Jump to content

Mavericks kernel testing on AMD (formerly Mountain Lion kernel testing on AMD)


theconnactic
 Share

6,414 posts in this topic

Recommended Posts

sinetek ,just throwing a guess in regarding Xcode

 

since in SL you had to patch libSystem.B_debug.dylib with anvs amd_insn_patcher in order to run xcode smoothly even in 32 bit or 64bit userland

 

worth a shot

Link to comment
Share on other sites

sinetek ,just throwing a guess in regarding Xcode

 

since in SL you had to patch libSystem.B_debug.dylib with anvs amd_insn_patcher in order to run xcode smoothly even in 32 bit or 64bit userland

 

worth a shot

 

 

Akimoa, there's no need to patch insn with Bronzovka's RC7 10.7.5 kernel. I'm using it, and the solution i found was simply toggling it to always run in 32-bit mode, without altering important system files.

 

whoooahhh !!!

 

 

comment as tu fais ça sous ML 10.8 !!?? moi je n'y arrive pas avec mes cartes graphiques ATI !! :(

 

 

how did you do it under ML 08.10!?? I can not do it with my ATI graphics cards! :(

 

Gilles, i didn't really have to do anything besides the plist edits that are needed even with Intel hacks for it to work. Which AMD cards do you have?

Link to comment
Share on other sites

Hi, Gilles!

 

AMDCedarGraphicsAccelerator

ATIEnableWideBlitSupport

ATIUseTearingWideBlit

CFBundleIdentifier

com.apple.AMDRadeonAccelerator

GpuDebugPolicy

0

IOClass

AMDCedarGraphicsAccelerator

IODVDBundleName

AMDRadeonVADriver

IOKitDebug

0

IOMatchCategory

IOAccelerator

IOPCIMatch

0x68E01002>

 

You'll have also to add the same ID under IOPCIMatch in the info.plist of the ATI5000Controller.kext.

 

All the best!

  • Like 1
Link to comment
Share on other sites

@theconnactic,

 

I have tested the 10.7.4 kernel against an install of 10.7.5, and it fails. By that I mean that the WindowServer shows a black screen with a functional cursor. I'm more and more inclined to believe that this is a user-side software problem. We need to test the 10.7.4 kexts on 10.7.5 now, to see if anything gets any better.

 

This is assuming 10.7.4 works in x64 mode.

What do you say?

Link to comment
Share on other sites

@theconnactic,

 

I have tested the 10.7.4 kernel against an install of 10.7.5, and it fails. By that I mean that the WindowServer shows a black screen with a functional cursor. I'm more and more inclined to believe that this is a user-side software problem. We need to test the 10.7.4 kexts on 10.7.5 now, to see if anything gets any better.

 

This is assuming 10.7.4 works in x64 mode.

What do you say?

Hi Sinetek, on my Phenom II X4 with Radeon HD 6570, iAtkos L2 updated to 10.7.5 works fine in both arch=i386 and arch=x86_64 (no GPU glitches). Bronya 10.7.5 kernel. What are you trying to do?

Link to comment
Share on other sites

Hi SS01! My messages were also addressed to you. I'm trying to see what's up with nVidia. With nVidia the GUI doesn't show up under 10.7.5 x64, nor under ML.

 

ATI users are also concerned, they report glitches under x64 as well.

Link to comment
Share on other sites

Hi SS01! My messages were also addressed to you. I'm trying to see what's up with nVidia. With nVidia the GUI doesn't show up under 10.7.5 x64, nor under ML.

 

ATI users are also concerned, they report glitches under x64 as well.

 

With a retail 10.7.5 installation on my ATI, I had glitches under x64. With iAtkos L2 updated to 10.7.5 and ATY_Init selected in customize menu, full acceleration and QE/CI, no glitches under x64.

 

Is there ATY_INit equivalent for NVIDIA maybe?

Link to comment
Share on other sites

Well, using those older KEXTs the GUI does start up, but it then locks up a while later, randomly. The cursor stops moving.

I think this may be due to the lack of i386 support in my kernel. Hmm.

QE/CI works with 10.7.4 drivers and x32, but it seems slow.

 

(for the record, i'm running 10.7.5)

 

UPDATE: Yeap, running 10.7.5 x86_64 on nVidia right now! Performance is {censored}, but it works. this should open the door to ML nVidia

 

QE/CI working, only slow.

 

 

Here's a list of the kexts I substituted:

 

GeForce.kext

GeForce7xxx.kext

GeForce7xxxGLDriver.bundle

GeForce7xxxGA.plugin

GeForce7xxxVADriver.bundle

GeForceGA.plugin

GeForceGLDriver.bundle

GeForceVADriver.bundle

Link to comment
Share on other sites

Well, using those older KEXTs the GUI does start up, but it then locks up a while later, randomly. The cursor stops moving.

I think this may be due to the lack of i386 support in my kernel. Hmm.

QE/CI works with 10.7.4 drivers and x32, but it seems slow.

 

(for the record, i'm running 10.7.5)

 

UPDATE: Yeap, running 10.7.5 x86_64 on nVidia right now! Performance is {censored}, but it works. this should open the door to ML nVidia

 

QE/CI working, only slow.

 

Congratulations. Any glitches?

 

:king:

Link to comment
Share on other sites

Amazing! As you said, this will certainly lead to ML nVidia. Now I just need more testers for ATY_Init..

 

For ATY_Init, you mean ATI users? Does it solve issues on ATI?

I feel like the fix on nVidia will be to backtrack a few kexts/bundles. It will take a while to indentify which one is crucial though. OpenGL driver/kext version mismatch and all...

Link to comment
Share on other sites

For ATY_Init, you mean ATI users? Does it solve issues on ATI?

I feel like the fix on nVidia will be to backtrack a few kexts/bundles. It will take a while to indentify which one is crucial though. OpenGL driver/kext version mismatch and all...

Yes, ATY_Init solved my ATI issues. I have yet to see if it works for everyone.

Link to comment
Share on other sites

I am currently testing out different Frameworks and kexts for ML just now. So far I'm stuck on a login loop with no lag when the login screen opens which seems to me like QE/CI is working to some extent.

Link to comment
Share on other sites

 Share

×
×
  • Create New...