Mac Pro (Late 2013) GPU (Driver) Issues

[chama98]

I had to uninstall the last beta as it was crashing the computer a lot. I mean freezing the screen while doing a time machine backup. I have now installed the up to date version and will see what happens.

I noticed in the release notes that there is nothing in there about any graphics fixes.

 

[chama98]

Well that didn't last long! FFS.

 

[poematik13]

I assumed it was all glued/soldiered on nonsense that apple tends to build now days.

Thats because a) you've never owned one and b) you're influenced by the "new mac pro is a failure" narrative that manifests itself on this board by a specific 5-6 troll members that constantly bash the machine.

 

[merlinmage]

Or you are a non reflexting Apple fanboy which salutes to form over function.

 

[axantas]

It is a gorgeous machine - I did not discover any glue yet

The kind of hangs are changing here. The newest and most sympathetic version: The GUI freezes for a second, I am getting logged out - I am logging in again - that's it. The apps and my work are still running. Furthermore they are actually kind of "startup hangs", once it runs (lets say, the motor is warmed up) it runs.

 

[geeksquaddci]

Experiencing the same problem with one of our support users. Upgraded to the El Capitan 10.11.12. No changes

 

[AndyUnderscoreR]

Apple Launches Repair Program for Late 2013 Mac Pro Video Issues

https://www.macrumors.com/2016/02/06/late-2013-mac-pro-video-issues-repair-program/

 

[axantas]

As quite some of our nMP's discussed in this thread ar not manufactured between February 8, 2015 and April 11, 2015 they don't seem to be affected by that problem...
...strange repair program
At least, the finally seem to have discovered something particular that is wrong, which should help to avoid re-re-repairs.

 

[poematik13]

They repaired mine for GPU resets / kernel panics in September and I purchased it in June but the manuf. date was in April, so it lines up in my case. I'm happy that this program exists now as it gives me piece of mind and I can work without worry. And then as soon as the 2018 date comes I'm gonna sell it and get whatever new model of the nMP is out then.

 

[merlinmage]

My 6,1 with D700s was from Mid 14, still had defective GPUs after repairing them 3 times. Glad to be back on the 5,1. I guess Apple never bothered selling us POS AMD cards, you can always buy an iMac or an 7,1 nMP. Thanks Apple.

 

[ShadovvMoon]

I bought my 6,1 with D700s a few hours after the announcement and it's had no issues (aside from various crappy OS X releases). Glad they've discovered the issue with ones made in 2015 though.

 

[mauro16]

Apple Launches Repair Program for Late 2013 Mac Pro Video Issues

https://www.macrumors.com/2016/02/06/late-2013-mac-pro-video-issues-repair-program/

could somebody tell me where i find the manufacturing date?
[doublepost=1454878882][/doublepost]

could somebody tell me where i find the manufacturing date?

this site

http://www.chipmunk.nl/cgi-fast/applemodel.cgi

says mine was built on december 2014

 

[MMcCraryNJ]

.

 

[Riwam]

So what about those of us who are clearly having GPU lockups/freezes/other problems related to video, but don't have machines in the specified manufactured range? My 6c/D500 system was manufactured in October of 2014, but I'm getting all types of random freezes related to video. Playing a YouTube video in Chrome, even letting the screensaver run for a bit without forcing the machine to sleep, will cause the system to lock up on occasion. If I pull the TB cable to my display, and then plug it back in and wait for a signal, it unlocks itself, but runs very very slowly until I reboot.

I'm worried about making an appointment at a Genius Bar, because it's very sporadic. Sometimes it will lock up 2-3x per week, sometimes I won't get a lockup for a month or more. There's nothing I can seemingly do to reproduce the symptoms, and my Console logs don't show anything noteworthy (unless I am looking in the wrong places for them).

*****
Your computer certainly has a problem and you should have it checked even if it happens only sporadically as you say.
I also have a mac with a similar configuration.
While from time to time some small problem is there for a while, I do not suffer from the serious graphic faults you mention.
You should insist to have it checked even if it takes time to find the problem.
That is my modest opinion and what I would do.
Good luck!
Ed

 

[poematik13]

******
Of course.
If the user would have bought the cards you sell then they would be perfect.
A very objective and neutral opinion no doubt...

