View Single Post
 
Old 04-29-2009, 11:55 PM
andyvand's Avatar
andyvand andyvand is offline
 
Join Date: Apr 2009
Location: Tienen
Posts: 515
Quote:
Originally Posted by naquaada View Post
Very interesting news: I removed the 10.5.6 system on my A8N-E FM/S and restored a 10.5.3 system which was based on Leo4All V3 10.5.2 and an AMD patched 10.5.3 update. Then I installed Andys new kernel and made a second copy of it in the root directory, mach_kernel_960.

I don't know if the 9.6.0 kernel was made for 10.5.3, but it worked. Then I tried the TBU - Total Brutal Update. A direct ComboUpdate from 10.5.3 to 10.5.6 on AMD. And it worked! After the update was installed I used the backup kernel for booting, the normal mach_kernel was overwritten with the vanilla kernel from the update. One more reboot was neccessary, then the system booted - to a black screen. OK, no gfx drivers anymore ;-)

I installed from my main system the drivers for the HD2600XT and rebooted. The system was there. But not everything is working, sure. 'About this Mac' leads to a logout/login, no PS/2 drivers, no hardware and RAM information in System Profiler, Geekbench shows 3.70 GHz (should be 2.41 GHz) an so on. But this can be restored. Here's a screenshot.

Much more intersting is, the speed is back! I got about 80-85 points in Xbench in comparison to 59-75 on the direct 10.5.6 installation! Geekbench has 75 points more. Even after the update to Quicktime 7.6, which I thought was the main error source. Mainboard and BIOS are correctly detected by Geekbench, don't know if this is good or bad.

I think I won't mess around with a 10.5.6 image now, I'll test a new installation of 10.5.2 with a direct update to 10.5.7! If this would work on AMD it would be a hammer!
Normally with my SMBIOS alone it should enable the BIOS to be spoofed as MacPro3,1 BIOS...
Little note: for AMD MacPro3,1 tends to not fool Geekbench but if you use macmodel=MacPro2,1 with my kernel... you'll be surprised...
If you want to get a faster system you can try my decrypts... they are fully patched up without code signatures and really tend to speed stuff up (especially boot...)
To the question why it show 0 Mhz? well I think that something else was interfering with the SMBIOS and it thus misdetected it...
As to the point of using the kernel on 10.5.3... sure this should work as long as System.kext and seatbelt.kext match the kernel...
Also 10.5.7 should provide no problems... I'll also make a custom kernel for it when it comes out...
Also AppleSMBIOSEFI, SMBIOSEnabler, SMBIOSResolver, ... tend to conflict with my AppleSMBIOS...
It overwrites the values set by my driver... (not wanted ofcourse)
Reply With Quote