I have a Dell Precision T7610, 2 Xeon 2650 v2, 128Gb of ram, MB NK70N. Bought off of ebay in December and shipped with 1 processor and 64 gb of ram. I recently filled the other side of the board with processor #2 and 64 more Gb of ram. There are two gpu’s, Nvidia K420 for video and Nvidia P40 for accelerated graphics. I have shelled out my entire budget on electronics tools and building this ‘homelab’ computer. I do not have the money to buy another motherboard at this time, otherwise I would!
The motherboard that shipped with the computer needed replacement as socket #2 wasn’t detecting the cpu. Bought another motherboard off ebay, but it came with a broken cmos socket. Seller shipped another, so I now have 3 motherboards on hand. There’s a reason I mention this, as I’ve been using the extra boards to help me troubleshoot what’s going on.
So, with the latest motherboard, I set everything back up and was able to boot into the os just fine. A part of my project is enabling a Nvidia Tesla P40 gpu to work within a Windows vm running on Proxmox. Since the gpu has a large amount of ram, I read that it would be good to enable memory mapped IO > 4Gb. After finding the setting in the bios menu, I enabled it and rebooted.
Upon powering back on, the computer ‘seemed to turn back on’ but was rather lifeless. No video signal being sent and the 4 disk array of sas drives would not initialize (as in I couldn’t hear them working, they would spin up but that’s all), the computer was also showing diagnostic lights indicating some hw failure but the manual was not more specific than that.
I dug out from storage the motherboard with the broken cmos socket and fixed it with solder, wire, and tape. I set this motherboard back up and was able to get to the bios splash screen with the dell logo and load the grub bootloader for proxmox but after first showing a FW and ME error and then ‘loading initial ramdisk’ message appeared in the prompt, the computer would turn off.
I had a Proxmox flash usb live disk laying around so I stuck it in and was able to boot into the installer but the prompt again showed the FW and ME error before doing so. I selected ‘advanced options’ and saw there was a memory test feature. I chose to run the memtest to observe any errors.
The test would begin and run for some time but then the computer would shut off, mid test. I had no problem with the computer when it first shipped, so I knew that it was likely the ram that shipped with the computer at that time was good. I decided to remove the 2nd bank of ram/processor and ran the memtest again. This time, no crash, the computer was able to perform the test without crashing.
Seeing that I could boot into Proxmox, and access it via web gui, I was curious to see how operable the system was at this point. I started a Windows 11 vm but the vm would freeze during the Windows splash screen, not immediatly, but before Windows had a chance to boot up completely.
Before turning on the memory mapped IO > 4Gb feature in bios, the computer ran just fine. Proxmox reported both cpu’s and listed 128 gb of available ram. I was able to create vm’s at will and had no problems. Then, after turning on the memory mapped feature it almost seemed like it somehow damaged the hardware.
I am curious to get the 3rd motherboard working again and see how the system runs with that. I want to figure out a way to reset the bios on that 3rd motherboard, because it was working before I changed the memory mapped setting in the bios.
Hopefully these details help make the picture clear enough for people to chime in.
Thank you,
el
