Contribute
Register

Graphics blues, black screen after verbose boot stage/before login.

Status
Not open for further replies.
Joined
Mar 6, 2016
Messages
13
Motherboard
Clevo p750dm-g (z170)
CPU
6700k
Graphics
970m
Mac
  1. 0
Classic Mac
  1. 0
Mobile Phone
  1. 0
Hi guys, I am having an issue with getting my graphics working well. I have a clevo p750dm-g, i7-6700k, 32gb (2x16), nvidia 970m, 4k screen @60 on a 950 m.2. I am able to boot in and have most things working except graphics. If I boot with nv_disable I can see the screen but it has flashing areas (it seems to be where any animations are being drawn). I have installed the nvidia web drivers (...05f02) and when I boot with nvda_drv=1 the screen sleeps after the verbose boot phase is done. I have tried using inject EDID and customEDID but it seems like the os is not understanding my display properties...

I have attached my EFI (which includes a boot log and F4 output) a dump of my EDID and ACPI from linux and the output from IOReg. Please let me know if you have any ideas -- banging my head against the wall.

View attachment p750dm-g.zip

Edited to add nv_disable flashing areas are more like .5fps refreshes for portions of the screen that are updating..
 
Looks like it is both the display panel and card not being fully recognized. No idea where to go from here -- I looked at the DSDT and I have no idea how to start rewriting that.

ss1.png
 
Hi guys, I am having an issue with getting my graphics working well. I have a clevo p750dm-g, i7-6700k, 32gb (2x16), nvidia 970m, 4k screen @60 on a 950 m.2. I am able to boot in and have most things working except graphics. If I boot with nv_disable I can see the screen but it has flashing areas (it seems to be where any animations are being drawn). I have installed the nvidia web drivers (...05f02) and when I boot with nvda_drv=1 the screen sleeps after the verbose boot phase is done. I have tried using inject EDID and customEDID but it seems like the os is not understanding my display properties...

I have attached my EFI (which includes a boot log and F4 output) a dump of my EDID and ACPI from linux and the output from IOReg. Please let me know if you have any ideas -- banging my head against the wall.

View attachment 180785

Edited to add nv_disable flashing areas are more like .5fps refreshes for portions of the screen that are updating..

You're using nvda_drv=1. Did you install the Nvidia web drivers?

Did you patch IOKit? (required for UHD, see FAQ).

http://www.tonymacx86.com/el-capita...faq-read-first-laptop-frequent-questions.html

Note: You should remove the invalid flags you have in config.plist/Boot/Arguments. GraphicsEnabler, MountEFI, and GraphicsMode are not valid kernel flags.
 
You're using nvda_drv=1. Did you install the Nvidia web drivers?

Did you patch IOKit? (required for UHD, see FAQ).

http://www.tonymacx86.com/el-capita...faq-read-first-laptop-frequent-questions.html

Note: You should remove the invalid flags you have in config.plist/Boot/Arguments. GraphicsEnabler, MountEFI, and GraphicsMode are not valid kernel flags.


I am trying to use ncda_drv -- that would be the goal. Yes I have the web driver installed. When I turn it on and try to run with nvda_drv instead of nv_disable the display goes black/to sleep after the verbose text stops right before login. I think that the EDID and GPU are both not being sensed correctly but am unsure what to do to fix it. The way I read that thread it appears that the DVMT patches are related to driving that many pixels with a low mem/shared mem igpu. The IOkit patch looks like it is needed specifically for the intel gfx to cheat on driving those pixels? If I could get the gpu/display recognized correctly I don't know that I need to worry about that as I would not need to use the gpu in 3mb of vram fallback mode instead of 6gb or am I misunderstanding what that is patching? Does the DVMT prealloc even apply to a full dedicated GPU or is that just igpu shared memory?

I noticed those were dead options -- I was posting the question after 12 hours of bashing my head on the wall trying to get the edid/gpu recognized and trying random stuff by that point. =)

The only active gfx on this laptop is the 970m -- the igpu is disabled and not wired out to any ports.
 
