Mac Pro (Late 2013) GPU (Driver) Issues

[axantas]

One system freeze with 10.12.4 identical to Pardo83 - with one exception: I don't have Chrome and never had it installed. It happened while attempting to klick the menu bar of the Finder.

 

[Pardo83]

Just got another freeze (the second in two hours). This time the mouse was moving but nothing was clickable. I decided to wait and after 5 minutes everything came back to normal.

This is what I found in console, this time is not a GPU driver not responding, it's a GPU reset.

Thu May 4 09:32:54 2017

Event: GPU Reset
Date/Time: Thu May 4 09:32:54 2017
Application: WindowServer
Path:
Tailspin: /Library/Logs/DiagnosticReports/gpuRestart2017-05-04-093254.tailspin
GPUSubmission Trace ID: 0
OS Version: Mac OS X Version 10.12.4 (Build 16E195)
Graphics Hardware: AMD FirePro D300
Signature: e

Report Data:

GPURestartReportStart



I think I'll go back to Mavericks at this point. I wonder if Apple will trade this mac pro for the new one when it will be out

 

[bax2003]

Unfortunately, classic Mac Pro experience in Mac OS X 10.10+

10.10 - 10.11 - spinning wheel (GUI not clickable, audio can continue playing in bg) - only SSH reboot via termnal can help to avoid hard reboot.

10.12 - spinning wheel, freeze ends up crashing all your apps while sending you to login screen.

Difference in console logs is that in 10.10 - 10.11 OS mostly reports GPU.restarts, and 10.12 reports WindowServer crashes because of unresponsive GPU driver - if you ask me: its the same thing.

 

[pierre1610]

I'm having the same issue as many.
System is as follows:
Mac Pro (Late 2013)
OSX El Capitan 10.11.6
3.5GHz 6-core
64GB 1866 MHz DDR3 ECC (OWC memory)
FirePro D300

The error is repeated in the console over and over again.

05/05/2017 20:57:32.116 WindowServer[235]: GPU Driver appears to be hung (over 5 continuous seconds of unreadiness)


I have discovered I can put the Mac to sleep with the keyboard shortcut (cmd-alt-eject) wait until the computer actually goes to sleep, power button on a slow blink. I can then wake it up and it's back to working normally.

 

[spillahcrown]

I just want to chime in here. I've been dealing with this issue for almost 3 months.

In December 2016, I purchased:

• Mac Pro (Late 2013)
• OSX El Capitan 10.11.6
• 3.5GHz 6-core
• 32GB 1866 MHz DDR3 ECC (OWC memory)
• FirePro D300

Since then, I've done every software related solution possible. OS update. PRAM reset. Clean install.

Nothing.

I've had my machine sent for 3rd party diagnostics and, everything appears to check out but as soon as I get it back - random freeze, reboot. GPU related kernel panic.

So then, I booked my appointment at the local Genius Bar. I showed them all the crash logs and the diagnostics which showed multiple GPU resets. Sometimes running multiple apps, sometimes NONE. Everything's under warranty so, they took it in. After a week I got it back with a new Logic Board, I/O Board and Wireless Card.

Machine failed within the first 4 hours. Again, multiple GPU resets.

Here's one:

Event: GPU Reset
Date/Time: Thu May 11 16:54:27 2017
Application:
Path:
Tailspin: /Library/Logs/DiagnosticReports/gpuRestart2017-05-11-165427.tailspin
GPUSubmission Trace ID: 0
OS Version: Mac OS X Version 10.12.4 (Build 16E195)
Graphics Hardware: AMD FirePro D300
Signature: 40
​

Here's another:

Mon May 15 09:19:16 2017
Event: GPU Reset
Date/Time: Mon May 15 09:19:16 2017
Application:
Path:
Tailspin: /Library/Logs/DiagnosticReports/gpuRestart2017-05-15-091916.tailspin
GPUSubmission Trace ID: 0
OS Version: Mac OS X Version 10.12.4 (Build 16E195)
Graphics Hardware: AMD FirePro D300
Signature: e​

I've taken it back to Apple again - today. Apparently, they're replacing the Graphics Card. After that - there's nothing left. It's pretty much a new machine - with the exception of the 512 SSD and the processor.

I've used MACS for almost 20 years. I've NEVER had a problem like this.

 

[CC88]

Dear all

same problem here almost with 10.12.5 recently released.

