MP 1,1-5,1 Mac Pro 5,1 no chime/no boot after CPU upgrade...

[Sample]

Hi to all, finally found some free time (yes, I know it Christmas...?) to swap the old 2 original quad core intel Xeons 2,26 CPUs in my dual Mac Pro (4,1 to 5,1) with 2 used delidded hexa-core X5675.
Been very careful in the repasting process and succesfully done it in the past, also already replaced the old Northbridge clips and repasted it and it had been working ok.

When I boot though, I get no chime and black screen, bot CPU fans are working at normal speed as are all the others, no solid red light on the board apart from the brief red flash of the led close to the frontal RAM bank at startup.

I can't even try to reset the PRAM because it won't even chime..
Any ideas?

Thanks a lot guys... and Merry Christmas

 

[Macschrauber]

is efi done lit when pressing the diagnose button?

even when no audible chime - try 4 nvram resets in one row.

maybe sound volume was set to zero

 

[Sample]

is efi done lit when pressing the diagnose button?

even when no audible chime - try 4 nvram resets in one row.

maybe sound volume was set to zero

Thanks for your help Macschrauber!

here is the status of the 7 diagnostic leds:

CPUA: off
CPUB: off
PWROK: solid green
3v STBY: solid amber

SYS: solid green
EFI DONE: solid green
GPU OK: flashing green

I tried replacing my GPU (R9 280 3Gb) with the old original Nvidia gt120 but screen is still black.

I also tried, as you suggested, command/alt +P+R pressed from the start even if there is no chime but nothing happens....

 

[Koka016]

Hi you have the same problem as i have..
Did you updated boot rom to 144.0.0.0.0 ?
And did you upgraded with Mojave or High Sierra installers or other way?

 

[Macschrauber]

To my experience the flashing gpu led is a bad cpu board.

also take out all ram sticks but one

 

[Sample]

Hi you have the same problem as i have..
Did you updated boot rom to 144.0.0.0.0 ?
And did you upgraded with Mojave or High Sierra installers or other way?

exactly! I followed tsialex's detailed instructions in upgrading first to High Sierra and then Mojave, my boot rom is 144.0.0.0.0.
Mojave worked perfectly, but now with these X5675 I get no chime nor boot

Anyone's got some clues? thanks a ton..

 

[Sample]

To my experience the flashing gpu led is a bad cpu board.

also take out all ram sticks but one

Hi Macschrauber, I just tried with just one Ram first in the cpu A bank an then just one in the cpu B bank but no results...
To the touch, the CPU A heatsink feels warmer that the CPU B heatsink.

Do you mean that a green flashinf GPU OK led means my CPU board is toast? Are you sure? It's been working until 2 hours ago before I upgraded the CPUs...

 

[Macschrauber]

Hi Macschrauber, I just tried with just one Ram first in the cpu A bank an then just one in the cpu B bank but no results...
To the touch, the CPU A heatsink feels warmer that the CPU B heatsink.

Do you mean that a green flashinf GPU OK led means my CPU board is toast? Are you sure? It's been working until 2 hours ago before I upgraded the CPUs...

thats my experience with 3 different single 4,1 CPU Boards I have in stock to repair some day, hopefully

place back the original processors to double check, and check for bent pins in the cpu sockets.

also you can test with the back processor allone.

 

[Sample]

thats my experience with 3 different single 4,1 CPU Boards I have in stock to repair some day, hopefully

so what would you advise me to do now, try and put back the old CPUs to see if the board still works? Will it work if I just make a test with just CPU A? Or should I fisrts try and make a test with the new ones, trying first one and then the second before swithching back to the old ones?

 

[Macschrauber]

I would take upgraded front cpu out and test, dont mind for full fans.

dont run too long, chime and a booting system is enough.


if it runs, front cpu or board is dead

so go from one processor to the next.

back processor allone should survice self test (chime).

check backplane and processor socket pins

 

[Sample]

Ok so just to be sure I understand well, I have to take CPU B and its heatsink off, and try and start with just CPU A (the one over the Northbridge)?

