While my Mac Pro is being repaired, I am using exactly the same applications and almost the same workflow on my MacBook Air with no problems. I am still confident that this is combo of bad drivers and badly put together hardware (GPUs).
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Mac Pro (Late 2013) GPU (Driver) Issues
- Thread starter bax2003
- Start date
- Sort by reaction score
While my Mac Pro is being repaired, I am using exactly the same applications and almost the same workflow on my MacBook Air with no problems. I am still confident that this is combo of bad drivers and badly put together hardware (GPUs).
Still impressed that you insist on using a faulty device, instead of using a cheese grater or a Windows equivalent. It's nearly 2017 now and you try to fix things broken since end of 2013. Kudos to your patience, I just refunded this "pro" product after Apple failed to repair my faulty D700s for the third time.
I don't need to have newest and fastest hardware. Mac Pro (Late 2013) is fine for me in every single way except that is not working properly
I have AC until Nov 2018, and Apple will fix it even if they have to replace every component until the last screw.Mac Pro 2006 - 2016 - no (USB 2, SATA 2, no Thunderbolt, slow WiFi, horrible Bluetooth, 980W, don't need internal storage)
PC - really no
Last edited:
Did you use Windows alot? GPUs ended up getting fried and/or logic boards? How did you get refunded?
Mac Pro came back home with these two replaced:
A. Graphics Board, 2GB VRAM, D300, AMD, FIREPRO 661-7533
B. Graphics Board, 2GB VRAM, D300, AMD, FIREPRO 661-7549
Check these in your machines:
AMD FirePro D300:
Chipset Model: AMD FirePro D300
Type: GPU
Bus: PCIe
Slot: Slot-1
PCIe Lane Width: x16
VRAM (Total): 2048 MB
Vendor: ATI (0x1002)
Device ID: 0x6810
Revision ID: 0x0000
ROM Revision: 113-C4012F-687
Automatic Graphics Switching: Supported
gMux Version: 4.0.11 [3.2.8]
EFI Driver Version: 01.0B.687
Metal: Supported
A. Graphics Board, 2GB VRAM, D300, AMD, FIREPRO 661-7533
B. Graphics Board, 2GB VRAM, D300, AMD, FIREPRO 661-7549
Check these in your machines:
AMD FirePro D300:
Chipset Model: AMD FirePro D300
Type: GPU
Bus: PCIe
Slot: Slot-1
PCIe Lane Width: x16
VRAM (Total): 2048 MB
Vendor: ATI (0x1002)
Device ID: 0x6810
Revision ID: 0x0000
ROM Revision: 113-C4012F-687
Automatic Graphics Switching: Supported
gMux Version: 4.0.11 [3.2.8]
EFI Driver Version: 01.0B.687
Metal: Supported
i have exactly the same, and os = sierra 10.12.1. Freezings occur with variable frequence, from 1 or more per day to 1 or less per week.
The frequence depends on how and how much the computer is used, if i use it for some hour watching video in live streaming with chrome and meanwhile browsing with firefox it is likely to freeze. E.g. these days i was following the world chess championship, it was broadcast live with flash and i used chrome to watch it (i have not installed flash on the mac). Many of the games lasted more than 3-4 hours and in many cases i had the freezing. It seems like an overheating problem, i'm not even sure what to do to diagnose it. I have windowserver crash reports in the console each time. Sometimes the freezing is related to gpu restart, and in this case i find the gpu restart report log (actually many of them for each event) in the console.
Last edited:
Thanks for the update. I am very interested to find out how El Capitan works with your newly serviced MacPro. The 'new' GPUs seem to be the same model and generation as the previous ones. I wonder if freshly applied thermal compound plays a significant role in avoiding the lookups (provided they are indeed related to excessive heat).
Following up - so with even with the 1.0B EFi firmware and both GPU's replaced - you are still getting WindowServer crashes in Sierra?
Correct and it is not very user friendly. Every app is forced closed, you are at the login screen and you can wave all your work good bye.
