAked
Posted
Posted
Received an e-mail telling that the iMac order is pushed back to Friday. What a bummer. I hate waiting.

Let the speculation begin! QA didn't approve the 10.6.7 iMac12 in time, so they delayed shipment und will bring us the update on friday :(

 

Btw, Tony found some nice Geekbench results for the new iMac's with processor info: http://www.tonymacx86.com/viewtopic.php?f=...965&start=0

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