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 Mac Pro 2,1 Random Restart Overtemp

M

Maanghod

macrumors newbie
Original poster
Sep 14, 2017
9
0
Hi guys,

I am having problems with a 2007 Mac Pro 2,1.

The symptoms are the following: Random restart. Cannot actually boot again (restarts before mac os finishes loading) unless you let it sit for 20-30 minutes. Sometimes the CPU OVERTEMP leds illuminate after. Sometimes they don’t. Either way, fans do not throttle up before crashing! They never throttle up, they stay at 500. Sometimes one of the DIMM leds on one of the RAM plates illuminates. Very rarely.
After failure northbridge heatsink feels very hot. RAM hot as well. CPU heatsink not hot.

Reported temps: Initially: CPUs never above 60c. It only restarted when Northbridge heatsink went up to 75-80c. Ram 75c as well. Very hot to touch.

Now: Now it even restarts when northbridge heatsink is at 55c. But you still have to wait 30 mins before it can actually boot without crashing.

There must be something overheating.

What I tried: Every possible RAM configuration. Removing the DIMM that had illuminated. Replaced thermal paste on northbridge and southbridge. Another GPU. Another OS X. SMC/PRAM reset.

What works: SMCFanControl worked for 2 weeks with front fans set at 1000 RPM. Then it started crashing again. Now I can only keep it from crashing with fans at 1800 RPM. So the problem seems to be evolving.

I don’t know what to try anymore. Please help me.

Thank you!
 
Last edited:
Maanghod said:
Hi guys,

I am having problems with a 2007 Mac Pro 2,1.

The symptoms are the following: Random restart. Cannot actually boot again (restarts before mac os finishes loading) unless you let it sit for 20-30 minutes. Sometimes the CPU OVERTEMP leds illuminate after. Sometimes they don’t. Either way, fans do not throttle up before crashing! Sometimes one of the DIMM leds on one of the RAM plates illuminates. Very rarely.
After failure northbridge heatsink feels very hot. RAM hot as well. CPU heatsink not hot.

Reported temps: Initially: CPUs never above 60c. It only restarted when Northbridge heatsink went up to 75-80c. Ram 75c as well. Very hot to touch.

Now: Now it even restarts when northbridge heatsink is at 55c. But you still have to wait 30 mins before it can actually boot without crashing.

There must be something overheating.

What I tried: Every possible RAM configuration. Removing the DIMM that had illuminated. Replaced thermal paste on northbridge and southbridge. Another GPU. Another OS X. SMC/PRAM reset.

What works: SMCFanControl worked for 2 weeks with front fans set at 1000 RPM. Then it started crashing again. Now I can only keep it from crashing with fans at 1800 RPM. So the problem seems to be evolving.

I don’t know what to try anymore. Please help me.

Thank you!
Click to expand...
Ok, now it started rebooting with the northbridge at 45c (fans on 1800), so it might not be the northbridge. As I said, all the temps are on point. But I still feel that something is overheating because you have to wait a bit before you can boot again. But I cannot figure out what is overheating!
 
What about the power supply? I would suspect that too after 13 years of use. Maybe try cleaning it up from dust and try again.
 
I had one of these before it got stolen - it was an absolute joy to own. Would love to see this resolved. Few questions:

  1. When you try reboot after a shutdown, what happens?
  2. Can you try boot from a known good (even fresh) boot of El Capitan? (if I recall this was the latest OS supported by the hardware). What happens?
  3. You can use Console to review the reason for the restart or Shutdown. What does this report? It may provide more of a clue than the logic board LEDs?
  4. As suggested by mikas, I would take a can of compressed air and a vacuum to this entire unit
  5. Have you got access to ASD for this unit? I'd minimise what was connected to the MLB and run the test
Good luck!
 
mikas said:
What about the power supply? I would suspect that too after 13 years of use. Maybe try cleaning it up from dust and try again.
Click to expand...