I own a 6core macpro with 2xD300 Gpu. Sometimes it freeze and I can only move mouse cursor. Other time it logout me without any apparently reason.

Are there any macpro 2013 that work fine out there? Or I have to switch to a Windows machine?

Can you tell me how to check the error event? Every time I reboot seems I loose log. Is it right or I'm wrong?

 

[richmlow]

Hi CC88,


I sympathize with the lockups, crashes, etc. that you and many others are constantly experiencing with the 2013 Mac Pro. Even though Apple has not been forthright about the cause of the problems, there apparently is some sort of design flaw(s) with the 2013 Mac Pro.

For me, I've been fortunate so far. I got a "refurbished" 2013 Mac Pro not that long ago and it currently works well.

I wish you and others best of luck in resolving these problems. Hopefully, Apple will take some responsibility in this fiasco.


Sincerely,
richmlow

Dear all

same problem here almost with 10.12.5 recently released.

I own a 6core macpro with 2xD300 Gpu. Sometimes it freeze and I can only move mouse cursor. Other time it logout me without any apparently reason.

Are there any macpro 2013 that work fine out there? Or I have to switch to a Windows machine?

Can you tell me how to check the error event? Every time I reboot seems I loose log. Is it right or I'm wrong?

 

[ManuelGomes]

Maybe 10.12.6 beta 1 (out for devs) will do the trick.

 

[CC88]

For me, I've been fortunate so far. I got a "refurbished" 2013 Mac Pro not that long ago and it currently works well.

I got it for months without issue. Than upgraded to Sierra 10.12.4

In the meantime I have added a thunderbolt external raid that I didn't have in El Capitan.

I have worked for hours without any problem at all with FCPX, DaVinci etc.

The freezes occur primarly with Luxology Modo and Safari (now I have disabled WebGL in the browser, hope it help).

Finger crossed for 10.12.6

 

[spillahcrown]

Dear all

same problem here almost with 10.12.5 recently released.

I own a 6core macpro with 2xD300 Gpu. Sometimes it freeze and I can only move mouse cursor. Other time it logout me without any apparently reason.

Are there any macpro 2013 that work fine out there? Or I have to switch to a Windows machine?

Can you tell me how to check the error event? Every time I reboot seems I loose log. Is it right or I'm wrong?

In CONSOLE - all recorded activity should be under system.log. Record the time of the freeze then, scroll through the log. You should be able to find your error.

Under /Library/Logs - you should be able to find your app logs.

 

[CC88]

In CONSOLE - all recorded activity should be under system.log. Record the time of the freeze then, scroll through the log. You should be able to find your error.

Under /Library/Logs - you should be able to find your app logs.

Thank you very much next time I will check it.

Today I worked in Modo without problems at all. Worked flawlessly.

 

[CC88]

In CONSOLE - all recorded activity should be under system.log. Record the time of the freeze then, scroll through the log. You should be able to find your error.

Under /Library/Logs - you should be able to find your app logs.

Although I didn't have any crash, I have checked the system log. But it stores only few hours. Do I have to open system.log.1.gz and so on to view the history and look for past freeze?

Thanks a lot for help.

 

[scoobs69]

Hi Everyone,

I've been reading through this thread with interest. I'm sorry to hear so many people have been going through this frustration. I've been looking at the 6-Core/D500 machine over the past few weeks, ready to pull the trigger at the end of May, the clearance page pricing is too good to pass up. Unless....

One question I have: is this primarily a D300 GPU issue? Most of the people I've seen commenting have the D300 GPU, and it seems apple pulled the D300 off their store offerings, unless you buy a refurbished machine. That's telling enough in and of itself.

Are people with the D500 and D700 GPU experienced the same issue?

Thanks in advance.

 

[bax2003]

Hi Everyone,

I've been reading through this thread with interest. I'm sorry to hear so many people have been going through this frustration. I've been looking at the 6-Core/D500 machine over the past few weeks, ready to pull the trigger at the end of May, the clearance page pricing is too good to pass up. Unless....

One question I have: is this primarily a D300 GPU issue? Most of the people I've seen commenting have the D300 GPU, and it seems apple pulled the D300 off their store offerings, unless you buy a refurbished machine. That's telling enough in and of itself.

Are people with the D500 and D700 GPU experienced the same issue?

Thanks in advance.

Its not just D300s. Apple sold much more those that D500/D700s, that´s the reason we can´t say D300 is more often a problem. Unortunately, there is no problem free nMP.

