DSDT - Vanilla Speedstep - Generic Scope (_PR)

[kdawg]

That's the default value for Vcore in bios, you can also see it using Core Temp on Windows.

Sorry I though thought you said VCore for the chip, so you mean the default VCore value for the motherboard.

[keeza]

After running a quick initial test using Keeza's DSDT patch, the voltage now is locked at maximum while my P states work normally. I need to run a few more tests to determine what happened. Using the CST from my earlier post C and P states seemed to work well even when overclocked.

 

Thats odd, considering we have almost identical board and processor.

You resolved it by setting voltage to normal in BIOS?

 

I've chucked out my processor box so have to run the intel utility in windows to identify the processor model but I have C1 C2 and C4 in BIOS.

 

@all,

I've removed my legacy kext - don't want to create confusion with it.

It doesn't appear to do anything - I'll run some more experiments when I get a chance but till then...

[kdawg]

Thats odd, considering we have almost identical board and processor.

You resolved it by setting voltage to normal in BIOS?

 

I've chucked out my processor box so have to run the intel utility in windows to identify the processor model but I have C1 C2 and C4 in BIOS.

 

@all,

I've removed my legacy kext - don't want to create confusion with it.

It doesn't appear to do anything - I'll run some more experiments when I get a chance but till then...

Changing VCore to "Normal" has fixed that problem.

[mm67]

Sorry I though thought you said VCore for the chip, so you mean the default VCore value for the motherboard.

 

That's the default value for cpu, not for motherboard. Put in a different cpu and you see that the value changes.

[FKA]

Oh my friend! My italy played as a team of amateurs. Chiellini is the defensor more poor that i have seen ever play, in absolute. And not only him. I was very disappointed after 2006 and I still have not completely disposed of anger. By the way... uffff

Enough for me that the world will not be defeated from France (already out, i'm enjoying, i hate France), the Germany ( I hate Germany--grosso, del piero-2006), and Brazil (a world cup than Italy): footballing talking, obviously.

 

This is it...

 

Common Capello!

 

 

what can i say!

thanks for the PM BTW it did make me laugh!

 

Anyway NO MORE FOOTBALL TALK!

[anibalin]

you are using rc4-aserebln-1.1.9 bootloader -

 

I presume you are taking this into account :-

 

• Filenames configured in com.apple.Boot.plist (DSDT=..., SMBIOS=...) must be given with the full path. The Booter does not check automatically the Extra Folder. So if you used DSDT=mydsdt.aml and the mydsdt.aml file is in the Extra folder, then you must use now DSDT=/Extra/mydsdt.aml.
  

D

 

 

 

simply loading the DSDT will allow you to boot and update without nullcpu kext.

 

As far as the others go, I'm not sure but would definatly check you cleared caches since you removed them.

 

snow lepoard cache cleaner is very handy tool!

 

D

 

Ufff, same stuff. I tried anval bootloader too but still.

Took away the ss stuff from the dsdt. Going bananas mate. Bananas.

 

One thing to notice, though, is that the BSD points:

BSD process name corresponding to current thread: kernel_task

(not unknown as before).

[mm67]

Today I got tired of guessing what C-states do to power usage so I went to shop and bought a cheap power meter. I just finished a short test sequence of running Windows with different C-states enabled at different speeds, power usage on my system looks like this:

GHz	C-states off	C1	C2	C4
3,91	143 W	       133 W	130 W	130 W
3,61	141 W	       131 W	129 W	129 W
2,83	140 W	       130 W	129 W	128 W

 

Next I'm going to check if get similar results on OS X using all kind of method CST's.

 

edit.

First results from OS X, all tested with system running at 3.91 GHz:

CST methods with FFixedHW addressing, MSI style

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x5)
		{
			0x4, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x020,0x3,)},0x3,0x11,0xFA}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x39,0x64} 
		})
	}
132 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x4)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x39,0x64} 
		})
	}
131 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x4)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x3,0x39,0x64} 
		})
	}
128 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x4)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x020,0x3,)},0x3,0x11,0xFA}
		})
	}
128 W



	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x3)
		{
			0x2, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4} 
		})
	}

128 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x1)
		{
			0x1, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8} 
		})
	}
130 W

CST methods with SystemIO addressing, Gigabyte style

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x05)
		{
			0x4, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x414,,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x415,,)},0x3,0x55,0xFA}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x416,,)},0x4,0x96,0x64} 
		})
	}
130 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x04)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x414,,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x416,,)},0x3,0x96,0x64} 
		})
	}
130 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x04)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x414,,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x415,,)},0x3,0x55,0xFA}
		})
	}
130 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x03)
		{
			0x2, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (SystemIO,0x08,0x00,0x414,,)},0x2,One,0x1F4}
		})
	}
130 W


Non-functional CST + working P-states
151 W

No powermanagement
180 W

 

Couple of things seem to be clear. It's a real bad idea to run OS X without working power management, using a stock cooler in a warm room even without overclocking will push temps real high, difference between fully working power management and no power management at all is a whopping 50 W, that's some workload for stock cooler.

