- Joined
- Jun 20, 2015
- Messages
- 5
- Motherboard
- Gigabyte GA-97MX-Gaming 5
- CPU
- Intel i7 4790K
- Graphics
- EVGA GeForce GTX 970 4GB FTW+ GAMING ACX 2.0+
- Mac
- Classic Mac
- Mobile Phone
First, here's the system:
This system had been rocking under OS 10.10 and PS CC 2014. However, some must have functionality was released, and so an upgrade was performed. Now, it was not that the system never crashed before; they were infrequent and irksome, but so rare that it could be excused. I considered the system stable. However, during my investigations, the crashes as they occurred could not be reliably produced. That is, until now...
Issue
When using the Healing tool on any image in PS CC 2015, I can reliably crash the entire system to disk. I do not pass go. I do not collect $200. I do not have any logs of the crash. The system just clicks, and then it restarts. It prevents using the environment for anything Adobe CC related, which was the reason for this build.
In addition, it appears that Lightroom has a similar instability. Periodically, LR will crash the system completely. I have tested this situation by leaving LR running (it will be doing a network sync with the Adobe Cloud), not doing any other work on the computer. However, it is a random crash and not reproducible, so I believe that since PS appears to expose the underlying issue,
Investigation
I first tested the same apps on a MacBook Pro (Retina 2015) running El Capitan. These same apps do not cause a crash. So, I know that they should not theoretically be the source of the issue. However, their actions may be creating some instability, one that the Hackintosh just can't handle.
I've scoured the Adobe boards, and I have reviewed the versions installed. I checked all support issues available, and while there were reports when El Capitan came out, they were resolved in later builds. I have installed the OS X Compatibility plugin to ensure that there are no issues remaining, but the logs indicate that the plugin is not required under the current PS version (2015.1.2).
I am told that downgrading is a high hurdle, since there are new features in the latest Adobe CC that are critical for use. So, while I would normally just consider going back to CC 2014, that does not appear to be an immediate option. Moreover, I see this situation as an uncovered instability, not something to avoid by downgrading.
I'd put a test matrix here, to show you all the ways I've tested it, but alas, there's no table. So, here are the major variables I tested. I've run the full factorial of all the tests, so there's not a combination I have not tried.
Video Driver: Apple vs. nVidia (2 options)
When using the Apple driver, the system will crash to disk within 1-2 seconds of tool use. When using the nVidia web driver, the system will crash to disk within 5-7 seconds of tool use.
OS Version: 10.10 vs. 10.11 (2 options)
I installed a second HD and then a clean 10.10 install. It was upgraded to the latest version. I then installed the latest nVidia web driver for that version. The system will crash under both environments; although, under 10.10, it lasts approximately 2 seconds longer when using the nVidia driver than 10.11. Otherwise, both OS versions manifest the issue.
App Settings: GPU Acceleration (5 options)
PS has a number of performance settings related to use of the graphics coprocessor. None of these options seem to affect the crash. I have tested PS with the acceleration turned off; it crashes. I have tested with the basic acceleration turned on; again, it crashed. I have enabled all options save OpenCL, and it still crashes. Lastly, with all options, it crashes. There does not appear to be a variance in the time to crash directly, more that both crash.
Analysis
This kind of crash has to be some buffer overflow or something similar that would cause OS X to completely crash to disk. It cannot be just a simple bug in the API calls.
I don't think it's a hardware related issue, since I have run the system components here for some time, and they all have not generated this kind of reliable error before. In fact, the system was stable and very fast, generating a Geekbench score well above many of the MacPro configurations.
Searching here has identified my next avenue of investigation. I have recently identified that there may be an issue with the ECC specification of the MacPro6,1 profile, and that I should change it to iMac14,2. Otherwise, I should remove the AppleTyMCEDriver.kext extension, since my system does not have ECC memory. Both of these issues revolve around memory, and this type of crash bears the hallmarks of a memory error.
Request
When I first looked on these boards, I was surprised I could not find a direct solution. That either said I was missing something, or that others aren't seeing it. I hope it's the former rather than the latter. Obviously, I'd like to resolve the issue. I'm not a kernel debugger, so beyond what I have done, I'm a bit out of my depth. So, I'm making these requests:
All help is greatly appreciated, and feel free to make any requests for information. I will make sure that regardless of the outcome, I update the thread so that we do not lose the lessons learned here.
Additional Note 1:
Prior to LR 6, the system used a EVGA GeForce GTX 760 Super Clocked ACX 2GB GDDR5 SLI Ready graphics card. However, LR indicated that the card was not supported under El Capitan for GPU acceleration, so the decision was made to upgrade. LR now indicates that the card is OpenGL 4 compliant, so it will enable the acceleration. Nonetheless, it remains unstable.