I don't know why that guy hasn't been banned yet, it's baffling. He makes problematic and argumentative posts in literally every nMP thread.

 

[ActionableMango]

http://www.chipmunk.nl/cgi-fast/applemodel.cgi

This is a great website and I've recommended it myself.

However, over the years I've noticed people checking their own known systems and seeing the content on chipmunk.nl was wrong, especially with BTO/CTO systems. I suspect that they don't have access to an actual database, but are extrapolating at least some of the data based off of the serial number.

MMcCraryNJ, if you are outside of the date range for this recall on chipmunk, it isn't 100% certain outside of the date range. If you are having problems, I'd take it in to Apple. Documenting the problems wouldn't hurt.

Intermittent problems suck. I had an intermittent knocking noise in one of my fans, but they couldn't replicate it in the store over 48 hours, so they refused to replace the fan. I had to replace the fan myself at my own expense and time, while under AppleCare warranty.

 

[edanuff]

I'm worried about making an appointment at a Genius Bar, because it's very sporadic. Sometimes it will lock up 2-3x per week, sometimes I won't get a lockup for a month or more. There's nothing I can seemingly do to reproduce the symptoms, and my Console logs don't show anything noteworthy (unless I am looking in the wrong places for them).

It's clear that machines outside of the official date range can exhibit the problem. I myself had the GPUs replaced under warranty for an nMP that was built in 2014 and the symptoms were the exact same ones Apple is conducting the repair program because of. My guess is if you bring your machine in and it's having these symptoms regularly, that they'll do the repair even if it's not in the exact date range. It also wouldn't surprise me if they expanded the date range as they've done that on previous repair programs.

 

[Infrared]

I'm surprised the D300s are not included, but this is, nevertheless, good news for some owners.

 

[axantas]

I'm surprised the D300s are not included, but this is, nevertheless, good news for some owners.

My "D300" is definitely not behaving THAT bad as Apple describes the symptoms. The strangest thing is, that the freezes changed in behavior. Initially it froze completely, so pushing the powerbutton was the only solution. Now the screen freezes from time to time and all I have to do is to unplug the TB-Monitor, reconnect it and wait a couple of seconds. I get the picture back, save my work and reboot. As I am not fond of repeated "trial and error"-repairs I will live with that. It can run a whole day without any problem. It is just doing the wrong thing at the wrong time, which includes "doing nothing" as well.

All the freezes do not only affect the video output, but do crash "everything Thunderbolt".

 

[twomartoe]

My "D300" is definitely not behaving THAT bad as Apple describes the symptoms. The strangest thing is, that the freezes changed in behavior. Initially it froze completely, so pushing the powerbutton was the only solution. Now the screen freezes from time to time and all I have to do is to unplug the TB-Monitor, reconnect it and wait a couple of seconds. I get the picture back, save my work and reboot. As I am not fond of repeated "trial and error"-repairs I will live with that. It can run a whole day without any problem. It is just doing the wrong thing at the wrong time, which includes "doing nothing" as well.

All the freezes do not only affect the video output, but do crash "everything Thunderbolt".

I have the same setup, except 6core. I am also doing the same work around. Does your console message say WindowServer: GPU Driver appears to be hung (over 5 continuous seconds of unreadiness) x 1000s when it hangs?

Have u brought it in for repair? Seriously contemplating bringing it to the apple authorised people. No official apple store where I live. I am now afraid its gonna take a couple weeks repair time and it still having the same problem or spending time telling them that there is an unpredictable hang. Unfortunately this is my main work machine.

I have a feeling apple may not know whats going on with the nmp.

 

[voyager77]

I have the same setup, except 6core. I am also doing the same work around. Does your console message say WindowServer: GPU Driver appears to be hung (over 5 continuous seconds of unreadiness) x 1000s when it hangs?

Same symptoms here, also a D300. The screen freezes, but the mouse still moves, just can't click anything. After a hard reset there's a ton of gpu driver hung messages in the log.

And mine is a november 2014 build.

 

[koyoot]

The problem looks to be more serious than faulty GPUs. It looks like it has something to do with PCIe management errors that cause problems with graphical drivers. Repair program on GPUs is only getting rid of symptoms not the source of the problem.

The question is: where the hell is the source for the problem?

