Jump to content

[NEED HELP] Games making my system hard-freeze!


7 posts in this topic

Recommended Posts

Hi, nice people!

 

As some of you (hopefully) know by now, i've been very active these days in this community, trying to help building momentum for AMD hackintoshing, so we AMD users (again, hopefully) someday manage to have at least the opportunity to have an up-to-date OSX install (yeah, i'm talking about Mountain Lion).

 

Meanwhile, i decided to downgrade from the handicapped Lion installation thad i used to have, since it was 32-bit only and some key features of Lion are 64-bit, to Snow Leopard, which i could get full-featured. All that was a smooth and somehow enjoyable process, and now i have 10.6.8, running stable with 64-bit user land, all peripherals working (well, almost all of them, as you'll see) and sleep working flawlessly.

 

Seriously, it could be a really awesome build, and my Geekbench score in fact exceeds anything i got running Windows 7. But, as has been proven for so long, in hackintoshes, specially AMD ones, the sweet always comes with the sour. I told you that my peripherals are almost working, and i say working because of my nVidia Geforce G210. Yes, it got QE/CI. Yes, all three outs are working nicely, and even HDMI audio worked oob. The card is CUDA compatible, and CUDA drivers are up-to-date. Movies play smoothly. What is making the trouble?

 

Games!

 

Not all of them. Only the most cool ones. Oh, i can play App Store's Jet Dudes, Lep's World or U-Boot silly windowed games, and full screen games like Forest Zombies or Battle Monkeys. Ultraviolet Dawn appears to work either windowed or full-screen. Uberstrike also didn't give me any issues untill now. But games like Amnesia, Trine 2, Jason vs Zombies and the supposedly low profile Kindom Chronicles are a certainty to make my whole system crash. Heck, even the said Forest Zombies did it at least once, as well as Crazy Monster Truck Escape a couple of times.

 

the symptoms are the following: after some minutes into the game, usually less than ten, the game will crash giving me a black screen with no sound, and my HDD led starts to show no activity. The system don't reboot or give me a panic, just stay frozen like that. My only option is a hard reboot.

 

At first, i thought the problem could be related to those games being 32-bit, since they used to work with my 32-bit only 10.7.4 with no issues, and i had to use the amd_insn_patcher to get my 10.6.8 to run both 32-bit and 64-bit user mode. I could be an instability inherent to the patch, so i tried to boot with -legacy flag, but the crashes continued to happen.

 

Investigating a solution, i tried using other combinations of audio and video resolution. Changed for VGA and DVI monitors and tried both my built-in Via Audio and my KX-driven Soundblaster to no improvement. In a last ditch attempt, i simply removed CUDA drivers and re-tried all of the other troubleshootin methods i'd already used, to no avail.

 

Then i remembered to look in the Console logs. Funny i waited so long to do it, since it's the first thing i myself recommend for anyone. Anyway, the common messages i found at the times the crashes happened were these two: KGCErrorFailure (always) and CGSGetWindowBounds: Invalid connection (sometimes).

 

I google a lot, and, when i found i was running in circles with it and couldn't find a solution for my specific situation, since these error messages seems to be also related to other issues, i decided to bring it here, in the hopes someone could point me a light at the end of the tunnel here.

 

Here's an excerpt of my last WindowServer log at the exact time the system crashed for the last time (testing with Amnesia.app):

 

 

Oct 28 22:47:45 [75] kCGErrorIllegalArgument: CGXMoveWindowListToWorkspace: invalid window id: 66

Oct 28 22:47:47 [75] kCGErrorIllegalArgument: CGXMoveWindowListToWorkspace: invalid window id: 67

Oct 28 22:47:50 [75] kCGErrorIllegalArgument: CGXMoveWindowListToWorkspace: invalid window id: 68

Oct 28 22:47:55 [75] kCGErrorIllegalArgument: CGXSetWindowListTags: Operation on a window 0x6 not owned by caller loginwindow

Oct 28 22:47:57 [75] dict count after removing entry for window 0x5d is 0

Oct 28 22:48:17 [75] CGXRestartSessionWorkspace: session workspace exited for session 256 ( on console )

Oct 28 22:48:17 [75] loginwindow connection closed; closing server.

Oct 28 22:50:10 [72] Server is starting up

Oct 28 22:50:48 [72] CGXMappedDisplayStart: Display0 : no display alias property

Oct 28 22:50:48 [72] CGXMappedDisplayStart: Display1 : no display alias property

Oct 28 22:50:50 [72] GLCompositor: GL renderer id 0x01022614, GL mask 0x00000003, accelerator 0x00003123, unit 0, caps 0x00000003, vram 1024 MB

Oct 28 22:50:50 [72] GLCompositor: GL renderer id 0x01022614, GL mask 0x00000003, texture units 8, texture max 8192, viewport max {8192, 8192} extensions 0x0000000f

Oct 28 22:50:50 [72] agc_attach: Couldn't find any matches

Oct 28 22:50:50 [72] CGXPerformInitialDisplayConfiguration

Oct 28 22:50:50 [72] Display 0x43000000: MappedDisplay Unit 0; Vendor 0x410c Model 0x0 S/N 0; online enabled (0,0)[1920 x 1080], Rotation 0, base addr 0x106100000

Oct 28 22:50:50 [72] Display 0x3f003d: MappedDisplay Unit 1; Vendor 0xffffffff Model 0xffffffff S/N -1; offline enabled (2944,0)[1 x 1], Rotation 0, base addr 0x106100000

Oct 28 22:50:52 [72] Display 0x43000000: MappedDisplay Unit 0; ColorProfile "HD 709-A" (MD5 58a7456cff434584b31ed0a11f2da7be)

Oct 28 22:51:45 [72] kCGErrorFailure: MPHWFill: failed (-536870212)

Oct 28 22:53:51 [72] kCGErrorFailure: MPHWFill: failed (-536870212)

 

 

Thanks in advance!

