I need some help from the many Mac Pro experts that populate this outstanding forum 
I was commissioned to upgrade a dual 4,1 (dual quad 2.93GHz) to a fully fledged 5,1 running dual X5680’s. The machine temps were running a little high and the fans were running speedier than usual, but I thought, no worries we will get fixed with a fresh coat of thermal paste. NorthBridge Temps - Were pretty standard (better than CPU’s)
Step 1: Firmware update 4,1 to 5,1 - OK!
Step 2: CPU delliding - OK
Step 3: Replace the Processors - reapply the termal paste (place back the heatsinks) - OK
Step 4: Nice smooth booting. Time to start running some benchmarks - Cinebench R23 (10 minutes test) - great scores to begin with, but after a few minutes CPU’s reached 81º C and eventually the machine crashed.
Step 5: A bunch of successive Kernel Panics.
Step 6: RAM Diagnostic Lights - Lid up
Step 7: Change RAM around (it was running on OWC ram - 48GB). No success
Step 8: Remove the new CPU’s - visually inspect the socket pins (all good).
Step 9: Re-Apply everything, close it up. No boot, no post! Same results
Step 10: Use my own ram. Ram led’s finally shutdown but the machine doesn’t post or boot.
Step 11: Go back to the original CPU’s (Perhaps I messed up one of the processors during the delliding). Try placing only CPU - A (NO Success, no post, no boot) - Plastic Bracket CPU holder is on most of the times.
Step 12: Despair sets in!!! Numerous attempts to tighten or loose the heatsink a bit better are done all the way to 1/4 of a turn.
Step 13/14/15/16: Try multiple combinations: Place my own 4,1-5,1 Single processors tray inside the machine. With my RAM perfect boot (always), with the clients Ram (dimm by dimm attempts) - No boot).
Step 120: Place costumer tray inside one of my “real” - 5,1 (Mid 2010) - No post, no boot! (My RAM). Not willing to the opposite place my tray inside his Mac and risk ruining a precious 5,1 dual CPU tray of my own inside his machine… The costumer has performed a slight MOD to his PSU (not Pixlas-MOD) material, basically he is bring in some extra power from the superdrive sata bus to the GPU, but placing a direct connection from the plug (not sure of the correct wording for this but let’s call it a "stent")
The guy has mentioned that he was facing occasional crashes when using two GPU’s inside the machine. Even if he was only powering up one at a time. I’ve received the Mac with a single GTX 680 (mac edition) but he uses RX 590 OC in windows to do some flight simulator gaming. From what I could gather he normally left the GTX 680 (unplugged but still seated inside the PCI-E Slot when gaming with the RX 590 in windows), so I’m assuming that the GTX 680 was still consuming energy from the PCI slot.
Why is all of his RAM suddenly dead? Additional information: There was an extra OWC Dimm that I’ve left out of the machine during the whole process (8GB more from OWC to a total sum of 56GB). Like the remaining 48GB, that were actually inside the machine, this “extra” 8GB were NOT inside the machine during the upgrades but still currently they're NOT recognised in any of my machines. Again these additional 8GB were left OUT of the tray during the whole process!!!! They were inside when I received but I took them out to maximize triple channel and matching number of occupied slots.
Is it possible that his tray was “fried” due to a PSU problem? The Dual x5680 are 130W processors vs the 95W drawn from the original X5570 could this make the difference?
Any suggestion of things to try and see if I can bring back his tray back to life?
Cheers!
I was commissioned to upgrade a dual 4,1 (dual quad 2.93GHz) to a fully fledged 5,1 running dual X5680’s. The machine temps were running a little high and the fans were running speedier than usual, but I thought, no worries we will get fixed with a fresh coat of thermal paste. NorthBridge Temps - Were pretty standard (better than CPU’s)
Step 1: Firmware update 4,1 to 5,1 - OK!
Step 2: CPU delliding - OK
Step 3: Replace the Processors - reapply the termal paste (place back the heatsinks) - OK
Step 4: Nice smooth booting. Time to start running some benchmarks - Cinebench R23 (10 minutes test) - great scores to begin with, but after a few minutes CPU’s reached 81º C and eventually the machine crashed.
Step 5: A bunch of successive Kernel Panics.
Step 6: RAM Diagnostic Lights - Lid up
Step 7: Change RAM around (it was running on OWC ram - 48GB). No success
Step 8: Remove the new CPU’s - visually inspect the socket pins (all good).
Step 9: Re-Apply everything, close it up. No boot, no post! Same results
Step 10: Use my own ram. Ram led’s finally shutdown but the machine doesn’t post or boot.
Step 11: Go back to the original CPU’s (Perhaps I messed up one of the processors during the delliding). Try placing only CPU - A (NO Success, no post, no boot) - Plastic Bracket CPU holder is on most of the times.
Step 12: Despair sets in!!! Numerous attempts to tighten or loose the heatsink a bit better are done all the way to 1/4 of a turn.
Step 13/14/15/16: Try multiple combinations: Place my own 4,1-5,1 Single processors tray inside the machine. With my RAM perfect boot (always), with the clients Ram (dimm by dimm attempts) - No boot).
Step 120: Place costumer tray inside one of my “real” - 5,1 (Mid 2010) - No post, no boot! (My RAM). Not willing to the opposite place my tray inside his Mac and risk ruining a precious 5,1 dual CPU tray of my own inside his machine… The costumer has performed a slight MOD to his PSU (not Pixlas-MOD) material, basically he is bring in some extra power from the superdrive sata bus to the GPU, but placing a direct connection from the plug (not sure of the correct wording for this but let’s call it a "stent")
The guy has mentioned that he was facing occasional crashes when using two GPU’s inside the machine. Even if he was only powering up one at a time. I’ve received the Mac with a single GTX 680 (mac edition) but he uses RX 590 OC in windows to do some flight simulator gaming. From what I could gather he normally left the GTX 680 (unplugged but still seated inside the PCI-E Slot when gaming with the RX 590 in windows), so I’m assuming that the GTX 680 was still consuming energy from the PCI slot.
Why is all of his RAM suddenly dead? Additional information: There was an extra OWC Dimm that I’ve left out of the machine during the whole process (8GB more from OWC to a total sum of 56GB). Like the remaining 48GB, that were actually inside the machine, this “extra” 8GB were NOT inside the machine during the upgrades but still currently they're NOT recognised in any of my machines. Again these additional 8GB were left OUT of the tray during the whole process!!!! They were inside when I received but I took them out to maximize triple channel and matching number of occupied slots.
Is it possible that his tray was “fried” due to a PSU problem? The Dual x5680 are 130W processors vs the 95W drawn from the original X5570 could this make the difference?
Any suggestion of things to try and see if I can bring back his tray back to life?
Cheers!
