Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MP 1,1-5,1 System crash when gaming

D

dontpokebearz

macrumors regular
Original poster
Feb 16, 2018
155
108
Maine
Hi all,

My 4,1 Mac Pro seems to be crashing when playing any games - macOS or Windows. When opening a game (CS:GO, Fortnite, Valheim) the game opens and loads to the main menu, sometimes into the game - and then the computer will crash. Screens go black, fans spin up and that's it. I can't do anything except for a hard restart. Audio, however still plays. When looking at the attached GPU crash log I am seeing 'GPU RESET' as an error.

My Specs:
  • Mac Pro 4,1 (Flashed 5,1)
  • 2x 5680
  • AMD Radeon RX580 Sapphire Pulse 8GB (No Pixlas, just using dual 4 -> 8 pin)
  • 8GB DDR3 RAM
  • SATA SSD
  • Mojave + latest OpenCore w/ Radeon Boost enabled.
  • Windows 10 on a separate drive.
Temperatures seem to be fine, but I could be wrong. I have a few theories as to what the issue may be:
  1. An issue with OpenCore (however, this is unlikely as this also is an issue with Windows)
  2. My PSU is failing
  3. I have the wrong BIOS on the GPU. I flashed what I believed to be the stock BIOS, but Techpowerup doesn't make it 100% foolproof that you have the right one.
I did a Unigin Heaven test for an hour - GPU temps never went above 83 degrees Celsius and CPU temps were fine.

Any ideas? has anyone run into anything similar?
 

Attachments

  • Crash.txt
    1.7 KB · Views: 129
My RX580 had a wonky BIOS that would cause a crash AFTER gaming. Never during, always after. That would be my first guess, as GPU irregularities are often the cause of bad behavior.

PSU would be my second guess.
 
  • Like
Reactions: dontpokebearz
dontpokebearz said:
Hi all,

My 4,1 Mac Pro seems to be crashing when playing any games - macOS or Windows. When opening a game (CS:GO, Fortnite, Valheim) the game opens and loads to the main menu, sometimes into the game - and then the computer will crash. Screens go black, fans spin up and that's it. I can't do anything except for a hard restart. Audio, however still plays. When looking at the attached GPU crash log I am seeing 'GPU RESET' as an error.

My Specs:
  • Mac Pro 4,1 (Flashed 5,1)
  • 2x 5680
  • AMD Radeon RX580 Sapphire Pulse 8GB (No Pixlas, just using dual 4 -> 8 pin)
  • 8GB DDR3 RAM
  • SATA SSD
  • Mojave + latest OpenCore w/ Radeon Boost enabled.
  • Windows 10 on a separate drive.
Temperatures seem to be fine, but I could be wrong. I have a few theories as to what the issue may be:
  1. An issue with OpenCore (however, this is unlikely as this also is an issue with Windows)
  2. My PSU is failing
  3. I have the wrong BIOS on the GPU. I flashed what I believed to be the stock BIOS, but Techpowerup doesn't make it 100% foolproof that you have the right one.
I did a Unigin Heaven test for an hour - GPU temps never went above 83 degrees Celsius and CPU temps were fine.

Any ideas? has anyone run into anything similar?
Click to expand...
I have the same problems with my vega 56.

During gaming it'll either:

1. Black screen, fans on full (I think it's the GPU fans rather than the system), which I think suggests thermal issue with the vega.
2. BSOD with thread_stuck warning, indicating driver issue.
3. Complete freeze on screen, no audio, indicating driver issue.
4. Artifacts on screen, again, driver issue (only had this once).
5. Crash to desktop, due to driver resetting.

I've tried official AMD, AMD PRO, bootcampdrivers red, and bootcampdrivers blue drivers and they have the same problems. At this point I can just about feel a crash coming when the FPS suddenly drops.

On bootcampdrivers red at the moment, with low power bios, and a slight undervolt with fan curve to keep the thermals lower and it feels OK for now.

I've also done the following tweaks:

1. Disable ULPS
2. Enable TDRdelay to allow the driver time to recover.
3. In AMD radeon, set state 1 to minimum state for core and memory.

I power mine via a secondary PSU so I think my mac psu us fine.

Edit; the only other thing I could think of is a dud GPU......

I would like to water-cool mine to see if it's the thermals but I can't justify the cost.

Edit 2:

Have to say it's rock solid during everything else but gaming so I'm inclined to think it's a thermal issue coupled with drivers being dumb.

