Contribute
Register

Editing AGPM to prevent GTX4XX Freeze

Status
Not open for further replies.
Agreed. Somehow I managed to fix it partly, not really sure how as I thought it didn't work. Now I do see AGPM in the console. My freezes are less frequent, and certainly not anything to really care about, considering Lion is around the corner.

I nearly broke my entire system doing this, so can't afford to screw it up again!
 
I've edited AGPM.kext but I used the code meant for MacPro4,1 and instead marked it as iMac12,1 because it will kernel panic at boot with MacPro in SMBIOS. At first I could see AGPM loading and changing states in Console, now it doesn't show up when I search for it, there was another person out there who said this happened, I don't know what the problem is.
My real concern is the use of UseNvidiaRom=Yes in the Boot.plist. I have edited my original EVGA GTS 450 bios rom to have only one power state and corresponding voltage setting. In System Profiler it shows up as;
Vendor: NVIDIA (0x10de)
Device ID: 0x0dc4
Revision ID: 0x00a1
ROM Revision: /Extra/10de_0dc4.rom
Note: I have not flashed my bios, only loaded my edit at boot with UseNvidiaRom.

Will it now use the edited bios rom? I have the feeling this is an old solution for older 7-200 series Nvidia cards and not actually meant for what I'm trying to do. If it does use this rom then there shouldn't be any P state changes.

Update: Still getting kernel panic. I would be interested in knowing what this boot.plist Nvidia command is actually doing.
 
yeah..after few days, the KP is back. I broke down and ordered a 5770 instead.
 
Grooby said:
yeah..after few days, the KP is back. I broke down and ordered a 5770 instead.
You and me both. Everything runs so god damn smooth. Just put my GTX on sale on kijiji.
 
Sorry - long time no posts :) Was in the hospital for a bit and have busy with work since.

First - this isn't a perfect fix, obviously. It, to me at least, seems to mitigate crashes based on power management/throttling, but it seems like there's a host of things wrong with the driver. I've gotten a couple of OpenGL crashes that have brought down the system, for instance.

Second - the issue we're having with these cards are only temporary. Lion's updated Fermi drivers don't cause the same type of kernel panics we're seeing now in 10.6 - trust me on this ;)

Just wait a bit longer and the majority of issues we've seen with this card and drivers will be gone.
 
hi Scorcho,
hope all is well. I am definitely looking forward to Lion.
 
is this the same chameleon in MultiBeast 3.7.2?
 
Mutlibeast 3.7.2 has Chameleon 2.0 RC5 r877 in it already, what's linked above is an older version Chameleon 2.RCr851.
 
Last edited by a moderator:
eelhead said:
Mutlibeast 3.7.2 has Chameleon 2.0 RC5 r877 in it already, what's linked above is an older version Chameleon 2.RCr851.


so far I found R1000 here (http://www.kexts.com/view/1348-chameleo ... r1000.html)

with the change logs (http://forge.voodooprojects.org/p/chameleon/timeline/)

looking through the change logs right now..might give it a try though...

[EDIT]
install R1000 and also updated to the latest CUDA. Current uptime is 3 hours..will update.
The only changelog that might impact us is the HDMI audio fix for NVIDIA. Wonder if that's causing the crashes.

[EDIT again]

yup..crashed...yay!
 
Last edited by a moderator:
Status
Not open for further replies.
Back
Top