MP 1,1-5,1 4,1 CPU overheating after CPU upgrade

[rorydaredking]

I picked up a 4,1 that’s upgraded to a 5,1 which the guy said had a few bent pins on socket A. Was cheap so I thought I would give it a go.
have used a magnifying glass and tiny spudger to unbend them but have also found that there are two pins which are bent off before it got to me.

The upside is it now boots, however after about a minute the lower level fans go mental and it kernel panics. Upon it auto rebooting, it crashes again within about 15 secs. Heat sink is only what I would say is warm to the touch, definitely not hot. I have a good fluke, so I wedged The thermocouple into the gap between the copper heat pipes and the copper plate, and that was only reaching about 28 Celsius

Running AHT isn’t much help as it fails within a second with issue 4SMC/1/4000000 SMC

strangely with both CPUs in, the fans don’t go mental but it does still crash in the same time frame

Also unable to run with just CPU B, it always does front LED single flash sequence

CPUs both delidded and both in CPU A have the same outcome.

Anyone got any ideas what my issue might be?

 

[Macschrauber]

Test it with one only one Ram Stick in each Slot.

most of the pins go to the Ram Slots as this Xeon has an internal Memory Controller.

 

[rorydaredking]

Only got a single stick in slot 1 at the moment, will try a different slot and see how I get on

 

[h9826790]

Heat sink is only what I would say is warm to the touch, definitely not hot. I have a good fluke, so I wedged The thermocouple into the gap between the copper heat pipes and the copper plate, and that was only reaching about 28 Celsius

Which means the heat is trapped inside the CPU, but can't transfer to the heatsink.

Anyway, did you remove the black silicon seal when delid? If that's not 100% cleaned, you should not install the plastic retainer, that will avoid the heatsink to touch the CPU die.

 

[rorydaredking]

Which means the heat is trapped inside the CPU, but can't transfer to the heatsink.

Anyway, did you remove the black silicon seal when delid? If that's not 100% cleaned, you should not install the plastic retainer, that will avoid the heatsink to touch the CPU die.

I did remove the silicone, but I’m on a 4,1 so no plastic retainer

will try a reapplication of thermal paste tomorrow.

Does anyone know if it should run With just CPU B installed? Or is single CPU mode only for A?

 

[h9826790]

I did remove the silicone, but I’m on a 4,1 so no plastic retainer

will try a reapplication of thermal paste tomorrow.

Does anyone know if it should run With just CPU B installed? Or is single CPU mode only for A?

Only dual processor 4,1 has that plastic retainer. Other model has metal clip in the socket.

Only CPU B can't boot, this is normal.

Only CPU A can work, but fans run at full speed.

 

[rorydaredking]

Unfortunately different RAM slots have the same result, as does a reapplication of thermal paste and also replaced thermal paste on northbridge. The most frustrating part is it doesn’t stay alive long enough to open istat to find out which temp is off the charts.

 

[rorydaredking]

Had ordered two new CPUs, tried one of those and same result. This is frustrating, but starting to seem to me like a tray fault.

Had another look at the socket and it has two pins missing, D21 and AL12. D21 is DDR1 CLK and AL12 is VCC PWR. No other pins look out of place, a couple are a little tarnished.

Any more ideas would be appreciated

 

[minifridge1138]

I picked up a 4,1 that’s upgraded to a 5,1 which the guy said had a few bent pins on socket A. Was cheap so I thought I would give it a go.
have used a magnifying glass and tiny spudger to unbend them but have also found that there are two pins which are bent off before it got to me.

So two pins are broken off?


Did you try booting it after you bought it but before you did anything to it?

 

[rorydaredking]

So two pins are broken off?


Did you try booting it after you bought it but before you did anything to it?

No, the pins were overlapping slightly and I thought it likely to cause more issues.

After unbending I did get it to boot and it was not going mental, tried to install the OS but it got stuck on 1 second remaining and didn't copy any files to the drive. System was still fully responsive when I cancelled the install

 

[rorydaredking]

Had it booting with 2 x 6 core deluded 2.8 GHz CPUs for a while. CPU A has DIMM 3 and 4 unavailable. CPU B had DIMMS 5, 7 and 8 unavailable. Then for one boot had only 5 unavailable. With both CPUs working, it wasn't able to wake from sleep, just hung on a black screen.

Tried adjusting the heatsink to recognise all RAM and now CPU B will never work, have tried both my 2.8GHz CPUs and also 2 3.06 GHz ones, all of them work in CPU A. Dual always attempted with the pairs, and each has booted successfully in single CPU A.

Situation with CPU B installed is either (depending on heatsink tightness)
1. Boots up, no lights on tray, no chime, just sits there
2. Boots up, lights on for slots 3 and 4 RAM (as expected) - then does nothing
3. Boots up, lights on for slots 3 and 4 RAM, chime, KP during load. If it does this and I then boot in verbose, during load it says 1 CPU detected during the first few lines.

This is certainly an education, never dealt with a computer as temperamental as this. Feels very much like I have a dud tray now.

Now able to boot if I have no RAM in slot 5 or 7. But in 6 or 8, no RAM detected once booted.