Hi, thank you for your suggestion. I didn’t actually clean the power supply, but it never reports more than 43c, so I don’t think it could be overheating, unless there’s so much dust that it’s shorting it...
 
Ok. Might be a bad capacitor too in there. You don't happen to have an extra one to try?
 
FireArse said:
I had one of these before it got stolen - it was an absolute joy to own. Would love to see this resolved. Few questions:

  1. When you try reboot after a shutdown, what happens?
  2. Can you try boot from a known good (even fresh) boot of El Capitan? (if I recall this was the latest OS supported by the hardware). What happens?
  3. You can use Console to review the reason for the restart or Shutdown. What does this report? It may provide more of a clue than the logic board LEDs?
  4. As suggested by mikas, I would take a can of compressed air and a vacuum to this entire unit
  5. Have you got access to ASD for this unit? I'd minimise what was connected to the MLB and run the test
Good luck!
Click to expand...

Hi,

Thank you for your answer.

1. It automatically restarts, but cannot finish booting without restarting again. In this case if I then shut it down from the power button and then turn it back on, it still fails. If I shut it down or restart it normally and start it again, it works fine.

2. I tried booting from a modded El Capitan, which is not actually supported on this machine, it behaves like Lion, meaning it restarts after some time.

3. Console says previous shutdown cause 3, which is simply bad shutdown.

4. I did do that, I also replaced the thermal paste on northbridge and southbridge. I cannot replace CPU thermal paste because the heatsink has fused to the CPU... either way, the temperatures for the CPU are ok.

5. I also tried booting from Apple Hardware Test. It restarts mid test because of heat I think. Northbridge gets hot hot hot. I was able to complete the test with fans on manual. No faults found. Also ran Apple Service Diagnostics. Same story. No faults found.

Thanks.
[automerge]1588780066[/automerge]
Two thoughts:

1) the northbridge heatsink gets hot hot hot to touch. at that moment it restarts immediately. but the fans do not speed up at all until it restarts, all the way to 75-80c. Should not the fans speed up? Could the SMC firmware become corrupted? Can I reinstall it?

2) I can reduce the northbridge heatsink temperature by manually setting fans on high. This makes it last for a while. However, sometimes (very rarely), it restarts with the heatsink at barely 45c. Could the northbridge have been damaged by heat causing it now to misfire even at low temperatures?
 
When was the last time all the thermal paste was replaced on all chips with heatsinks with this machine ?

It's over a decade old . Don't worry about onboard diagnostic LEDs and ASD reports . She's overheating .

Remove the old , dried factory paste and re-apply a new layer of paste on both CPUs , the Southbridge controller , the Northbridge controller and even the GPU , if that's stock . None of this is hard to do and all heatsinks are removable ( although they might stick a bit ) .

Buy a couple tubes of Arctic MX4 thermal paste .
Obtain an anti-static black stick , to apply new paste and remove the old . I like the fine control this provides .
Some higher purity ( 70 - 100 percent ) isopropyl alcohol as a solvent .
Wooden stick cotton swabs to remove paste .
Drivers for heatsink removal .

That's all you need .

Shoot me a message if you need help with the procedures , as I've done this many times . You'll get a new Mac when you're done .

One word of warning .

Most of the Mac Pro 2,1 ( 2007 ) 3.0 GHz Eight Core ( the only factory configuration ) and all Mac Pro 3,1 ( 2008 ) 3.2 GHz Eight Core factory models used a dangerous CPU thermal paste seal called Krytox . It is dangerous to your health if you are exposed to heated Krytox in an enclosed environment . It begins to shut down your lungs . So , let your Mac cool for an hour before removing the CPU heatsinks . It's safe when cool . You'll know Krytox has been used in a System when you see the distinctive shiny silver metallic crystal deposits on the thermal areas . This is what old Krytox looks like :

IMG_0891.JPG



In the old days , there were warnings not to let these thermal materials be exposed to air for more than 30 minutes , as the paste / seal would denature and harden . These days , the thermal materials are so old they have dried out naturally anyways .