I am trying to use ncda_drv -- that would be the goal. Yes I have the web driver installed. When I turn it on and try to run with nvda_drv instead of nv_disable the display goes black/to sleep after the verbose text stops right before login. I think that the EDID and GPU are both not being sensed correctly but am unsure what to do to fix it. The way I read that thread it appears that the DVMT patches are related to driving that many pixels with a low mem/shared mem igpu. The IOkit patch looks like it is needed specifically for the intel gfx to cheat on driving those pixels? If I could get the gpu/display recognized correctly I don't know that I need to worry about that as I would not need to use the gpu in 3mb of vram fallback mode instead of 6gb or am I misunderstanding what that is patching? Does the DVMT prealloc even apply to a full dedicated GPU or is that just igpu shared memory?

I noticed those were dead options -- I was posting the question after 12 hours of bashing my head on the wall trying to get the edid/gpu recognized and trying random stuff by that point. =)

The only active gfx on this laptop is the 970m -- the igpu is disabled and not wired out to any ports.

DVMT is specific to IGPU. But patching IOKit is required for either (UHD).
 
Code:
"3cec8ae287ee52a3622082bfc049bb86")
  echo "         --> El Capitan 10.11.3 IOKit (${GREEN}unpatched${OFF})"
  sudo perl -i.bak -pe 's|\xB8\x01\x00\x00\x00\xF6\xC1\x01\x0F\x85|\x33\xC0\x90\x90\x90\x90\x90\x90\x90\xE9|sg' /System/Library/Frameworks/IOKit.framework/Versions/Current/IOKit
  sudo codesign -f -s - /System/Library/Frameworks/IOKit.framework/Versions/Current/IOKit
  echo "         Patched"

The IOkit patch looks like it only applies to specific versions (that match md5 sigs) and falls through to a noop on unmatched. the version i have is not matching any of the signatures in the script. I also can't seem to find any info for what the patch is even meaning to patch (not that I would be able to make a different version myself anyways).

d41d8cd98f00b204e9800998ecf8427e /System/Library/Frameworks/IOKit.framework/Versions/Current/IOKit
 
It it looks like that patch is only relevant to pushing 4k 30hz over hdmi or displayport -- apple's site says most systems since 10.10 support 4k @60hz even on those connectors. https://support.apple.com/en-us/HT202856.

I really think the issue is either the EDID is not being recognized or that the GPU is not outputting to the right port/not recognized or something. This is just my first time doing this and I have no idea how to read the assembly or understand the internals well enough to understand if there is a patch I am missing or how to make one.
 
Code:
"3cec8ae287ee52a3622082bfc049bb86")
  echo "         --> El Capitan 10.11.3 IOKit (${GREEN}unpatched${OFF})"
  sudo perl -i.bak -pe 's|\xB8\x01\x00\x00\x00\xF6\xC1\x01\x0F\x85|\x33\xC0\x90\x90\x90\x90\x90\x90\x90\xE9|sg' /System/Library/Frameworks/IOKit.framework/Versions/Current/IOKit
  sudo codesign -f -s - /System/Library/Frameworks/IOKit.framework/Versions/Current/IOKit
  echo "         Patched"

The IOkit patch looks like it only applies to specific versions (that match md5 sigs) and falls through to a noop on unmatched. the version i have is not matching any of the signatures in the script. I also can't seem to find any info for what the patch is even meaning to patch (not that I would be able to make a different version myself anyways).

d41d8cd98f00b204e9800998ecf8427e /System/Library/Frameworks/IOKit.framework/Versions/Current/IOKit

From where did you download OS X?
 
It it looks like that patch is only relevant to pushing 4k 30hz over hdmi or displayport -- apple's site says most systems since 10.10 support 4k @60hz even on those connectors. https://support.apple.com/en-us/HT202856.

I really think the issue is either the EDID is not being recognized or that the GPU is not outputting to the right port/not recognized or something. This is just my first time doing this and I have no idea how to read the assembly or understand the internals well enough to understand if there is a patch I am missing or how to make one.

You should try InjectEDID (and CustomEDID) if needed.
 
You should try InjectEDID (and CustomEDID) if needed.

Re: Where did I get OSX from my mackbook pro late 2009, with the createinstall method.

I have tried the injectEDID and CustomEDID flags; both separately and in unison but I do not know that they were correct. In the zip I attached I have the xrandr output and show the CustomEDID block i was using in the config.plist if you could take a look and verify I am doing it right.
 
Status
Not open for further replies.
Back
Top