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 2009 cpu and heatsink issue.

Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
H

Hershey52

macrumors member
Original poster
Dec 27, 2022
38
7
Melbourne Australia
Hi Brains Trust, Greetings from Oz.

In my mac pro both my 3.46 GHz Xeon cpus (12cores) stopped working two days ago and would NOT boot up.

So I checked and re-pasted CPU A under Heatsink A, all good, she booted fine. However, when I repasted CPU B and put it back under Heatsink B the computer would NOT BOOT at all.

So then I put cpu B under heatsink A and again, it booted fine, all good. So this confirmed that both my CPU's were working.

Now here is the kicker. When I put cpu A or B which are confirmed working, under heatsink B the computer would not boot up, no chime no nothing. Holding down Option + Command P + R did nothing. Holding in power button on 10 to 15 seconds, nothing ...grrr

So what was for years a 12 core system is now a 6 core system with both cpu's working fine under heatsink A. However, neither cpu A or B will work under heatsink B. Very strange indeed don't ya think...!
 
Yep, check those pin connections in socket B.

bax2023 is referring to the delidded CPUs originally used in 2009s, vs the regular CPUs used from 2010 onwards. I have a 2009 that got upgraded ... using lidded CPUs. I had to set the heatsinks just right, with spacers, or there were problems with RAM sticks not being recognized.

Whoever upgraded my machine before I bought it, seems to have discarded a few heatsink parts. I didn't have the option to go back to delidded, so my dual-3.33 upgrade required heatsink fiddling. At least I didn't encounter boot failures.
 
  • Like
Reactions: bax2023
bax2023 said:
I just have to ask, is this Mac Pro 4,1 (Early 2009) or Mac Pro 5,1 (Mid 2010/2012) ?
Click to expand...
My Mac pro is an early 4,1 2009 flashed to 5,1 and worked flawlessly for 5 years until 4 days ago. Am still running OSX Sonoma, thanks to Open Core Legacy Patcher :)
Macschrauber said:
Check the socket B for bent or burnt pins. Microscope or good camera helps, Grandpa's reading glasses also.
Click to expand...

reader50 said:
Yep, check those pin connections in socket B.

bax2023 is referring to the delidded CPUs originally used in 2009s, vs the regular CPUs used from 2010 onwards. I have a 2009 that got upgraded ... using lidded CPUs. I had to set the heatsinks just right, with spacers, or there were problems with RAM sticks not being recognized.

Whoever upgraded my machine before I bought it, seems to have discarded a few heatsink parts. I didn't have the option to go back to delidded, so my dual-3.33 upgrade required heatsink fiddling. At least I didn't encounter boot failures.
Click to expand...
Hi Reader50 Thanks for the reply. My cpu's are the de-lidded ones as I did it myself, no issues with either one. Also have checked the cpu B bass for any bent pins ect, and no bent pins fluff or dirt (see pic attached)
20250409_111252.jpg
 
Macschrauber said:
Check the socket B for bent or burnt pins. Microscope or good camera helps, Grandpa's reading glasses also.
Click to expand...
Thanks mate, I'm the grandfather, lol and I am using my prescription close focus glasses as I do electronics. So here is the pic showing no bent pins, dirt or fluff ect. Still a mystery :(
20250409_111252.jpg
 
The Xeons have links. Even when working solo, the links could not work.

Check Ram sticks, by reducing.

Are the CPUs properly polished?

Have you tried, if the original CPUs working?
 
Last edited:
I'm not seeing any imperfections in the socket array, but the socket pic is out of focus - focus plane is above the pins. Camera seems to have latched onto the heatsink bolts. So it's hard to say for sure.

Agree with Macschrauber - RAM sticks are the next suspect. The RAM is not handled by a separate Northbridge chip - that's an old design today. Instead, each CPU drives the RAM slots associated with it. ie - if you have a single bad stick associated with socket B, then your system will be unstable whenever there's a CPU in socket B. If there's no CPU there, then the bad stick is not seen by the system, so everything would seem OK.

Test it by running with one pair of RAM at a time, trying different pairs. Until you narrow down a particular stick.
 
reader50 said:
I'm not seeing any imperfections in the socket array, but the socket pic is out of focus - focus plane is above the pins. Camera seems to have latched onto the heatsink bolts. So it's hard to say for sure.

Agree with Macschrauber - RAM sticks are the next suspect. The RAM is not handled by a separate Northbridge chip - that's an old design today. Instead, each CPU drives the RAM slots associated with it. ie - if you have a single bad stick associated with socket B, then your system will be unstable whenever there's a CPU in socket B. If there's no CPU there, then the bad stick is not seen by the system, so everything would seem OK.

Test it by running with one pair of RAM at a time, trying different pairs. Until you narrow down a particular stick.
Click to expand...
The CPU B socket pic may be slightly blurred, however, as you can see by the pin points, not one is bent, burnt or damaged in any way. As an retired navel engineer, I checked it and triple checked it, line by line in the daylight and 100% no issues there.