 

[Riwam]

I have the same setup, except 6core. I am also doing the same work around. Does your console message say WindowServer: GPU Driver appears to be hung (over 5 continuous seconds of unreadiness) x 1000s when it hangs?

Have u brought it in for repair? Seriously contemplating bringing it to the apple authorised people. No official apple store where I live. I am now afraid its gonna take a couple weeks repair time and it still having the same problem or spending time telling them that there is an unpredictable hang. Unfortunately this is my main work machine.

I have a feeling apple may not know whats going on with the nmp.

******
We certainly live in crazy times.
While my own nMP thanks God behaves in almost 2 years reasonably OK, people depending on their computer to make a living should seriously consider having a second computer able to perform more or less OK their usual work for sad situations like this one, in which the most expensive mac surprisingly shows a serious problem.
Unfortunately not everyone can afford such an alternative.

I do not share your view that Apple does not know what is going on with the nMP since every human product is always far from perfect.
Therefore sooner or later some fault can -unfortunately- be found out.

However I believe they should supply another computer during the repair time precisely for all those who cannot wait idle until their nMP is repaired.
Since the nMP is supposed to be the right tool for professional work, such a replacement program until repair finished should be taken seriously by Apple.
If I would be in such a situation I would certainly put pressure on Apple to obtain a replacement until my mac is back.
Ed

 

[voyager77]

The problem looks to be more serious than faulty GPUs. It looks like it has something to do with PCIe management errors that cause problems with graphical drivers. Repair program on GPUs is only getting rid of symptoms not the source of the problem.

The question is: where the hell is the source for the problem?

If the problem can be solved by replacing the GPU, it must have something to do with that. That a faulty GPU will affect the PCIe bus seems likely to, and it would explain the other type of crash i get, namely a full and total system freeze, followed by a restart. That type of crash doesn't even leave a trace in the logs, since the system is so crashed it can't even write anything into the logs anymore.

 

[axantas]

I have the same setup, except 6core. I am also doing the same work around. Does your console message say WindowServer: GPU Driver appears to be hung (over 5 continuous seconds of unreadiness) x 1000s when it hangs?

Have u brought it in for repair? Seriously contemplating bringing it to the apple authorised people. No official apple store where I live. I am now afraid its gonna take a couple weeks repair time and it still having the same problem or spending time telling them that there is an unpredictable hang. Unfortunately this is my main work machine.

I have a feeling apple may not know whats going on with the nmp.

No, the most recent ones i remarked were:

Kernel_2016-02-09-133204_Mac-Pro
WindowServer_2016-02-09-133205_Mac-Pro.userspace_watchdog_timeout

The most interesting this time, there was a slight hang of everything on screen - and it recovered itself.

I am not bringing it in for service, just because I do not have something to point at, saying "do replace that". Many years ago I did that game with Apple and an clamshell iBook. On battery it just switched off - woosh. Service "did not find any error". Got it back - the error was still there - servicing it again - nothing found - got it back - error was still there. I agree with the statement here above, that Apple has absolutely no clue, what is the SOURCE of the problem. So instead of fixing it in service, they are tinkering something, finally pointing at me, saying that I am "holding it the wrong way".

Yes, I am somewhat disappointed about Apple. But as for now, the software tinkering seems to be somewhat successful in my case, because it is definitely becoming better. So I wait for another sticking plaster.

I do have driver crashes in Bootcamp as well from time to time. But they happen on "normal" PC's as well. "Driver was hanging and had to be restarted". If OS X is managing it the same way - thats ok for me.

I need that machine for working and not for having it in continuos and useless service. As Apple writes in the repair program that the D300 are "not affected", we do apparently have perfectly working machines... I have no time for fighting with Apple.

[doublepost=1455042194][/doublepost]

The problem looks to be more serious than faulty GPUs. It looks like it has something to do with PCIe management errors that cause problems with graphical drivers. Repair program on GPUs is only getting rid of symptoms not the source of the problem.

The question is: where the hell is the source for the problem?

There are not only problems with graphical drivers. All of Thunderbolt crashed in the past in case of a crash. I do have an Elgato Thunderbolt Dock V1. Everything on it was dead after a crash until reboot. Graphics itself could be recovered by unplugging and plugging the TB monitor again (stuttering picture, but ok for a proper shutdown). But devices connected to the Dock did never recover.

And: No I am not really mad - just making the best of it