My latest report on this saga including summary of past repairs:
Repair 1: 2014 - took machine to store, for 1 week they had it running but could not reproduce issue.
- took back home
Repair 2: - took machine to store, they replaced video cards, and other components. After they had the machine for 5 weeks I cried so hard they gave me a new machine.
2015 - Machine crashes every 2 days or so:
DumpGPURestart[3623]: Saved gpuRestart report for kernel to /Library/Logs/DiagnosticReports/Kernel_2016-12-09-130919_mammoth.gpuRestart
- - I mostly ignore the problem while my ulcer grows in size.
2016:
- Repair #3 - Take machine to store, GeniusBar diagnostics finds no trouble (no surprise there) they refuse to take the machine in, instead have me take it home so that I can capture data for engineering.
- Repair #4 - Engineering sees the logs and deems one card as faulty. Apple replaced both D300 cards for the D700 model.
- - - - -
As a test today: left a 3 hour long youtube video playing through Chrome AirPlayed to an Apple TV.
2 hours later:
Event: GPU Reset
Date/Time: Fri Dec 9 14:43:08 2016
Application: AirPlayXPCHelper
Path:
Tailspin: /Library/Logs/DiagnosticReports/gpuRestart2016-12-09-134508.tailspin
GPUSubmission Trace ID: 0
OS Version: Mac OS X Version 10.12.1 (Build 16B2555)
Graphics Hardware: AMD FirePro D700
Signature: 100
Report Data:
GPURestartReportStart
------------------------
Hung Channels: VCELLQ
------------------------
[08] AccelChannel: VCELLQ
Pending Command from : VideoCtx
PendingCommandTimestamp: 0x000f7d58, TotalDWords: 0x00000000, GART Offset=0x0000000180d2d300, stamp_idx=8, estamp=0x000f7d58
PendingCommandStart:
PendingCommandEnd
------------------------
[08] VCELLQ HWChannel : Enabled, Idle
LastReadTimestamp : 0x000f7d57
NextSubmitTimestamp : 0x000f7d59
[VCELLQ, ts:0x000f7d58]: No semaphore wait
[VCELLQ, ts:0x000f7d58]: semaphore signal: 0xffffff80b9c2eac0
HWSemaphore 0xffffff80b9c2eac0 Signal Event: [VCELLQ channel[8] TS:0x000f7d58 lastReadTS:0x000f7d57]
- - - -
I miss you Steve Jobs.
Repair 1: 2014 - took machine to store, for 1 week they had it running but could not reproduce issue.
- took back home
Repair 2: - took machine to store, they replaced video cards, and other components. After they had the machine for 5 weeks I cried so hard they gave me a new machine.
2015 - Machine crashes every 2 days or so:
DumpGPURestart[3623]: Saved gpuRestart report for kernel to /Library/Logs/DiagnosticReports/Kernel_2016-12-09-130919_mammoth.gpuRestart
- - I mostly ignore the problem while my ulcer grows in size.
2016:
- Repair #3 - Take machine to store, GeniusBar diagnostics finds no trouble (no surprise there) they refuse to take the machine in, instead have me take it home so that I can capture data for engineering.
- Repair #4 - Engineering sees the logs and deems one card as faulty. Apple replaced both D300 cards for the D700 model.
- - - - -
As a test today: left a 3 hour long youtube video playing through Chrome AirPlayed to an Apple TV.