Stress testing didn't pick anything up.

Also, I can game for between 30 to 60 minutes before I see any crashing
 
Last edited:
eicca said:
My RX580 had a wonky BIOS that would cause a crash AFTER gaming. Never during, always after. That would be my first guess, as GPU irregularities are often the cause of bad behavior.

PSU would be my second guess.
Click to expand...
Thanks for response - I've flashed the BIOS twice now, with what I believe is the stock BIOS.
 
the bios switch (if present) is set to gaming?

If there is no switch show the Stickers of the card for reference.
 
This is typical AMD behaviour and should make you question why Apple still uses their crappy GPUs instead of Nvidias. Look up 'amd black screen' and you're going to get a ton of results.

For the record; I've had about 8 different GPUs in the last couple of years: Asus GTX 660/960/970/980/1080 (All Strix) Sapphire 5700 XT Nitro+, Pulse 5700 and 580 Nitro+ (4gb). They've all ran fine EXCEPT for the 5700s, and ESPECIALLY the 5700 XT: black screens are common, also happen either before a game loads or after it's finished; never during.

I'm on a Hackintosh and I've swapped motherboards, PSU, RAM, HDD and SSD, updated motherboard bios, cleared bios, no overclocking, underclocking, etc. Nothing changed the behaviour. AMD's are unreliable trash. Not a single Nvidia I've ever used had any issue... not a single crash with any of them either.

I wouldn't wish the problems I've had with my Radeons on my worst enemy.
 
blazerunner said:
This is typical AMD behaviour and should make you question why Apple still uses their crappy GPUs instead of Nvidias. Look up 'amd black screen' and you're going to get a ton of results.

For the record; I've had about 8 different GPUs in the last couple of years: Asus GTX 660/960/970/980/1080 (All Strix) Sapphire 5700 XT Nitro+, Pulse 5700 and 580 Nitro+ (4gb). They've all ran fine EXCEPT for the 5700s, and ESPECIALLY the 5700 XT: black screens are common, also happen either before a game loads or after it's finished; never during.

I'm on a Hackintosh and I've swapped motherboards, PSU, RAM, HDD and SSD, updated motherboard bios, cleared bios, no overclocking, underclocking, etc. Nothing changed the behaviour. AMD's are unreliable trash. Not a single Nvidia I've ever used had any issue... not a single crash with any of them either.

I wouldn't wish the problems I've had with my Radeons on my worst enemy.
Click to expand...

If I understand your post right , the only issues you had were with the 5700's.
This doesnt necessarily imply all AMD cards are crap i'd say.

Comparing AMD and Nvidia is not really helping anybody here (or is it complaining?).
Neither is a Hackintosh a reference , when troubleshooting issues on a MP4,1/5,1

Just my point of view.
 
Last edited:
KeesMacPro said:
If I understand your post right , the only issues you had were with the 5700's.
This doesnt necessarily imply all AMD cards are crap i'd say.

Comparing AMD and Nvidia is not really helping anybody here (or is it complaining?).
Neither is a Hackintosh a reference , when troubleshooting issues on a MP4,1/5,1

Just my point of view.
Click to expand...
The black screens go well beyond the 5700s, as is clearly stated by OP... the whole point of this thread. AMD has serious issues with this.

Hackintosh or not, it's irrelevant, the point is it happens in OS X on other reliable systems as well. So it's definitely a GPU problem that AMD doesn't give a damn to fix or even address.
 
I'm currently using the Mac, so I'll post the SKU soon.

In regards to AMD's black screens - I haven't heard of these issues. I'm not saying it's not an issue or they don't exist, but you'd think it would be more known.

Is their a way to test my power supply? I'm wondering if my PSU is failing when the computer needs to draw more power.
 
dontpokebearz said:
Is their a way to test my power supply? I'm wondering if my PSU is failing when the computer needs to draw more power.
Click to expand...
AFAIK not , or not an easy way with e.g. a multimeter.
If possible I'd test a PSU from another MP 4,1 or 5,1.

Just noticed something (post #1):
Are you sure there's only 8GB RAM installed?

If so, in combination with a (slow) SATA SSD I guess there's not much performance to expect
 
Last edited:
dontpokebearz said:
I'm currently using the Mac, so I'll post the SKU soon.

In regards to AMD's black screens - I haven't heard of these issues. I'm not saying it's not an issue or they don't exist, but you'd think it would be more known.