 

[Macschrauber]

Yes, the back one, I prefer back and front as this is not so confusing.

the back one of dual cpu boards has the same position as the one processor of a single board.

and so at least start a mac pro when front cpu of a double cpu board is left empty.

 

[Sample]

Ok, got it, so just before removing CPU B (front) I thought as a first step I might untighten by just one turn the bolts of heatsink B, just in case they were too tight, and now my Mac starts perfectly, however it only sees CPU A and the single RAM stick that I had left in its relative Ram bank A.Diagnostic led GPU OK is solid green again.

CPU B and its single RAM stick I left there are not seen, at the touch heatsink B is cold and off.
So I tried tightening CPU B's bolts 1 turn again and we're back to no chime no start.
Can just 1 single turn of the bolts turn off a CPU?

Btw at this stage might it be that the new (used) X5675 CPU B I just bought is faulty? Should I try and swap it in CPU A?

 

[tsialex]

Btw at this stage might it be that the new (used) X5675 CPU B I just bought is faulty? Should I try and swap it in CPU A?

Sure, it's the next logical step.

 

[Sample]

Sure, it's the next logical step.

Hi tsialex!
ok so I put hexa CPU B in place of hexa CPU A and it starts ok, so the new "used" hexa-core CPUs are ok.

Then I tried putting my old original quad-core CPUs and sometimes there's no chime and stalls, sometimes it starts but a red led just at the right of CPU B switches on, once Mojave is started fans max out.
Tha strange thing is that when it starts it sees CPU A I guess, but just 8 of the 16 Gb or RAM in banks 1-2-3-4 are seen (one stick is 8Gb 1333 and the others are 2Gb 1060).

My god, have I wrecked my Mac while installing the new "used" hexa CPUs?
I gotta work, am I screwed?
Please, give me some advice tsialex...

 

[tsialex]

Hi tsialex!
ok so I put hexa CPU B in place of hexa CPU A and it starts ok, so the new "used" hexa-core CPUs are ok.

Then I tried putting my old original quad-core CPUs and sometimes there's no chime and stalls, sometimes it starts but a red led just at the right of CPU B switches on, once Mojave is started fans max out.
Tha strange thing is that when it starts it sees CPU A I guess, but just 8 of the 16 Gb or RAM in banks 1-2-3-4 are seen (one stick is 8Gb 1333 and the others are 2Gb 1060).

My god, have I wrecked my Mac while installing the new "used" hexa CPUs?
I gotta work, am I screwed?
Please, give me some advice tsialex...

Get a very good light and inspect the pins of the CPU socket.

Btw, take a rest and do this tomorrow morning. This is exactly the the type of thing that needs a good night of sleep, 2732 pins to inspect…

 

[Macschrauber]

Are the CPU pads clean? Clean it with Isopropyl.

For the pins a good camera with Macro and flash is also of use.

I have a stereo Microscope for such work, cause I am an ole dog

 

[Sample]

Ok guys thanks, at a fisrt glance the pins looked ok before the last CPU install but I'll check again with my pro camera...
Yes I had always cleaned the pads and heatsinks with isopropyl.
Merry Christmas tsialex and Macschrauber, I really appreciate your help..

 

[arche3]

Before you go crazy. Try loosening the screws that hold down the heatsinks. It that doesn't work try tightening them. A lot of the time it's just the wrong tension on the screws holding down the cpus. But be careful not to over tighten. That will break it

 

[Macschrauber]

Before you go crazy. Try loosening the screws that hold down the heatsinks. It that doesn't work try tightening them. A lot of the time it's just the wrong tension on the screws holding down the cpus. But be careful not to over tighten. That will break it

undelidded cpus here,

if the black seal is cut completely theres no issue with the screws.

if the seal is there either cut it completely off to the ground or dont use the pastic spacers.

using the spacers is recommended, if there is no reason for them Apple had not used them.

 

[motulist]

I can't provide any info but i just wanna say how much i enjoy all this super techy nerding out =)

 

[Sample]

Well, looks like a story with a happy ending, everything’s working at last! (it's been a long night...?)
Still not sure which of the following actions fixed the issue, but just in case it might be useful for others I will make a list, first two are the most probable I think:

1) TSIALEX: Found a very slightly “out of line” pin on CPU B and gently pushed it back in line it with a tiny screwdriver and a magnifying glass (see picture below). Could such a little bend in a pin cause this havoc?