- Clover boot loader (using MacPro 6,1 information)
- Gigabyte Z97MX-Gaming 5
- Intel i7 4790K (not overclocked)
- Crucial Ballistix Sport XT 32GB Kit (8GBx4) DDR3 1866 MT/s
- EVGA GeForce GTX 970 4GB FTW+ GAMING ACX 2.0+
- OS X 10.11.4 (along with MacPro6,1 video fix)
- Latest nVidia web drivers
This system had been rocking under OS 10.10 and PS CC 2014. However, some must have functionality was released, and so an upgrade was performed. Now, it was not that the system never crashed before; they were infrequent and irksome, but so rare that it could be excused. I considered the system stable. However, during my investigations, the crashes as they occurred could not be reliably produced. That is, until now...
Issue
When using the Healing tool on any image in PS CC 2015, I can reliably crash the entire system to disk. I do not pass go. I do not collect $200. I do not have any logs of the crash. The system just clicks, and then it restarts. It prevents using the environment for anything Adobe CC related, which was the reason for this build.
In addition, it appears that Lightroom has a similar instability. Periodically, LR will crash the system completely. I have tested this situation by leaving LR running (it will be doing a network sync with the Adobe Cloud), not doing any other work on the computer. However, it is a random crash and not reproducible, so I believe that since PS appears to expose the underlying issue,
Investigation
I first tested the same apps on a MacBook Pro (Retina 2015) running El Capitan. These same apps do not cause a crash. So, I know that they should not theoretically be the source of the issue. However, their actions may be creating some instability, one that the Hackintosh just can't handle.
I've scoured the Adobe boards, and I have reviewed the versions installed. I checked all support issues available, and while there were reports when El Capitan came out, they were resolved in later builds. I have installed the OS X Compatibility plugin to ensure that there are no issues remaining, but the logs indicate that the plugin is not required under the current PS version (2015.1.2).
I am told that downgrading is a high hurdle, since there are new features in the latest Adobe CC that are critical for use. So, while I would normally just consider going back to CC 2014, that does not appear to be an immediate option. Moreover, I see this situation as an uncovered instability, not something to avoid by downgrading.
I'd put a test matrix here, to show you all the ways I've tested it, but alas, there's no table. So, here are the major variables I tested. I've run the full factorial of all the tests, so there's not a combination I have not tried.
Video Driver: Apple vs. nVidia (2 options)
When using the Apple driver, the system will crash to disk within 1-2 seconds of tool use. When using the nVidia web driver, the system will crash to disk within 5-7 seconds of tool use.
OS Version: 10.10 vs. 10.11 (2 options)
I installed a second HD and then a clean 10.10 install. It was upgraded to the latest version. I then installed the latest nVidia web driver for that version. The system will crash under both environments; although, under 10.10, it lasts approximately 2 seconds longer when using the nVidia driver than 10.11. Otherwise, both OS versions manifest the issue.
App Settings: GPU Acceleration (5 options)
PS has a number of performance settings related to use of the graphics coprocessor. None of these options seem to affect the crash. I have tested PS with the acceleration turned off; it crashes. I have tested with the basic acceleration turned on; again, it crashed. I have enabled all options save OpenCL, and it still crashes. Lastly, with all options, it crashes. There does not appear to be a variance in the time to crash directly, more that both crash.
Analysis
This kind of crash has to be some buffer overflow or something similar that would cause OS X to completely crash to disk. It cannot be just a simple bug in the API calls.
I don't think it's a hardware related issue, since I have run the system components here for some time, and they all have not generated this kind of reliable error before. In fact, the system was stable and very fast, generating a Geekbench score well above many of the MacPro configurations.
Searching here has identified my next avenue of investigation. I have recently identified that there may be an issue with the ECC specification of the MacPro6,1 profile, and that I should change it to iMac14,2. Otherwise, I should remove the AppleTyMCEDriver.kext extension, since my system does not have ECC memory. Both of these issues revolve around memory, and this type of crash bears the hallmarks of a memory error.
Request
When I first looked on these boards, I was surprised I could not find a direct solution. That either said I was missing something, or that others aren't seeing it. I hope it's the former rather than the latter. Obviously, I'd like to resolve the issue. I'm not a kernel debugger, so beyond what I have done, I'm a bit out of my depth. So, I'm making these requests:
- If the memory issue does not pan out, what's my next route of investigation? Do I install a debugger and see if I can intercept the error as it occurs, identifying what's happening?
- Are there any other items I should be considering to resolve this issue?
All help is greatly appreciated, and feel free to make any requests for information. I will make sure that regardless of the outcome, I update the thread so that we do not lose the lessons learned here.
Additional Note 1:
Prior to LR 6, the system used a EVGA GeForce GTX 760 Super Clocked ACX 2GB GDDR5 SLI Ready graphics card. However, LR indicated that the card was not supported under El Capitan for GPU acceleration, so the decision was made to upgrade. LR now indicates that the card is OpenGL 4 compliant, so it will enable the acceleration. Nonetheless, it remains unstable.
