13 replies to this topic

[martisti]

I,m planning to buy i5 3470 which has Intel HD 2500 integrated. Will this work with QE/CI with ML? Somebody can help? I tried to search but didn't find anything. Thanks

[hellozx]

I'm sorry. It can't work now.
i5 3570K with hd4000 can work well.

[Zenith432]

I've been able to get HD 2500 working on ML for framebuffer only - no QE/CI.

The framebuffer kext is AppleIntelFramebufferCapri.kext (same as HD4000.)
The QE/CI driver is AppleIntelHD4000Graphics.kext.

Did as follows

• Remove AppleIntelSNBGraphicsFB.kext from /S/L/E so it doesn't collide with Capri.
• Edit Info.plist in AppleIntelFramebufferCapri and changed the personality for the Intel MEI driver to match on 1C3A instead of 1E3A. Note that this chip is on the motherboard, not the CPU. The MEI for Series 6 chipset (which is what I have) has device-id 1C3A. Its driver is normally in AppleIntelSNBGraphicsFB. The MEI for Series 7 chipset has device-id 1E3A. Its driver is in Capri, and the Capri driver can handle both chipsets!
• Injected "AAPL,ig-platform-id" as explained in the HD4000 thread using Chameleon boot-loader. All values for this field worked for me (except the ones with no-DVI/no-display).
• Removed AppleIntelHD4000.kext from /S/L/E. This is necessary, as the QE/CI driver doesn't work, and if you leave it in, the system will hang during boot and system log will be flooded with error messages from it.

After this, I have framebuffer driver which allows resolution changes, hardware mouse cursor. Also seems to work a bit faster than the vanilla framebuffer driver (IONDRVFramebuffer.)

There's no QE/CI as it hangs

[oliveiro]

Just to report a successful install of a HD2500 mobile (Ivy Bridge Pentium 2020M laptop) WITH QE/CI.

You have to update to a working 10.8.3 (on my laptop, it requires ACPIPlatform 10.8.2 rollback, and I believe PCIfamily too. Then you have to use Chameleon r2187 or at least r2179.

My 2500 has HDMI/VGA outputs and internal display. HDMI was working almost OOTB as soon as I booted with GraphicsEnabler Yes. FullHD. But the internal display was not. I noticed that in IoRegistry, under GFX tree, Ig-Platform-Id was locked to 01660009, whatever parameter I tried to inject through Chameleon's IGPlatformId key.

This is due to the fact that Chameleon uses injection of that parameter since version 2177, but from version 2179 on, the injection is done automatically for HD2500 and HD4000 both mobile and desktop versions, based on the Graphics Memory. 32 64 or 96Mo. And for my board, the 64 M config was automatically setting ig-platform-id to 01660009.

From 2179 version on, the igplatformid key is not used.

I tried to patch and compile 2187 version of chameleon, to re-enable the manual injection option (source taken from version 2177), and to enable it for both 2500 and 2500 mobile. I was able to change the igplatformid to 3 (01660003) and it was a succes. Now the internal display works. If I want dual monitor, I have to boot with the HDMI monitor unplugged, then plug it once OSX is launched, otherwise OSX seems to see the internal device, but it doesn't turn on... Instead of patching Chameleon, you can use two other methods : a dsdt.aml including ig-platform-id patch (but it creates problems with ivy bridge native power management if you want it) or deviceproperties injection in boot.plist

That post in chinese (http://bbs.pcbeta.co...297288-1-1.html) gave me the idea to the solution. If anyone reads or writes chinese, say thanks to that guy for me !

Edited by oliveiro, 23 March 2013 - 12:44 PM.

[theconnactic]

Well, now that's a breakthrough discovery: first time i hear about a totally unsupported card (no kexts for its chipset family, meaning no way to simply inject an ID to a plist to get it working) made to work.

Congratulations, but before crying victory i'd wait for someone to reproduce this success.

All the best!

[oliveiro]

Well technically, there are kexts for Ivy Chipset and Intel HD Graphics 2500 (they already include PCI primary match) the point was just to inject the correct info at boot, so these drivers got loaded properly. X86platform FramebufferCapri and HD4000 can be loaded thanks to methods I found around here, in forums. So thanks to them. And also, the chinese post was the first success report I found. So thanks to the guy too.

[gpatxxx]

Could this work for the Intel HD2000 as well? The one on the Celeron 847 Mobile CPU. I need it for an ASUS notebook (x201e) but could also help for an Intel NUC which has this CPU has well.

[Zenith432]

Just to report a successful install of a HD2500 mobile (Ivy Bridge Pentium 2020M laptop) WITH QE/CI.