2) MACSCHRAUBER: yes the CPUs are delidded, but about 0,5mm of the seal was still there, so I cut out as much as I could with a razor blade. Always used spacers, before and now. Is it possible that that extra thickness might have caused the thermal surface not to perform a perfect and even contact with the heatsink?

3) Before repasting, this time I more thoroughly scrubbed the thermal surface of the two X5675, they had looked pretty clean when they arrived, but I repeatedly scrubbed them again and again and lots of dirt kept coming and coming out on the cotton swab. They're never too clean.

3) Sprayed a little contact cleaner in both the heatsink’s connectors, and on the tray's female and the board's male connectors.

4) Even if all already pretty clean, I sprayed a little compressed air everywhere.

5) I tightened the heatsinks just as I always had done, just finger tight (using thumb and index) and stopping as soon as I feel a sudden increase in resistance (for delidded usually around 7-8 turns, in alrernated sequence of course).

6) I performed a PRAM reset but that was followed by the last failed attempt before the final repasting, however I thought I should mention it.

Here's a pic I made of the bent pin on CPU B before fixing it:

Yeee, thanks to you all guys! ?

 

[Koka016]

Im glad to hear!

 

[Macschrauber]

I turn the screws down to a good tight feeling, not just finger tight. In cross order of course.

Never had a problem Until yet, did more than a dozen boards.

I cut the seal down and polish the remains.

Also polish the die until it gets the same finish as the original Processors.

if the height is too much of the processor/spacer construction damage can occur.

never ever would do it with the lid on. Not just for the processor sockets, also the voltage regulator parts need decent contact to the heat sink.

 

[tsialex]

Well, looks like a story with a happy ending, everything’s working at last! (it's been a long night...?)
Still not sure which of the following actions fixed the issue, but just in case it might be useful for others I will make a list, first two are the most probable I think:

1) TSIALEX: Found a very slightly “out of line” pin on CPU B and gently pushed it back in line it with a tiny screwdriver and a magnifying glass (see picture below). Could such a little bend in a pin cause this havoc?

2) MACSCHRAUBER: yes the CPUs are delidded, but about 0,5mm of the seal was still there, so I cut out as much as I could with a razor blade. Always used spacers, before and now. Is it possible that that extra thickness might have caused the thermal surface not to perform a perfect and even contact with the heatsink?

3) Before repasting, this time I more thoroughly scrubbed the thermal surface of the two X5675, they had looked pretty clean when they arrived, but I repeatedly scrubbed them again and again and lots of dirt kept coming and coming out on the cotton swab. They're never too clean.

3) Sprayed a little contact cleaner in both the heatsink’s connectors, and on the tray's female and the board's male connectors.

4) Even if all already pretty clean, I sprayed a little compressed air everywhere.

5) I tightened the heatsinks just as I always had done, just finger tight (using thumb and index) and stopping as soon as I feel a sudden increase in resistance (for delidded usually around 7-8 turns, in alrernated sequence of course).

6) I performed a PRAM reset but that was followed by the last failed attempt before the final repasting, however I thought I should mention it.

Here's a pic I made of the bent pin on CPU B before fixing it:

View attachment 1700789

Yeee, thanks to you all guys! ?

Needs just one QPI pin out of contact to render the CPU tray useless. You probably did everything that I would do in this situation.

@Macschrauber is correct, you have to make the die surface completely clean and perfectly level, I always try to make is similar of the original un-lidded Xeons.

It's extremely important to remove any black sealant from the Xeon PCB if you are using the spacers, needs to be absolutely level and shittty things will happen:

• you will have poor heatsink contact and can overheat,
• you bend socket pins,
• both.

People reading this in the future - never use a contact cleaner that has oil in the formula, test on an empty PCB space if the contact cleaner that you are using don't leave any residue or you will have to spend hours cleaning it later. Please don't ask how I know this…