 

Other thing that looks clear is that CST methods using Gigabyte style addressing do not work with OS X, all such methods use same 130 W as simply running only C1E.

[keeza]

Today I got tired of guessing what C-states do to power usage so I went to shop and bought a cheap power meter. I just finished a short test sequence of running Windows with different C-states enabled at different speeds, power usage on my system looks like this:

 

GHz	C-states off	C1	C2	C4
3,91	143 W	       133 W	130 W	130 W
3,61	141 W	       131 W	129 W	129 W
2,83	140 W	       130 W	129 W	128 W

 

 		Method (CST, 0, NotSerialized)
		{
			Return (Package (0x4)
			{
				0x3, 
				Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
				Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
				Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x39,0x64} 
			})
		}
131 W

		Method (CST, 0, NotSerialized)
		{
			Return (Package (0x4)
			{
				0x3, 
				Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
				Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
				Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x3,0x39,0x64} 
			})
		}
128 W

 

This result surprised me. By defining C4 with 3 actually results in lower wattage. If I set the CState number to 3 (and not 4 as I currently have it set at) then byte 7 doesn't get set. (80). Could this have to do with you using Method CST? (I'm using Name CST)

 

Other thing that looks clear is that CST methods using Gigabyte style addressing do not work with OS X, all such methods use same 130 W as simply running only C1E.

So Gigabyte board users need to dump SysytemIO addresses in defining CStates yes?

 

Would it be too difficult to conduct the same test at stock settings i.e. no overclocking?

 

Good work mm67. Nice approach...

Out of interest how is the power metre connected to the computer? And was your computer at idle during the test?

[mm67]

This result surprised me. By defining C4 with 3 actually results in lower wattage. If I set the CState number to 3 (and not 4 as I currently have it set at) then byte 7 doesn't get set. (80). Could this have to do with you using Method CST? (I'm using Name CST)

 

 

So Gigabyte board users need to dump SysytemIO addresses in defining CStates yes?

 

Would it be too difficult to conduct the same test at stock settings i.e. no overclocking?

 

Good work mm67. Nice approach...

Out of interest how is the power metre connected to the computer? And was your computer at idle during the test?

 

Just finished tests running at stock speed, no surprises.

		Method (CST, 0, NotSerialized)
	{
		Return (Package (0x5)
		{
			0x4, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x020,0x3,)},0x3,0x11,0xFA}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x39,0x64} 
		})
	}
130 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x4)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x39,0x64} 
		})
	}
130 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x4)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x3,0x39,0x64} 
		})
	}
127 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x4)
		{
			0x3, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x020,0x3,)},0x3,0x11,0xFA}
		})
	}
127 W



	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x3)
		{
			0x2, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4} 
		})
	}

127 W

	Method (CST, 0, NotSerialized)
	{
		Return (Package (0x2)
		{
			0x1, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8} 
		})
	}
129 W

 

I'm still not sure about C4, I know that on my system Bluetooth module and Wlan card wake up the system from C4 all the time, on Windows cpu spends about 50 % of idle time in C2 and 50 % in C4. On OS X situation maybe even worse and I don't know what constant C-state switching does to power usage. I'll check later what happens without wireless stuff interfering with idle states. Maybe C4 will use less power that way, on Windows there was hardly any difference between C2 and C4.

 

Using Gigabyte SystemIO addresses doesn't seem to do any harm, it just doesn't do anything, you won't gain any power savings. Might just as well use only C1E. I did try one of those Gigabyte methods also on stock settings, same result as on overclocked settings. Exactly same power usage as using only C1E.

 

I bought a unit like this : http://cn-yidong.com/en/openwin.php?id=57 . It's probably not very accurate if you are trying to see exact power usage of some appliance but for this kind of testing purpose it should be fine. I'm not so interested in knowing the exact power usage, knowing the difference between different settings is enough and for that use the meter is surely good enough.

 

I tried do all tests same way, boot up system, close all programs I can (Dropbox, Pleasesleep etc.), wait 5 minutes without touching anything and then check the reading from meter.

[kdawg]

Thats odd, considering we have almost identical board and processor.

You resolved it by setting voltage to normal in BIOS?

 

I've chucked out my processor box so have to run the intel utility in windows to identify the processor model but I have C1 C2 and C4 in BIOS.

 

@all,

I've removed my legacy kext - don't want to create confusion with it.

