Become a MacRumors Supporter for $25/year with no ads, private forums, and more!

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
I have this Mac Pro 2013 which crashes with Kernel panic first when i turn it on then boots and works as intended....
Strange, no matter what test i run all seems good, replaced the CPU does the same, replaced the Flash same thing.
The GPU D700 runs without artifacting or crashing or overheating.
Memory was replaced same issue.
After a couple reboots the problem goes away strangly,.. like it would warm up or something:)
but if you leave it rest for a 20 mins it boots with kernel panic first then works just fine again... although it does panic a couple times more if you turn it of and on again before reaches warmed up state:)...

Any suggestion? Clean install already done 3 times by now Mojave latest, cleared the pram nvram...

Log is the following as pictured.
 

Attachments

  • 2019-11-07 14.30.29.jpg
    2019-11-07 14.30.29.jpg
    868.3 KB · Views: 191
  • 2019-11-07 14.30.18.jpg
    2019-11-07 14.30.18.jpg
    874.8 KB · Views: 110
  • 2019-11-07 14.30.06.jpg
    2019-11-07 14.30.06.jpg
    887.7 KB · Views: 114
  • 2019-11-07 14.29.53.jpg
    2019-11-07 14.29.53.jpg
    963 KB · Views: 108
  • 2019-11-07 14.29.36.jpg
    2019-11-07 14.29.36.jpg
    868.4 KB · Views: 131
  • 2019-11-07 14.28.37.jpg
    2019-11-07 14.28.37.jpg
    457.3 KB · Views: 94

netkas

macrumors 65816
Oct 2, 2007
1,179
336
either overheating or card is dying:

possible way to workaround:

1) increase fan speed with some app
2) if won't help - try to disable second card (opencl-only one) by modding AMDRadeonX4000.kext's Info.plist
Before Doing so It might have a sense to remove AMDRadeonX4000.kext to see if disabling acceleration will help at all
3) if main card ( with displays) is dead your only way is egpu.
 

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
either overheating or card is dying:

possible way to workaround:

1) increase fan speed with some app
2) if won't help - try to disable second card (opencl-only one) by modding AMDRadeonX4000.kext's Info.plist
Before Doing so It might have a sense to remove AMDRadeonX4000.kext to see if disabling acceleration will help at all
3) if main card ( with displays) is dead your only way is egpu.
in heaven it does not go over 56C the GPU and no crashes, does that only runs 1 GPU maybe?
It is strange that the Monitor attached to the second GPU and not the first, does that matter?
I use the HDMI port.
 

netkas

macrumors 65816
Oct 2, 2007
1,179
336
on macpro 6,1 all of the monitors are attached to second card, first one is used only for opencl workloads, so it can be disabled if it's the cause of errors

looking only to temp of the gpu is not right, there is other elements between gpu and cpu that can overheat, like pcie switch,
 

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
on macpro 6,1 all of the monitors are attached to second card, first one is used only for opencl workloads, so it can be disabled if it's the cause of errors

looking only to temp of the gpu is not right, there is other elements between gpu and cpu that can overheat, like pcie switch,
and why the probelm only exist when I boot the unit cold,? and after that it disappears if I use it for a few minutes? Very strange. Any way to test both video card?
 

tsialex

macrumors G3
Jun 13, 2016
9,511
10,253
One possibility is solder micro fractures are developing somewhere, when it's cold the solder is contracted and has fracture/no contact between two points, when it's warm the solder expands and now you have contact.

This weird problem started to happen when lead (Pb) was banned from the composition of the solder, RoHS/lead free solder, the first formulations were somewhat susceptible to this.
 

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
One possibility is solder micro fractures are developing somewhere, when it's cold the solder is contracted and has fracture/no contact between two points, when it's warm the solder expands and now you have contact.

This weird problem started to happen when lead (Pb) was banned from the composition of the solder, RoHS/lead free solder, the first formulations were somewhat susceptible to this.
by any chancve I can find this soldering point?
 

tsialex

macrumors G3
Jun 13, 2016
9,511
10,253
that i have none:)
This type of repair is not cost effective for mass produced goods of low value, like MP6,1 parts, the total cost of the investigative process will be higher than just replacing. Industry normally use this to find problems in prototypes submitted to test runs or extremely high value goods that are not easily replaced.