It is also very difficult to remove baked-on Krytox from thermal areas of the processors and heatsinks . You'll probably need to grind it off . It'll laugh at all your solvents .
 
  • Like
Reactions: Maanghod
Snow Tiger said:
When was the last time all the thermal paste was replaced on all chips with heatsinks with this machine ?

It's over a decade old . Don't worry about onboard diagnostic LEDs and ASD reports . She's overheating .

Remove the old , dried factory paste and re-apply a new layer of paste on both CPUs , the Southbridge controller , the Northbridge controller and even the GPU , if that's stock . None of this is hard to do and all heatsinks are removable ( although they might stick a bit ) .

Buy a couple tubes of Arctic MX4 thermal paste .
Obtain an anti-static black stick , to apply new paste and remove the old . I like the fine control this provides .
Some higher purity ( 70 - 100 percent ) isopropyl alcohol as a solvent .
Wooden stick cotton swabs to remove paste .
Drivers for heatsink removal .

That's all you need .

Shoot me a message if you need help with the procedures , as I've done this many times . You'll get a new Mac when you're done .

One word of warning .

Most of the Mac Pro 2,1 ( 2007 ) 3.0 GHz Eight Core ( the only factory configuration ) and all Mac Pro 3,1 ( 2008 ) 3.2 GHz Eight Core factory models used a dangerous CPU thermal paste seal called Krytox . It is dangerous to your health if you are exposed to heated Krytox in an enclosed environment . It begins to shut down your lungs . So , let your Mac cool for an hour before removing the CPU heatsinks . It's safe when cool . You'll know Krytox has been used in a System when you see the distinctive shiny silver metallic crystal deposits on the thermal areas . This is what old Krytox looks like :

View attachment 912675


In the old days , there were warnings not to let these thermal materials be exposed to air for more than 30 minutes , as the paste / seal would denature and harden . These days , the thermal materials are so old they have dried out naturally anyways .

It is also very difficult to remove baked-on Krytox from thermal areas of the processors and heatsinks . You'll probably need to grind it off . It'll laugh at all your solvents .
Click to expand...
Hi, thanks for your answer and for the warning. Pretty scary stuff.

It’s funny because that’s exactly what I did. I replaced the thermal paste on the northbridge and southbridge using Arctic MX4. I also tried another known good graphics card.

I also wanted to redo the thermal paste on the cpus, but they are fused to the heatsink. Thermal paste turned to glue. It seems that if I force it I’m gonna break the logic board! How could I remove the heatsink when it is fused to the cpu? This is the last thing that remains for me to try...

Finally, it is also my opinion that something is overheating. But why do the fans not speed up? They stay at 500 rpm until it crashes and I can watch the temperature increase by one centigrade every 30 seconds. It seems like it’s not even trying to stop itself from overheating!
Could the temperature sensors be bad, underreporting?
Thanks
 
Last edited:
I do believe that the cpus are overtemp because the heatsinks are always so cold! Possibly the thermal paste is done for. But I can’t figure out how to remove the heatsink since it feels glued to the cpu... I found that this is a common problem with the thermal paste turning to glue.
 
Maanghod said:
Hi, thank you for your suggestion. I didn’t actually clean the power supply, but it never reports more than 43c, so I don’t think it could be overheating, unless there’s so much dust that it’s shorting it...
Click to expand...

It's a really good idea to clean the PSUs on an occasional basis with a compressor , as dusty power supplies are a leading reason for System failure . Use 60 PSI at 6 inches and make certain everything dries thoroughly before you power up the System again , because it's normal that there are water particles in the air hose . Try to use a water trap . I remove and clean my PSUs . And let my PSUs dry for at least 24 hours after cleaning , before re-installation .
[automerge]1588799775[/automerge]
Maanghod said:
I do believe that the cpus are overtemp because the heatsinks are always so cold! Possibly the thermal paste is done for. But I can’t figure out how to remove the heatsink since it feels glued to the cpu... I found that this is a common problem with the thermal paste turning to glue.
Click to expand...