It doesn't appear to do anything - I'll run some more experiments when I get a chance but till then...

If I set my vcore to "normal" per mm67's suggestion I now see voltage fluctuate again with core temperatures lower than I've ever seen them. My stock vcore for my processor is 1.225. I've been running the FSB at 460 but of course I have stability problems 1.225v surely isn't enough juice for the processor to run at 3.91GHz. Running mprime eventually results in a KP but I'm not sure if it's playing with C states that caused the KP. Lowering the FSB to seems to be better but I have a lot of work ahead of me to see what I can do about getting it stable again. I think perhaps my RAM is getting flaky because a startup from a shutdown or KP results in an endless boot loop.

[mm67]

If I set my vcore to "normal" per mm67's suggestion I now see voltage fluctuate again with core temperatures lower than I've ever seen them. My stock vcore for my processor is 1.225. I've been running the FSB at 460 but of course I have stability problems 1.225v surely isn't enough juice for the processor to run at 3.91GHz. Running mprime eventually results in a KP but I'm not sure if it's playing with C states that caused the KP. Lowering the FSB to seems to be better but I have a lot of work ahead of me to see what I can do about getting it stable again. I think perhaps my RAM is getting flaky because a startup from a shutdown or KP results in an endless boot loop.

 

You have to find the max FSB for your cpu, each chip is individual. You probably start with something like 400 and do some stress testing. If that works go to 410 and repeat and so on. That's how i figured out max FSB values for stock Vcore. My chip can do 460 with LLC enabled and 430 with LLC disabled. And I know that the 460 settings I'm using at the moment are pretty stable, I ran 24+ h of Prime95 small FFT's, large FFT's and blend test. Before those Prime95 runs I let system run 50 passes of Linpack with max memory settings and after all that couple of hours of OCCT PSU test. That by the way seems to be hardest test I have ever seen, basically it is running Linpack and Furmark tests at same time. Of course I had to test my new power meter with that test, power usage was 386 W And I only have a GTS 250.

 

edit. Had to test 4.1 GHz also, 430 W. No wonder my old 460 W PSU was starting to feel too small.

[msm]

mm67, kdawg, keeza and anyone overclocking: Are you adjusting your dsdt values (fid, vid) for each change in overclocking? It seems the story on c-states is still evolving.

 

I also recently got a q9550. The first chip I got had a Vcore of 1.2125, but 2 of the cores never changed temp on Prime 95 and it didn't seem to overclock well (bad sensors perhaps). The replacement from Intel has a Vcore of 1.270 and can run 3.8GHz on air without adjusting the Vcore at all. So, not sure stock Vcore is as important to a stable overclock as other factors.

 

Thanks for the great guidance!

msm

[mm67]

mm67, kdawg, keeza and anyone overclocking: Are you adjusting your dsdt values (fid, vid) for each change in overclocking? It seems the story on c-states is still evolving.

 

I also recently got a q9550. The first chip I got had a Vcore of 1.2125, but 2 of the cores never changed temp on Prime 95 and it didn't seem to overclock well (bad sensors perhaps). The replacement from Intel has a Vcore of 1.270 and can run 3.8GHz on air without adjusting the Vcore at all. So, not sure stock Vcore is as important to a stable overclock as other factors.

 

Thanks for the great guidance!

msm

 

Using stock Vcore is by no means not necessary for stable overclock, it's just one more stupid Gigabyte bug that C-states don't work if you use a manually set Vcore value in bios.

[keeza]

Just finished tests running at stock speed, no surprises.

 

Shouldn't C4 be:

(FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x96,0x64}

for Gigabyte boards?

 

mm67, kdawg, keeza and anyone overclocking: Are you adjusting your dsdt values (fid, vid) for each change in overclocking? It seems the story on c-states is still evolving.

 

Sorry but I'm not overclocking.....yet.

[mm67]

Shouldn't C4 be:

 

for Gigabyte boards?

 

FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x96,0x64}

 

Sorry but I'm not overclocking.....yet.

 

That 0x39 came from MSI table, but there is no difference using 0x39 or 0x96, same power usage on both.

 

Just tried what happens when I remove Wlan board and Bluetooth module, on OS X nothing happened, results with different kind of C4 definitions are exactly same. But on Windows side power usage went down to 128 W from original 130 W, using Performance Monitor I can see that system now uses 70 % of idle time in C4 and 30 % in C2, original levels were about 45 % and 55 %.

 

Looks like there is still something wrong with our C4 definition and OS X, then again difference between C2E and C4E seems to be only 2 W even on Windows, that's probably 1-2 degrees difference in core temperatures if one is using stock cooler. With good aftermarket cooler I can't see any difference in temps using C2E or C4E.

 

edit.