Some repair shops use optical microscopy and some optical filters to do the repair process, UV lights too, but you can't do that to BGA parts.
 

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
This type of repair is not cost effective for mass produced goods of low value, like MP6,1 parts, the total cost of the investigative process will be higher than just replacing. Industry normally use this to find problems in prototypes submitted to test runs or extremely high value goods that are not easily replaced.

Some repair shops use optical microscopy and some optical filters to do the repair process, UV lights too, but you can't do that to BGA parts.
sad:(
 

ActionableMango

macrumors G3
Sep 21, 2010
9,606
6,868
If you've done a true clean install, (wiped first, not just installed over existing installation), then software is out of the picture and the problem is likely hardware. The KP log sure points the finger at GPUs, and this also fits with the poor reputation of the GPUs in this model. Furthermore, it sounds like you've replaced all of the other hardware components anyway.

Whether you are under warranty or not, I'd try to get a new GPU out of Apple. Apple knows there is a problem. They had an extended repair campaign for 6,1 GPUs. In the big thread about 6,1 GPU problems there are people who got GPUs replaced by Apple even out of warranty. There was also a class-action lawsuit about 6,1 GPUs, but I lost track of what happened to that.
 

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
If you've done a true clean install, (wiped first, not just installed over existing installation), then software is out of the picture and the problem is likely hardware. The KP log sure points the finger at GPUs, and this also fits with the poor reputation of the GPUs in this model. Furthermore, it sounds like you've replaced all of the other hardware components anyway.

Whether you are under warranty or not, I'd try to get a new GPU out of Apple. Apple knows there is a problem. They had an extended repair campaign for 6,1 GPUs. In the big thread about 6,1 GPU problems there are people who got GPUs replaced by Apple even out of warranty. There was also a class-action lawsuit about 6,1 GPUs, but I lost track of what happened to that.
the repair program finished in the end of 2018, and yes I did all above I even repartitioned the Flash drive and the only thing I had no option to replace is the GPU and motherboard.

Frustrating as the unit works just fine after i manage to boot it and use it for a few minutes, up until i turn it off... Seems to me that the main GPU running the monitor works but the secondary fails during the boot hardware checks, anyway thanks for all the info.
 

ActionableMango

macrumors G3
Sep 21, 2010
9,606
6,868
the repair program finished in the end of 2018, and yes I did all above I even repartitioned the Flash drive and the only thing I had no option to replace is the GPU and motherboard.

Frustrating as the unit works just fine after i manage to boot it and use it for a few minutes, up until i turn it off... Seems to me that the main GPU running the monitor works but the secondary fails during the boot hardware checks, anyway thanks for all the info.

I should elaborate. People in that thread reported getting a new GPU from Apple even though they were out of warranty and not covered by the extended campaign. They didn't even have a model that was covered by the campaign (D300). So it seems to be worth a try, if you have the time and willpower to open a ticket and push forward.

I could see someone shrugging their shoulders and just living with it, because I've done that myself for similar problems in the past. Especially when I'm busy with life problems. But I also feel like the failure could worsen at any unknown random time.
 

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
I should elaborate. People in that thread reported getting a new GPU from Apple even though they were out of warranty and not covered by the extended campaign. They didn't even have a model that was covered by the campaign (D300). So it seems to be worth a try, if you have the time and willpower to open a ticket and push forward.

I could see someone shrugging their shoulders and just living with it, because I've done that myself for similar problems in the past. Especially when I'm busy with life problems. But I also feel like the failure could worsen at any unknown random time.
do you have the thread?
 

ActionableMango

macrumors G3
Sep 21, 2010
9,606
6,868
do you have the thread?

Yes, I believe this is the one:

I know the subject line says "driver issues", but that's misleading because it's not what things turned out to be.
 

bsbeamer

macrumors 601
Sep 19, 2012
4,040
2,148
Absolutely talk to your LOCAL Apple Store if you have one. It's hit/miss with success, but many will straight up repair or replace if they have the parts on hand.
 

MacZoltan

macrumors member
Original poster
May 18, 2016
92
7
Absolutely talk to your LOCAL Apple Store if you have one. It's hit/miss with success, but many will straight up repair or replace if they have the parts on hand.
will check if I have the original receipt, as it was purchased used long time ago
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.