MSI Mainboards > Intel Core-iX boards

Spontaneously lost both PCI-e slots on PD67A-GD65 B3

(1/3) > >>

spaten:
I have a long working (+10 months) dual headed (single card) build (see below for full details) that I use as my daily PC.  I've had no issues, glitches, or other indicators that anything has been wrong.

I sat down to the PC and logged another user off.  Then logged on to my account.

When my account logged in, the Primary display went black, panel said 'no signal'.  The Secondary display continued to work fine.

Figuring something went buggy in the driver/windows I did a reboot.  The PC seemed to boot fine (usual clicks and beeps) but neither display would kick on, both said they saw no signal and went to sleep.

I run a remote desktop manager on the pc, so went to my phone to see if I could log on.  Sure enough, i could see my desktop from my phone, but not on the actual screens.

I opened the display properties and windows seemed to think all was fine.  It saw my display adapters with no problems, and even let me change the resolution.  All the while, both panels continued to say there was no signal.

I rebooted again, this time didn't hear the usual beeps, etc.  And when I tried to connect with the remote desktop, I could no longer see it.

I proceeded to:

remove the GTX480 GPU and swap it with another working PCI-e card (8800GT) I had on hand...... no change
tried the 8800GT in the second PCI-e slot ..... no change
removed the 8800GT and put the GTX480 in the second PCI-e slot..... no change
put the GTX480 back into the original first PCI-e slot then removed all but one DRAM..... no change
Tried various combinations of the DRAM.... no change

Then I dug out an old PCI (that's regular PCI) video card from the garage, got it hooked up to the VGA input to my panel, removed the PCI-e card and gave it a go.

It booted up without a problem, display worked fine.  I took this opportunity to update the bios to 1.17 (the last before adding support for Ivy Bridge - since I don't need it).  And rebooted again, still everything worked fine.

I pulled the PCI card out, put my PCI-e GTX480 back in, and was back to nothing working again.

I checked voltages on both of the PCI-e power leads with a multi meter and had a good solid 12v DC.

I've read other posts where people have had bent pins on the CPU pad leading to one or both PCI-e slots failing.... I just can't see this being the case for me.  Since the system was working fine for many months, and nothing physical had been touched in the box.

Also, this "in between failure mode" where the displays didn't see a signal from the video card, but I could remote in to the desktop, and saw that the OS thought everything was fine.... really has me confused.

I've built dozens of systems in my life, am quite tech savvy, and have solved many weird problems over the years... but here, I am stumped.

Any ideas would be greatly appreciated.....

mobo: MSI P67A-GD65 B3 - bios at 1.17
CPU: Intel i5 - 2500 - (no overclocking) - massive zallman cooler
RAM: Corsair 4x4GB (16GB) DDR3 - don't have the model number handy, but it's matched to the MSI approved list for the mobo
GPU: GTX480SE PCI-e dual headed from DVI connections
OS: win7 64bit

xmad:
Do you have another cpu you are able to test with or able to put that cpu in another computer?

Sounds like either the mobo or the cpu.

spaten:

--- Quote from: xmad on 20-August-12, 17:02:11 ---Do you have another cpu you are able to test with or able to put that cpu in another computer?

Sounds like either the mobo or the cpu.

--- End quote ---

I wish I did, but don't.  I've narrowed it to those two as well.  Seems really odd for electronics failures to up and do this past the usual break in times.

Wish there was a way to pinpoint one or the other.  I fear I'll be buying both out of assurance.

I'm kinda surprised that if there is such a failure, that the bios or otherwise wouldn't report it.  I would expect the motherboard diagnostics to realize that it's two pci-e slots are 'unavailable'

Bernhard:
Seeing that your system did work, yes agreed, that should eliminate the possibility of bent pins, but I would recheck the torque of the cooler. Maybe remove it and check that one of the securing fasteners hasn't come loose resulting in uneven torque and causing the CPU not to make full contact. Its a long shot but nothing is impossible. You may have to take the CPU out in any case to have it tested on another machine. The 16 PCIE lanes are provided by the CPU itself, so that may be the origin of the problem. Otherwise its the circuitry of the board.
 

spaten:

--- Quote from: Bernhard on 20-August-12, 17:26:35 ---The 16 PCIE lanes are provided by the CPU itself, so that may be the origin of the problem. Otherwise its the circuitry of the board.

--- End quote ---

Thanks Bernhard...

This does clarify the possibility.  And explains how the PCIe buses could have issues without the BIOS knowing.  Also explains why I couldn't find anything referencing the graphics output in the BIOS (back with AGP you would specify whether the BIOS should target booting to AGP or PCI..... etc).

I'll certainly take a look at the pins, etc.... if it's broke, i'll be pulling it if I end up replacing it anyway

Navigation

[0] Message Index

[#] Next page