How did you get QE/CI working? I have the HD2500. Don't need chameleon, as I patched DSDT as follows

		    Device (MEI)
		    {
			    Name (_ADR, 0x00160000)  // _ADR: Address
			    Method (_DSM, 4, NotSerialized)
			    {
				    Store (Package (2)
				    {
					    "device-id",
					    Buffer (4)
					    {
						    0x3a, 0x1e, 0x00, 0x00
					    }
				    }, Local0)
				    DTGP (Arg0, Arg1, Arg2, Arg3, RefOf (Local0))
				    Return (Local0)
				 }
		    }
	    Device (GFX0)
	    {
				 .... <original stuff omitted>
		    Method (_DSM, 4, NotSerialized)
		    {
			    Store (Package (2)
			    {
				    "AAPL,ig-platform-id",
				    Buffer (4)
				    {
					    0x09, 0x00, 0x66, 0x01
				    }
			    }, Local0)
			    DTGP (Arg0, Arg1, Arg2, Arg3, RefOf (Local0))
			    Return (Local0)
		    }
	 }

The kexts are
AppleIntelFramebufferCapri.kext - framebuffer driver.
AppleIntelHD4000Graphics.kext - QE/CI driver.
The QE/CI driver doesn't work. System hangs on boot stuck in gray apple screen and lots of msgs in syslog.
When I remove the QE/CI driver, system works with framebuffer only.
This is on 10.8.3.

[oliveiro]

Gpatxxx, I'm sorry, but hd2000 and hd2500 are different (different chipsets, cpu & graphics). Try to google "hd 2000", I remember seeing topics about workarounds and modded kexts.

Zenith432, except bios video set to 64 Mo, graphicsEnabler= Yes in boot.plist and the ig-platform-id parameter set to 3, I din't have to do anything special ( I'm using a vanilla 10.8.3 (only intelCPUpowermanagement is patched)). What is your video Ram in bios, 64 ? if it's the case, you can try 3,8, or 9 for platform-Id. Since your dsdt edit uses 9, you can try 3 or 8 and see what happens. Else, is this a laptop you're using ? have you tried hdmi/dvi outputs ? does it still hang on apple logo ?

[artur-pt]

hello

one question

if u remove the qe/ci driver to boot... u have qe/ci ?

good hack

[oliveiro]

@artur : I don't understand what you mean. Remove the qe/ci driver ? Do you mean the GraphicsEnabler injection in boot.plist ? or the HD4000 driver in SLE ? I didn't test because I believe the purpose of GraphicsEnabler with the HD4000 driver is precisely to have HD2500 or HD4000 correctly identified and configured, and enabled QE-CI.

[Lacedaemon]

Just to report a successful install of a HD2500 mobile (Ivy Bridge Pentium 2020M laptop) WITH QE/CI.

You have to update to a working 10.8.3 (on my laptop, it requires ACPIPlatform 10.8.2 rollback, and I believe PCIfamily too. Then you have to use Chameleon r2187 or at least r2179.

My 2500 has HDMI/VGA outputs and internal display. HDMI was working almost OOTB as soon as I booted with GraphicsEnabler Yes. FullHD. But the internal display was not. I noticed that in IoRegistry, under GFX tree, Ig-Platform-Id was locked to 01660009, whatever parameter I tried to inject through Chameleon's IGPlatformId key.

This is due to the fact that Chameleon uses injection of that parameter since version 2177, but from version 2179 on, the injection is done automatically for HD2500 and HD4000 both mobile and desktop versions, based on the Graphics Memory. 32 64 or 96Mo. And for my board, the 64 M config was automatically setting ig-platform-id to 01660009.

From 2179 version on, the igplatformid key is not used.

I tried to patch and compile 2187 version of chameleon, to re-enable the manual injection option (source taken from version 2177), and to enable it for both 2500 and 2500 mobile. I was able to change the igplatformid to 3 (01660003) and it was a succes. Now the internal display works. If I want dual monitor, I have to boot with the HDMI monitor unplugged, then plug it once OSX is launched, otherwise OSX seems to see the internal device, but it doesn't turn on... Instead of patching Chameleon, you can use two other methods : a dsdt.aml including ig-platform-id patch (but it creates problems with ivy bridge native power management if you want it) or deviceproperties injection in boot.plist

That post in chinese (http://bbs.pcbeta.co...297288-1-1.html) gave me the idea to the solution. If anyone reads or writes chinese, say thanks to that guy for me !

Did you delete the IntelHD4000.kext, like Zenith 432?

I've tried 3, 8, and 9 in the DSDT edit, still nothing.

[oliveiro]

Hi Lacedaemon, Zenith's method was pre-10.8.3.
Now there are working drivers and it's a combination of framebufferCapri.kext for framebuffer, AND hd4000.kext which is the driver for QE/CI for both hd2500 and hd4000, so keep it in SLE.

[oliveiro]

@Zenith

Did you solve your problem since april 1st ?
If not, check that you included DTGP method in your dsdt, since your dsdt edit includes a call to it.
Otherwise doublecheck in ioregistry in PCI0/GFX0 category, you will find the aapl igplatformid value, see there if your dsdt value is taken into account;
hope it helps