Kaby lake HD 630 problem

[Tiejun]

Assuming you're using ig-platform-id 0x19120000 and that is the only framebuffer patch, and the kext is in cache...
Then I think your problem is something other than the lack of the 0306 connector availability.
Note: You can set config.plist/KernelAndKextPatches/Debug=true to see which patches are being applied and how many times...

I turned on the debug mode, and saw that the generic patch had patched 9 times.

If the patch has effect, should I expect to see 4 frame buffers in ioreg? I've never seen 4, but 3 only.

 

[RehabMan]

I turned on the debug mode, and saw that the generic patch had patched 9 times.

It is the correct number of patches.

If the patch has effect, should I expect to see 4 frame buffers in ioreg? I've never seen 4, but 3 only.

It depends on whether you're using the correct ig-platform-id.
I would need full "Problem Reporting" files.
Or at least ioreg, and config.plist.

 

[pH24]

It is the correct number of patches.



It depends on whether you're using the correct ig-platform-id.
I would need full "Problem Reporting" files.
Or at least ioreg, and config.plist.

Hi, RehabMan. Any info updated with this problem? I currently have the same problem. What's worse, I haven't achieved qe with my HD 630. But I got ci. I did tons of reading about HD630 and experiment a lot with IntelSKLFrameBuffer. Let me explain.

Nearly all tutorials suggest fakeid of 19168086. I got that. It's probably because your FakePCI_Intel_HD has that string in it. But when I use it, the screen would be stuck with a white blank on the top left corner. Picture attached(with verbose boot picture).





It would get stuck here no matter how long you wait. No vnc either.
I am using a 4k display with DP. So just in case it's needed, I did the core display patch already. Now back to the fakeid of 19128086. I have seen one of your posts which said "showing frame buffer means you get ci but lacking of FakePCI means losing qe". As the picture shows, it's got ci. BTW, the display is recognized by the system. Correct Vendor and display id. Even HiDPI is auto-enabled. But just no acceleration.


So after all, what might be the cause of my failure of fakeid of 19168086? I think I almost got there. Just one step left to go. I did lots of research and test but to no avail. Here are the ioreg with 19128086 fakeid and config plist. since faking with 19168086, i can not get to desktop or vnc, so i wouldn't be able to provide ioreg with that. Hope you could help me.

 

[Tiejun]

It is the correct number of patches.



It depends on whether you're using the correct ig-platform-id.
I would need full "Problem Reporting" files.
Or at least ioreg, and config.plist.

Hi RehabMan,

Just to let you know, I recently bought a new monitor with DisplayPort, and then the problem is gone. It seems the driver only works for DisplayPort.

 

[shuhung]

Hi, RehabMan. Any info updated with this problem? I currently have the same problem. What's worse, I haven't achieved qe with my HD 630. But I got ci. I did tons of reading about HD630 and experiment a lot with IntelSKLFrameBuffer. Let me explain.

Nearly all tutorials suggest fakeid of 19168086. I got that. It's probably because your FakePCI_Intel_HD has that string in it. But when I use it, the screen would be stuck with a white blank on the top left corner. Picture attached(with verbose boot picture).


View attachment 251838
View attachment 251837

It would get stuck here no matter how long you wait. No vnc either.
I am using a 4k display with DP. So just in case it's needed, I did the core display patch already. Now back to the fakeid of 19128086. I have seen one of your posts which said "showing frame buffer means you get ci but lacking of FakePCI means losing qe". As the picture shows, it's got ci. BTW, the display is recognized by the system. Correct Vendor and display id. Even HiDPI is auto-enabled. But just no acceleration.
View attachment 251839

So after all, what might be the cause of my failure of fakeid of 19168086? I think I almost got there. Just one step left to go. I did lots of research and test but to no avail. Here are the ioreg with 19128086 fakeid and config plist. since faking with 19168086, i can not get to desktop or vnc, so i wouldn't be able to provide ioreg with that. Hope you could help me.

Condition has something like mine.

Pentium Base HD530 (mine is G4500) will not work with Sierra, will get windowserver XXX issue, can not reach desktop GUI, by inject wrong device-id and AAPL,ig-platform-id with customer ssdt, can reach Sierra desktop but no QE/CI.