Link to comment
Share on other sites

Hmmm... does HWSensors from kozlec run on Snow Leopard? Does it maintain a log so i can check it out after rebooting from the system crash?

 

Anyway, i must say it would be an atypical behavior if it's the case: overheating usually makes the motherboard shut down cold the computer. What happens here is that the computer stays on, just being there frozen with no image or sound. The lights of my keyboard, mouse and usb wi fi stick also stay on. The HDD led shows no activity though.

 

Besides all this, there is that KCG error message. Thank you for your suggestion, eep357, and i'll surely give it a go, but i think it's something else that's causing this very frustrating issue.

Link to comment
Share on other sites

IDK, I've never over heated and had a thermal shutdown before... But the way I watch temps you'd think I did all the time :) I have had freezes and kernel panics when stressing a new overclock, generally a combination of heat and/or voltage. Kozlek's HW monitor does have a graph view where if you run it, play for a while and stop before you freeze, you can get a clear picture of what your temps and voltages were looking like and how they changed as time when on or any spikes, can't see why it wouldn't work in SL, but One thing I've learned about console logs, they can lead you to the cause of a problem, send you in the wrong direction, or when looked at for the heck of it, make you think you have problems that don't exist. Your issue could be caused by many different things, but generally a OS/file system problem will be just as (or more) likely to occur at launch or at seemingly random times(to us). Aside from a mobo thermal shutdown, which means your at TDP and crazy boiling water hot, heat decreases electrical efficiency, possibly effecting anything from your PSU to VRMs, MOSFETs, chokes, etc. where a couple upticks in BIOS CPU, IOH, PLL, or PCIe voltage can potentially stabilize things, or as all the fans start ramping up, something else doesn't get the voltage it needs and gets pissed off. Honestly, it's probably something else, just explaining my train of thought on stuff :)

  • Like 1
Link to comment
Share on other sites

IDK, I've never over heated and had a thermal shutdown before... But the way I watch temps you'd think I did all the time :) I have had freezes and kernel panics when stressing a new overclock, generally a combination of heat and/or voltage. Kozlek's HW monitor does have a graph view where if you run it, play for a while and stop before you freeze, you can get a clear picture of what your temps and voltages were looking like and how they changed as time when on or any spikes, can't see why it wouldn't work in SL, but One thing I've learned about console logs, they can lead you to the cause of a problem, send you in the wrong direction, or when looked at for the heck of it, make you think you have problems that don't exist. Your issue could be caused by many different things, but generally a OS/file system problem will be just as (or more) likely to occur at launch or at seemingly random times(to us). Aside from a mobo thermal shutdown, which means your at TDP and crazy boiling water hot, heat decreases electrical efficiency, possibly effecting anything from your PSU to VRMs, MOSFETs, chokes, etc. where a couple upticks in BIOS CPU, IOH, PLL, or PCIe voltage can potentially stabilize things, or as all the fans start ramping up, something else doesn't get the voltage it needs and gets pissed off. Honestly, it's probably something else, just explaining my train of thought on stuff :) In your 6 min timespan of console logs, it looks like you couldn't have been gaming for more than 3 mins, and those 2 error messages 2 minutes apart would be odd for a freeze, unless it unfreeze and refreeze after 2 mins. I get these scary looking ones,

CGXSetTrackingAreaEnabled : Invalid tracking area 0x7fd6711f4b50

3 at a time. Looking in console, you'd think something wrong. Thing is I can recreate the same 3 logs any time, by clicking the drop down menu in istat. istat works great, if you didn't look in console, you'd never know anything is wrong, because nothing is wrong. I know, you obviously have something wrong, just remember to take console logs with a grain of salt. :)

  • Like 1
Link to comment
Share on other sites

Hey, eep357, don't get me wrong: you're being very helpful here. :)

 

I had set up my BIOS so the system shuts down on relatively mild overheatings. The chance my CPU temp is going too high is almost null. But i'm wondering: could it be the GPU temp? What exactly could i do to solve it? The voltage subject was also a good insight, since my power supply unit is not that champ.

 

Anyway, things used to do just fine with Lion, even it running strictly 32-bit (and i indeed tried to run Snow Leopard in legacy mode to see if it solved anything, it didn't). Perhaps i'll try a NVDHAL kext "rollforth", that is, using the 10.7.4 kexts in my system. It really worths a try.

Link to comment
Share on other sites

 Share

×
×
  • Create New...