Is their a way to test my power supply? I'm wondering if my PSU is failing when the computer needs to draw more power.
Click to expand...
OCCT in Windows has PSU stress test.

Of course, it can't show you the voltage / current etc. But it's a good enough test for system stability (under high power demand).
 
h9826790 said:
OCCT in Windows has PSU stress test.

Of course, it can't show you the voltage / current etc. But it's a good enough test for system stability (under high power demand).
Click to expand...

I'll give that a go.

Sorry for the delayed response - I haven't had much time to test anything.
 
If it were a PSU issue I'd expect the machine to reboot or shut off completely not sit there still playing audio.?
 
  • Like
Reactions: pertusis1
Hi, I'm going to update on this issue - found my culprit.

My GPU thermal paste was hard as a rock. Scraped it off, cleaned the die and re-applied new paste. All is working.

I've tested this on macOS and Windows on my 2009 Mac Pro, as well as a friends Intel based custom-built rig.

If you have this issue, check the thermal paste. Note that removing the heat sync will require reapplication of paste, even if it's not dried out.

EDIT: GPU Thermal paste. In case any others have this issue.
 
Last edited:
  • Like
Reactions: h9826790
dontpokebearz said:
Hi, I'm going to update on this issue - found my culprit.

My thermal paste was hard as a rock. Scraped it off, cleaned the die and re-applied new paste. All is working.

I've tested this on macOS and Windows on my 2009 Mac Pro, as well as a friends Intel based custom-built rig.

If you have this issue, check the thermal paste. Note that removing the heat sync will require reapplication of paste, even if it's not dried out.
Click to expand...
I am curious how much the cpu core temp dropped on average as a result of this thermal paste. I am think of doing the same. At idle mine is around 50-60 C with high load it gets in high 60s - 75 C
 
I hate to sound like a broken record but you should probably repaste the Northbridge too while you’re at it. You’ll need replacement clips to do the job.
 
jscipione said:
I hate to sound like a broken record but you should probably repaste the Northbridge too while you’re at it. You’ll need replacement clips to do the job.
Click to expand...
I should have done that, but I forgot too. :(

At this point I'm not sure if I will. I love my 09 Mac Pro, but the increased frustration regarding macOS updates has pushed me to Linux, and not having proper boot pickers makes installing distributions difficult. So maybe a new computer is on the horizon. Who knows.
 
If the thermal paste on the CPUs was hard as a rock, the thermal paste on the Northbridge must be too. It’s your Mac do what you want but you should probably repaste the Northbridge. It seems as though we’re starting to get to the limits of 2009 thermal paste effectiveness.
 
jscipione said:
If the thermal paste on the CPUs was hard as a rock, the thermal paste on the Northbridge must be too. It’s your Mac do what you want but you should probably repaste the Northbridge. It seems as though we’re starting to get to the limits of 2009 thermal paste effectiveness.
Click to expand...
Sorry, I'm not sure if my post was confusing or not but this was a GPU issue, not CPU.
 
The GPU part was not clear... I guess you are off the hook then. I'm happy that you got you machine working again.
 
poundman said:
I am curious how much the cpu core temp dropped on average as a result of this thermal paste. I am think of doing the same. At idle mine is around 50-60 C with high load it gets in high 60s - 75 C
Click to expand...
This is completely normal for CPU. If you want to lower it (absolutely not required if just max at 75C), you can simply spin up the fan.
 
poundman said:
I am curious how much the cpu core temp dropped on average as a result of this thermal paste. I am think of doing the same. At idle mine is around 50-60 C with high load it gets in high 60s - 75 C
Click to expand...
The (CPU) Temp depends on different factors like e.g. Climate and single vs dual CPU, fan setup etc.
So, if your MP is dust-free and you're not living in a hot climate, CPU Temp is indeed a bit high.

Anyway, to see if a repaste is necessary check the Delta CPU diode vs CPU heatsink .
With the MP running at idle Delta should be only a few degrees Celsius.
 
Last edited:
Glad you fixed it but........ 8GB DDR3 RAM..?

If you have a single CPU tray - it will cost $17 to upgrade to 12GB [3x4GB sticks] (the bare minimum anyone should have in 2021)
If you have a double CPU tray - it will cost $22 to upgrade to 12GB [6x2GB sticks] (the bare minimum anyone should have in 2021)
 
  • Like
Reactions: KeesMacPro
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back