2 hours later:
Event: GPU Reset
Date/Time: Fri Dec 9 14:43:08 2016
Application: AirPlayXPCHelper
Path:
Tailspin: /Library/Logs/DiagnosticReports/gpuRestart2016-12-09-134508.tailspin
GPUSubmission Trace ID: 0
OS Version: Mac OS X Version 10.12.1 (Build 16B2555)
Graphics Hardware: AMD FirePro D700
Signature: 100
Report Data:
GPURestartReportStart
------------------------
Hung Channels: VCELLQ
------------------------
[08] AccelChannel: VCELLQ
Pending Command from : VideoCtx
PendingCommandTimestamp: 0x000f7d58, TotalDWords: 0x00000000, GART Offset=0x0000000180d2d300, stamp_idx=8, estamp=0x000f7d58
PendingCommandStart:
PendingCommandEnd
------------------------
[08] VCELLQ HWChannel : Enabled, Idle
LastReadTimestamp : 0x000f7d57
NextSubmitTimestamp : 0x000f7d59
[VCELLQ, ts:0x000f7d58]: No semaphore wait
[VCELLQ, ts:0x000f7d58]: semaphore signal: 0xffffff80b9c2eac0
HWSemaphore 0xffffff80b9c2eac0 Signal Event: [VCELLQ channel[8] TS:0x000f7d58 lastReadTS:0x000f7d57]
- - - -
I miss you Steve Jobs.
One week later: not a single problem on OS X El Capitan 10.11.6.
I am using MP as usual, heavy multitasking: chrome, itunes, vmware, photoshop, finder, mail, vlc...which usually killed window server at least once a day on macOS Sierra 10.12.1.
Testing continues....
I am using MP as usual, heavy multitasking: chrome, itunes, vmware, photoshop, finder, mail, vlc...which usually killed window server at least once a day on macOS Sierra 10.12.1.
Testing continues....
I've had the same problem. Got my Mac Pro in May 2016, running mainly ProTools 12 with Apollo Thunderbolt Interface, OWC TBolt hub.
Within a week had the issue: screen freezes, audio continues, can't exit out of any software so I force quit. I had been running the same system on a Mac Mini with none of these problems (unfortunately it just was not powerful enough).
Console lists repeatedly: GPU Driver appears to be hung (over 5 continuous seconds of unreadiness.)
Troubleshooting included no internet, running one monitor, disconnecting OWC TBolt hub, adding more RAM.
Brought it into Apple last September and they replaced the Graphics Card A. The problem re-occurred and I did some diagnostics with the Apple Guy on the phone, which reduced the incidences for a while but they promptly returned. (These were: Reset System Mgmt Controller [unplug all connections and hold down power button for 15 secs]; Reset NVRAM [start holding down OPT COMM P R keys]; and reboot in safe mode)
I just got it back from the Apple Store (for the second time) and this time they replaced both Graphics cards as well as the Logic Board, and it's frozen twice in the hour since I booted it back up, straight from the Apple Store.
Oddly enough, they said they needed to erase the computer to run a full diagnostic but they did not erase it. They were very insistent on erasing it in order to properly diagnose it...
Has anyone been able to resolve this problem?
Within a week had the issue: screen freezes, audio continues, can't exit out of any software so I force quit. I had been running the same system on a Mac Mini with none of these problems (unfortunately it just was not powerful enough).
Console lists repeatedly: GPU Driver appears to be hung (over 5 continuous seconds of unreadiness.)
Troubleshooting included no internet, running one monitor, disconnecting OWC TBolt hub, adding more RAM.
Brought it into Apple last September and they replaced the Graphics Card A. The problem re-occurred and I did some diagnostics with the Apple Guy on the phone, which reduced the incidences for a while but they promptly returned. (These were: Reset System Mgmt Controller [unplug all connections and hold down power button for 15 secs]; Reset NVRAM [start holding down OPT COMM P R keys]; and reboot in safe mode)
I just got it back from the Apple Store (for the second time) and this time they replaced both Graphics cards as well as the Logic Board, and it's frozen twice in the hour since I booted it back up, straight from the Apple Store.
Oddly enough, they said they needed to erase the computer to run a full diagnostic but they did not erase it. They were very insistent on erasing it in order to properly diagnose it...
Has anyone been able to resolve this problem?
If you are still getting Kernel.GPURestarts, you have to take to Apple until you get 100% functional GPUs.