In terms of the RAM. I put all the sticks from my 4,1 into my 2010 5,1 and all were good, However, I will but the B cpu back in it's socket, in my 4,1 and try two ram sticks two at a time. Good idea :)
 
Last edited:
Macschrauber said:
The Xeons have links. Even when working solo, the links could not work.

Check Ram sticks, by reducing.

Are the CPUs properly polished?

Have you tried, if the original CPUs working?
Click to expand...
The cpu's were properly polished as then were both working for years. Also I swopped them over to cpu A Socket each cpu worked fine. However, when I tried both cpu's back in A and B sockets with new thermal paste, it would not boot up.
 
Hershey52 said:
The cpu's were properly polished as then were both working for years. Also I swopped them over to cpu A Socket each cpu worked fine. However, when I tried both cpu's back in A and B sockets with new thermal paste, it would not boot up.
Click to expand...
They have links to combine the Memory. For example they work with slots 5 to 8 empty. So CPU B needs access to the ram bank on CPU A.

This could fail, and of course a bunch of other components. So, do you have the pair of originals or other known good Xeons to test?
 
  • Like
Reactions: reader50
Macschrauber said:
They have links to combine the Memory. For example they work with slots 5 to 8 empty. So CPU B needs access to the ram bank on CPU A.

This could fail, and of course a bunch of other components. So, do you have the pair of originals or other known good Xeons to test?
Click to expand...
This is an interesting possibility I hadn't considered. The Xeons we use have the extra circuitry to work in pairs. If that circuitry were to fail in one of your Xeons, then they could no longer coordinate, not even during POST.

Hershey52, you've established that each of your Xeons will work by itself in socket A. Have you tried using them by themselves in socket B? (with socket A empty)

Perhaps the problem isn't to do with socket B. Rather, the issue may be that one CPU won't work in a pair any more. If this turns out to be the problem, give a thanks to Macschrauber. He mentioned the possibility twice before I caught on.
 
reader50 said:
Have you tried using them by themselves in socket B? (with socket A empty)
Click to expand...

Invalid configuration - with dual CPU tray when just one Xeon is installed, will only POST when the Xeon is installed to the CPU A socket.

Screen Shot 2025-04-10 at 21.07.10.png
 
tsialex said:
Invalid configuration - with dual CPU tray when just one Xeon is installed, will only POST when the Xeon is installed to the CPU A socket.

View attachment 2501018
Click to expand...
Thanks Tsialex for that.

I noticed that on point three, (Mac pro will not start if you use processor B Only) my fans do power up to 100% when booting up and powering down. Maybe this the problem, will check it out all connectors on processor board and back plane, and get back to you.
 
Hershey52 said:
Thanks Tsialex for that.

I noticed that on point three, (Mac pro will not start if you use processor B Only) my fans do power up to 100% when booting up and powering down. Maybe this the problem, will check it out all connectors on processor board and back plane, and get back to you.
Click to expand...

When you install only one Xeon - only works if the Xeon is installed to socket-A - to a dual CPU tray, the SMC will enter fail-safe mode and all the Mac Pro fans will run at full RPM and without any possibility of fan management, by design.

Running with just one Xeon installed to a dual CPU tray is just for diagnostic purposes and not intended for normal operation.
 
Thanks tsialex, have completely cleaned out my Mac Pro back to the backplane. However, after....

1. Taking out the Intake and Exhaust fan housing and cleaning the fans and their
contacts.
2. Checked and cleaned back plane
3. Put the cpu B back again, new paste.
4. Reset SMC, did a PRAM reset.
Unfortunately, my mac pro will still only work on CPU A
5. Fans still run at breakneck speed, and Red LED on RHS of processor board near RAM 4
flashes on start up.
6. Have installed Mac Fan Control, running Intake, Exhaust fans at half speed. Boost A is running at half capacity. Am still at a loss as to the problem...!
 
Last edited:
Macschrauber said:
They have links to combine the Memory. For example they work with slots 5 to 8 empty. So CPU B needs access to the ram bank on CPU A.

This could fail, and of course a bunch of other components. So, do you have the pair of originals or other known good Xeons to test?
Click to expand...
Unfortunately,
reader50 said:
This is an interesting possibility I hadn't considered. The Xeons we use have the extra circuitry to work in pairs. If that circuitry were to fail in one of your Xeons, then they could no longer coordinate, not even during POST.

Hershey52, you've established that each of your Xeons will work by itself in socket A. Have you tried using them by themselves in socket B? (with socket A empty)

Perhaps the problem isn't to do with socket B. Rather, the issue may be that one CPU won't work in a pair any more. If this turns out to be the problem, give a thanks to Macschrauber. He mentioned the possibility twice before I caught on.
Click to expand...
Have you tried using them by themselves in socket B? (with socket A empty)

My Mac Pro will not boot with a cpu in socket B, must be in slot A, as per troubleshooting manual.

Also, I don't have other cpu's to test other than buying de-lidded ones, or another pair of lidded which I may just have too...thanks :)
 
Hershey52 said:
One final question, till I get an early 2009 tray, how long could it be B4 I duel tray crashes completely. The reason I ask as, I may have to get a tray from the U.S
Click to expand...

I don't think that anyone can really predict that with any certainty…
 
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