Just did some testing with Gigabyte DES and that thing really does work, but only on stock speed. On stock speed power usage was reduced by about 4 W with DES enabled, on overclocked settings there was no changes in power usage.

[keeza]

VoodooMonitor only shows the VID, on iStat I see real Vcore changing like this, idle:

 

On full load:

 

Are you using another version of fakesmc to get fan speeds and cpu voltage in Istat?

I'm presently using fakesmc 2.5

[mm67]

Are you using another version of fakesmc to get fan speeds and cpu voltage in Istat?

I'm presently using fakesmc 2.5

 

I'm using this version : http://www.projectosx.com/forum/index.php?showtopic=1206

 

I just checked on my E7400 system and this CST:

Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8},

Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4},

Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x96,0x64}

 

is not good. Power usage with that definition is higher also on that system.

 

edit.

This definition gives best results on both systems:

		Method (CST, 0, NotSerialized)
	{
		Return (Package (0x3)
		{
			0x2, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4} 
		})
	}

[keeza]

I'm using this version : http://www.projectosx.com/forum/index.php?showtopic=1206

 

I just checked on my E7400 system and this CST:

Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8},

Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4},

Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x030,0x3,)},0x4,0x96,0x64}

 

is not good. Power usage with that definition is higher also on that system.

 

edit.

This definition gives best results on both systems:

		Method (CST, 0, NotSerialized)
	{
		Return (Package (0x3)
		{
			0x2, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x000,0x0,)},One,One,0x3E8}, 
			Package (0x4){ResourceTemplate (){Register (FFixedHW,0x01,0x02,0x010,0x1,)},0x2,One,0x1F4} 
		})
	}

 

Thanks.

 

So to sum up...

C1 & C2 entries only draws less power than C1-C4 entries?

[mm67]

Thanks.

 

So to sum up...

C1 & C2 entries only draws less power than C1-C4 entries?

 

That's how it seems to work, couple of definitions with C3 or C4 used same power but those probably just don't do nothing.

[keeza]

Tested the new Fakesmc kext.

Installed with SMCITEController.kext.

Broke C-States but all temps and fans seem to be OK (except GPU diode-117 degrees!)

 

tried with FakeSMCSuperIO, CStates are back GPU diode still hot. It appears my graphics card isn't supported yet.

 

Core voltage changes between 1.04 - 1.2 volts, so kinda nice to see some evidence of C-States working!

[anibalin]

Ufff, same stuff. I tried anval bootloader too but still.

Took away the ss stuff from the dsdt. Going bananas mate. Bananas.

 

One thing to notice, though, is that the BSD points:

BSD process name corresponding to current thread: kernel_task

(not unknown as before).

edit: solved. update bios 'till 144 build. anything newer => kp. Thanks spíder for the hint. This was slowly getting into my white whale.

[mm67]

Tested the new Fakesmc kext.

Installed with SMCITEController.kext.

Broke C-States but all temps and fans seem to be OK (except GPU diode-117 degrees!)

 

tried with FakeSMCSuperIO, CStates are back GPU diode still hot. It appears my graphics card isn't supported yet.

 

Core voltage changes between 1.04 - 1.2 volts, so kinda nice to see some evidence of C-States working!

 

You never used mark-i ? You could have seen C-states working long time ago with that one :mellow:

[FKA]

Tested the new Fakesmc kext.

Installed with SMCITEController.kext.

Broke C-States but all temps and fans seem to be OK (except GPU diode-117 degrees!)

 

tried with FakeSMCSuperIO, CStates are back GPU diode still hot. It appears my graphics card isn't supported yet.

 

Core voltage changes between 1.04 - 1.2 volts, so kinda nice to see some evidence of C-States working!

 

Are you using LegacyAGPM kext? does the GT 250 have support in AGPM yet?

working at the mo so can't check.

 

http://www.projectosx.com/forum/index.php?showtopic=1024

 

D

[tinush]

@ mm67

 

Hi there mm67, can i ask a question?

My previous Gigabyte board died and got a asus p5q-em

This board with a e8500 gives vanilla speedstep, but i'm not sure if all is OK (temps at idle 53c)

Is there anything i can change to get the temps down (on windows it idles at 34c)

 

Thnx

T

P5Q_EM_reg.zip

dsdt.dsl_tinush_p5q_em.zip

[mm67]

@ mm67

 

Hi there mm67, can i ask a question?

My previous Gigabyte board died and got a asus p5q-em

This board with a e8500 gives vanilla speedstep, but i'm not sure if all is OK (temps at idle 53c)

Is there anything i can change to get the temps down (on windows it idles at 34c)

 

Thnx

T

 

It looks like you don't have C-states, add C1E and your temps should be same as on Windows.