It work fine with El Capitan, currently have no chance to run Sierra with QE/CI.

So I guess, it might same as your HD630. As FakePCIID work, it should get same result. As FakePCIID is disabled, can reach desktop GUI but without QE/CI. It should work with El Capitan.

Pentium Base CPU (Skylake/Kabylake) is not supported, XCPM power control will not work under Sierra, speedstep still work with NullCPUPowerManagement.kext applied, need to check HWPEanble from Clover config.plist to get full CPU PM, resulting may be much better.

 

[Loloflatsix]

Condition has something like mine.

Pentium Base CPU (Skylake/Kabylake) is not supported, XCPM power control will not work under Sierra, speedstep still work with NullCPUPowerManagement.kext applied, need to check HWPEanble from Clover config.plist to get full CPU PM, resulting may be much better.

It's Ok : I'm running on my budget buid Celeron G1840 in Sierra 10,12,4 :
- curently use with Ge force GT730
- tested with IGP HD 4400 and work's fine to

Sleep and wake up is OK with NullCPUPowerManagement.kext

I use this hack for teleworking at home and also with Remotdesktop on my job TSE server.

I think with Kaby Lake it may be working to , I notice to have right config.plist and other parameter will be perhaps "difficult and long time training".

 

[shuhung]

It's Ok : I'm running on my budget buid Celeron G1840 in Sierra 10,12,4 :
- curently use with Ge force GT730
- tested with IGP HD 4400 and work's fine to

Sleep and wake up is OK with NullCPUPowerManagement.kext

I use this hack for teleworking at home and also with Remotdesktop on my job TSE server.

I think with Kaby Lake it may be working to , I notice to have right config.plist and other parameter will be perhaps "difficult and long time training".

I also had a Celeron G1840 built, now is fade out.

NullCPUPowerManagement.kext had no problem with Skylake/Kabylake base Pentium CPU, it does not distort speedstep function any more with above CPU, this kext must be applied for those CPU. My advise is, should enable HWP power control with those CPU while XCPM is not work under Sierra. Also, it should better than normal XCPM power control.

 

[pH24]

Condition has something like mine.

Pentium Base HD530 (mine is G4500) will not work with Sierra, will get windowserver XXX issue, can not reach desktop GUI, by inject wrong device-id and AAPL,ig-platform-id with customer ssdt, can reach Sierra desktop but no QE/CI.

It work fine with El Capitan, currently have no chance to run Sierra with QE/CI.

So I guess, it might same as your HD630. As FakePCIID work, it should get same result. As FakePCIID is disabled, can reach desktop GUI but without QE/CI. It should work with El Capitan.

Pentium Base CPU (Skylake/Kabylake) is not supported, XCPM power control will not work under Sierra, speedstep still work with NullCPUPowerManagement.kext applied, need to check HWPEanble from Clover config.plist to get full CPU PM, resulting may be much better.

Oops. So your g4500 used to work all right in El Capitan and now fail in sierra? How do you know the window server issue? I cannot get useful info in the new console in sierra.
I was really shocked to see Pentium based one would have such differences with the core series. The integrated gpu share the same name and device id. Thought they should be identical as far as the graphics part.
Yeah. Pm in sierra is not a big deal . One way or another, you can achieve this. The problem I have with hwp is that it would keep my CPU full speed reported by intel power gadget. Even though it seems not based on temp and the fan speed.
Last, pretty sad to see what sierra brought to the table for pentium based CPUs.

 

[pH24]

Ye

I also had a Celeron G1840 built, now is fade out.

NullCPUPowerManagement.kext had no problem with Skylake/Kabylake base Pentium CPU, it does not distort speedstep function any more with above CPU, this kext must be applied for those CPU. My advise is, should enable HWP power control with those CPU while XCPM is not work under Sierra. Also, it should better than normal XCPM power control.

You can patch kernel for xcpm. Btw it's been a long long time since my last hack. So I guess you could imagine how shocked I am when I find CPU speed stepping with nullcpu.

 

[Loloflatsix]

May be also patch SSDT with ssdtPRGen.sh