That's right, you can see the difference between: Kernel GPU.restart and WindowsServer crash (GPU unreadiness) in console. I recommend you that you stay away from Sierra for now. Even with replaced GPUs, I was still getting Windows Server crashed on Sierra 10.12.1, but everything is fine under El Capitan 10.11.6.
I have the same exact GPU hardware/firmware as @bax2003. My GPU-A was replaced Nov 24th and today was the first time it gave me a GPU restart on macOS 10.12.1. Almost 19 days. Apple previously said they'd replace the second one but that's another week in the Apple shop. After that, who knows...
FWIW, I had Chrome, Safari, Word, Terminal, Path Finder and Apple Remote Desktop active on the desktop.
P.S. This is on my replacement Mac Pro.
P.P.S. I really like the deal @krolingo got.
FWIW, I had Chrome, Safari, Word, Terminal, Path Finder and Apple Remote Desktop active on the desktop.
P.S. This is on my replacement Mac Pro.
P.P.S. I really like the deal @krolingo got.
Is there any chance that these problems are related to unstable power supply or to an UPS absence?
Who knows, could be. That is one of the reasons why I asked for everyone who had these problems to write, among others things, do they have UPS / Voltage Stabilizer. Answers are mixed so, I don't think that power source is the problem. PSU itself....maybe. Apple very rarely replaces PSU on nMP when these problems occur.
Unstable power on all Mac Pro ports could have something to do with the issue. I did notice that the power light on my mouse which is connected to the USB port goes out every once in a while for a very short time. Whenever that happens the cursor remains in the same position on the screen until the mouse is supplied with power again. That happens despite running OS X 10.9 which completely avoids all GPU issues (I didn't have a crash in months).
The USB power issue is affected by the USB extension chord. If the mouse is connected to the USB 2 hub of the Cinema 30 display the mouse stops working for 2-3 seconds before coming back. The video portion is not lost. The same holds true for a USB 3 hub without external power supply.
If the mouse is connected directly to the Mac Pro via a USB 2 extension chord then the light goes out for less than a second and almost comes back immediately. If I use no extension chord and connect the mouse directly to the Mac Pro the light stays on at all times.
I also have issues with external 2.5 USB drive enclosures. SSD drives which use less power do work with USB 3, USB 2 and FireWire (with adapter) enclosures. Regular 2.5 drives do not work reliably with the same USB 3 enclosure but do work fine with the USB 2 and FireWire enclosures.
These USB power issues could be related to the fact that I have six 30 inch monitors connected which all draw USB power via the dual link adapter. Either way the USB ports in the Mac Pro seem to be underpowered.
To use the Mac Pro without any issues I run OS X 10.9, have the mouse connected directly via the shortest possibly USB 2 extension chord, use a wireless keyboard and track pad and use hard drives in USB 2 or FireWire enclosures for backups.
The USB power issue is affected by the USB extension chord. If the mouse is connected to the USB 2 hub of the Cinema 30 display the mouse stops working for 2-3 seconds before coming back. The video portion is not lost. The same holds true for a USB 3 hub without external power supply.
If the mouse is connected directly to the Mac Pro via a USB 2 extension chord then the light goes out for less than a second and almost comes back immediately. If I use no extension chord and connect the mouse directly to the Mac Pro the light stays on at all times.
I also have issues with external 2.5 USB drive enclosures. SSD drives which use less power do work with USB 3, USB 2 and FireWire (with adapter) enclosures. Regular 2.5 drives do not work reliably with the same USB 3 enclosure but do work fine with the USB 2 and FireWire enclosures.
These USB power issues could be related to the fact that I have six 30 inch monitors connected which all draw USB power via the dual link adapter. Either way the USB ports in the Mac Pro seem to be underpowered.
To use the Mac Pro without any issues I run OS X 10.9, have the mouse connected directly via the shortest possibly USB 2 extension chord, use a wireless keyboard and track pad and use hard drives in USB 2 or FireWire enclosures for backups.
Last edited: