**Help me with understanding the anatomy of Nvidia kexts and GPU hardware acceleration**

Not open for further replies.
Jul 9, 2015
2x Xeon 5680
GTX 480
  1. MacBook Pro
  2. Mac Pro
Mobile Phone
  1. Other
one hackintosh build I continually have to make is a snow leopard build for the multiple Pro Tools 9 HD computers required in our facility. These builds have come in many different forms; all in the attempt to build the fastest snow leopard build possible. Using a vanilla kernel the fastest build I have is a dual xeon 5600 supermicro based 'mac pro 5.1' - using a patched kernel I can go up to a i7 4790k system -- these are certainly powerful enough machines in terms of 'CPU power' so in reality the main bottleneck for building my snow leopard systems is GPU support.

The higher power machines all use a GTX 480 flashed for mac [maybe the fastest card for snow leopard]

The lower power machines use an Apple GT120 Card

The portable 'even lower power' machines use intel HD3000 (patched snb frame buffer kext)

In a recent experiment on one of the 'guinea pig' builds I installed the nvidia retail drivers for OSX 10.7.5 after realizing that the kexts referenced the same OS library versions as the snow leopard kexts. Upon rebooting the machine I see that NVDANV50Hal.kext is 'loaded and registered' [this is a GT 120 build] and I have full resolution and the GPU vRAM is properly recognized in system profiler but there is no hardware acceleration.

I am just very curious as to what is the missing link since the kexts load a register just fine -

I have always wanted to see a diagram of how the GPU hardware support works in a driver/kernel relationship chart.

Does anyone know why kexts that load and register fine are not providing the desktop hardware acceleration they are meant for? Is it the difference in mach kernel ? Is it not a kext related situation ? etc....just lots of questions as I was surprised I could get kexts for Lion to load and register on Snow Leopard
Not open for further replies.