 

[scoobs69]

Thanks. I'm going to ask around to several of my photography message boards if people are having the same issue.

Will report back...

 

[CC88]

Dear all

about ten minutes ago I got my logout (were using Luxology Modo... the machines has frozen, spinball and then logout).

I now have the system.log. I can't find any realated GPU problems at all.

Here it is from dropbox shared link:
https://www.dropbox.com/s/sy4my4rajb4lndq/macpro2013.log?dl=0

I'm now thinking that I have problems only when I run Modo...

 

[voyager77]

I updated to 10.12.5 and all went well for a week.
Really tried to crash it, but it wouldn't.
Until today, another frozen screen with moving mouse. So much for the latest update.

I'll try to put it into service in the next weeks, but i dread to be without a working pc for weeks. (And apple or applecare doesn't offer any descent solution for that.)

Edit:
And when it rain it pours, it just crashed again.

 

[CC88]

I updated to 10.12.5 and all went well for a week.
Really tried to crash it, but it wouldn't.
Until today, another frozen screen with moving mouse. So much for the latest update.

I'll try to put it into service in the next weeks, but i dread to be without a working pc for weeks. (And apple or applecare doesn't offer any descent solution for that.)

Edit:
And when it rain it pours, it just crashed again.

Can you post the syslog.log or maybe the line about GPU crash? I can't find any in my syslog...

 

[MarkC426]

I have periodically viewed this thread, but have not kept up with it.
Now the nMP has been reduced, I am slightly tempted to get one.
However, if there is STILL an ongoing issue, I am not prepared to spend 5k on a 12 core machine which is going to give me problems.
Are brand new 'build to order' machines not being fitted with working gpu's.
Alternatively, would getting a refurb guarantee a fully tested machine with working gpu's.

 

[scoobs69]

That's kind of what I'm hoping Mark, as I'm looking at a 6-Core refurb. But if this GPU thing hasn't been resolved in four years, do I really want to risk getting a problematic computer - then fighting with it for the next few years.

For the first time since working on the Quadra 700 and 950's, boy am I dating myself, I'm seriously considering getting a custom built PC. Just not sure I want to deal with the hassle of swapping all of my software over.

 

[spillahcrown]

Although I didn't have any crash, I have checked the system log. But it stores only few hours. Do I have to open system.log.1.gz and so on to view the history and look for past freeze?

Thanks a lot for help.

Do you have the exact time the crash occured? Otherwise, that's a lot of code to go through.
[doublepost=1495202832][/doublepost]**UPDATE**

Apple has now replaced one of the two D300 Graphics Boards.

So, to date, they've replaced:

• Logic Board
• I/O Board
• Wireless Card
• 1/2 of the D300 Graphics Boards [A]

The only thing left from the old machine is the processor, the RAM, the 512 SSD and 1/2 of the d300 Graphics Boards .

I'd be hard-pressed to recommend this machine. Apple's been very accommodating but, I'm skeptical about these AMD graphics cards.

 

[4God]

I have the 6 core with D500's and 32Gigs of RAM. The only issue I've had is the pink startup screen on one of my monitors (the HDMI connected one) while still on El Capitan. That was resolved with upgrading to Sierra 10.12.5. Outside of that, no complaints here. Edit, color grade 4K material with no problems.

 

[voyager77]

Can you post the syslog.log or maybe the line about GPU crash? I can't find any in my syslog...

Nothing in the syslog.log, it's to long ago now.
But this is the kernel log of the event

Thu May 18 23:10:40 2017

Event: GPU Reset
Date/Time: Thu May 18 23:10:40 2017
Application:
Path:
Tailspin: /Library/Logs/DiagnosticReports/gpuRestart2017-05-18-231040.tailspin
GPUSubmission Trace ID: 0
OS Version: Mac OS X Version 10.12.5 (Build 16F73)
Graphics Hardware: AMD FirePro D300
Signature: 40

Report Data:

GPURestartReportStart
------------------------
Hung Channels: DMA1
------------------------
[04] AccelChannel: DMA1
Pending Command from : UnknownCtx
PendingCommandTimestamp: 0x00317be8, TotalDWords: 0x00000000, GART Offset=0x0000000080369000, stamp_idx=4, estamp=0x00317be8
PendingCommandStart:
PendingCommandEnd
------------------------
[04] DMA1 HWChannel : Enabled, NotIdle
LastReadTimestamp : 0x00317be7
NextSubmitTimestamp : 0x00317c63
[DMA1, ts:0x00317be8]: No semaphore wait
[DMA1, ts:0x00317be8]: semaphore signal: 0xffffff80354fddc0
HWSemaphore 0xffffff80354fddc0 Signal Event: [DMA1 channel[4] TS:0x00317c62 lastReadTS:0x00317be7]
------------------------
[0] PFP_HEADER_DUMP: 0xc0004200 0xc0013900 0xc0013900 0xc0053c00 0xc0034300
[0] ME_HEADER_DUMP : 0xc0013900 0xc0053c00 0xc0034300 0x00002010 0x0000a2a4
[1] PFP_HEADER_DUMP: 0xfbfffffe 0x2ecd0c81 0xcc5934c9 0x2d1453be 0x2c5f6cdc
[1] ME_HEADER_DUMP : 0xcaf3dc3f 0x2f0ddbfd 0xf6f4751f 0x90cfbff7 0xd1f93d68
[2] PFP_HEADER_DUMP: 0xc0013900 0xc0044702 0xc0001700 0xc0051000 0xc0034302
[2] ME_HEADER_DUMP : 0xc0013900 0xc0044702 0xc0001700 0xc0034302 0xc0004602
------------------------
SRBM_CNTL : 0x000003ff | CP_ME_CNTL : 0x15000000
SRBM_STATUS: 0x200000c0 | SRBM_STATUS2: 0x00000040
GRBM_STATUS: 0x00003028 | GRBM_STATUS2: 0x00000000
------------------------
GPU HangFlags 0x00000005: AsicHangState 0x00000005, AsicResetRequirement 0x00000001
IndividualEngineHang: 1
NonEngineBlockHang : 0
FenceNotRetired : 1
PerEngineReset : 1
FullAsicReset : 0
HangEngineBitmap : 0x00000040
Hung Engines Detected by Cail: DRMDMA1/SDMA
------------------------
AMDPitcairnGraphicsAccelerator PCIe Device: [6:0:0] State: ENABLED
Configuration: deviceBits: 0x50116810, capabilityBits: 0x30840300
TotalVideoRAMBytes: 0x0000000080000000 (2147483648)
------------------------
[00] PM4 Engine : Enabled, Idle
[00] GFX Channel: Enabled, Idle, lastReadTimestamp 0x04be249b, nextSubmitTimestamp 0x04be24ef
[01] C0 Channel: Enabled, Idle, lastReadTimestamp 0x00000000, nextSubmitTimestamp 0x00000001
[02] C1 Channel: Enabled, Idle, lastReadTimestamp 0x00000338, nextSubmitTimestamp 0x00000339
------------------------
[01] SDMA Engine : Enabled, Idle
[03] DMA0 Channel: Enabled, Idle, lastReadTimestamp 0x00048f54, nextSubmitTimestamp 0x00048f62
------------------------
[02] SDMA Engine : Enabled, NotIdle
[04] DMA1 Channel: Enabled, NotIdle, lastReadTimestamp 0x00317be7, nextSubmitTimestamp 0x00317c63
------------------------
[03] UVD Engine : Disabled, Idle
[05] UVD Channel: Disabled, Idle, lastReadTimestamp 0x00000000, nextSubmitTimestamp 0x00000001
------------------------
[04] SAMU Engine : Disabled, Idle
[06] SPU Channel: Disabled, Idle, lastReadTimestamp 0x00000000, nextSubmitTimestamp 0x00000001
------------------------
[05] VCE Engine : Disabled, Idle
[07] VCE Channel: Disabled, Idle, lastReadTimestamp 0x00000000, nextSubmitTimestamp 0x00000001
[08] VCEL Channel: Disabled, Idle, lastReadTimestamp 0x00000000, nextSubmitTimestamp 0x00000001
------------------------

 

[CC88]

Do you have the exact time the crash occured? Otherwise, that's a lot of code to go through.

Yes I have. Was 19.05.
[doublepost=1495257419][/doublepost]

Nothing in the syslog.log, it's to long ago now.
But this is the kernel log of the event

Can you tell me where I can find it? I'm running a nMP on Sierra 10.12.5.

Thanks a lot.

 

[voyager77]

Can you tell me where I can find it? I'm running a nMP on Sierra 10.12.5.

Thanks a lot.

Open console and look under System reports.
It's named Kernel.......
At least that's where mine was hiding, yours may be different.