What's the delta temperature between a CPU and its corresponding heatsink , at idle ?
 
Last edited:
Maanghod said:
Hi, thanks for your answer and for the warning. Pretty scary stuff.

It’s funny because that’s exactly what I did. I replaced the thermal paste on the northbridge and southbridge using Arctic MX4. I also tried another known good graphics card.

I also wanted to redo the thermal paste on the cpus, but they are fused to the heatsink. Thermal paste turned to glue. It seems that if I force it I’m gonna break the logic board! How could I remove the heatsink when it is fused to the cpu? This is the last thing that remains for me to try...

Finally, it is also my opinion that something is overheating. But why do the fans not speed up? They stay at 500 rpm until it crashes and I can watch the temperature increase by one centigrade every 30 seconds. It seems like it’s not even trying to stop itself from overheating!
Could the temperature sensors be bad, underreporting?
Thanks
Click to expand...

Could be the Xeons' built in automatic safeties are kicking in very quickly , because of an overtemp situation . You won't get thermal reports showing overheating in that situation , because the lag in reporting is too long .

If you have exhausted all other options and you need to remove an adhered CPU Heatsink from it's CPU , remove the memory cage , the processor heatsink cover and the front fan assembly . Loosen the CPU heatsink hex fasteners with your driver as much as you can . Then cup both sides of the topmost heatsink with both your hands and gently press first one way laterally and then the other way . Alternate this process until the heatsink is removed . There is the risk you'll damage the processor socket doing this operation , in which case you're looking at a logic board replacement . If you can remove the top heatsink , proceed to remove the bottom heatsink .

Worse case scenario is that you damage the processor socket but are unable to remove the CPU heatsinks . In that case , if you power up the System , you will damage more than the logic board .

If you are successful in removing both CPU heatsinks , remove the processors and very closely examine both socket pin arrays for any bent pins . You do not want so much as a single bent pin . I would have ready a pair of LGA771 socket protection covers before you attempt to remove the heatsinks , since if you are successful you want to protect both unpopulated sockets as soon as possible .

You should back up all data first from the System in case something goes horribly wrong .

Before you ask , I have successfully performed the above operation more than once and hated every second of it . The feeling of imminent disaster never left me .
 
Snow Tiger said:
Before you ask , I have successfully performed the above operation more than once and hated every second of it . The feeling of imminent disaster never left me .
Click to expand...

I recall upgrading the CPUs on 2.66GHz MP 1,1 to the top CPUs available for the socket. That feeling was so bad, I got one of the best Mac Genius' with me in my front room as he watched me do this.

I hope the op gets this going - its our little COVID project ;)
 
Agree, thermal paste sound likeiy needs reapp. I would also check the northbridge thermal tape and hold down rivets.
 
Last edited:
Hi guys, sorry to have gone missing. So, to recap, I reapplied thermal paste to northbridge and southbridge with no difference and the heatsinks have fused to the cpus so I can’t really redo that. Anyway, the temp difference between cpu diode and heatsink is always less than 7 degrees, mostly 5, so it’s probably still efficient.

BUT! I did this funny air diffuser tunnel and, sure enough, it works?! (The idea was to get more air to the motherboard) It just does not crash anymore and the northbridge is 10 degrees cooler at the same fan rpm... I keep both front fans at 1200.

I’m not sure what’s actually happening, but it works. Results are results :D
 

Attachments

  • 459A7641-F7E4-4775-959D-890B28375DC6.jpeg
    459A7641-F7E4-4775-959D-890B28375DC6.jpeg
    359.5 KB · Views: 355
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