Problems with new IOMMU build and possibly PCI / VGA passthrough

[lordoxide]

Guys,

 

My server is completely freezing randomly when I use the VM, and when I say freezing I mean 100% need to hold the power for 10 seconds to revive the box. It feels like a hardware issue, specifically because after a freeze I have to reboot 2 or 3x to even get it to post past the sas controller spin up. I spent 20 hours researching and testing things listed across the forums on similar problems, but I cannot find a fix. I pulled apart my unraid and gamer and built a single box, with a GTX 560 TI in the first PCIe and a GTX 770 in the 2nd PCIe. I am able to run unraid just fine (it does seem to be very sluggish to boot after the SAS controller posts, until I see the bios logo) but I have let a full parity check, while running a preclear and it never shuts down, processor never goes above 40c. I have been able to run a seabios passthrough to windows10 with really no problems, unless I have a parity check or preclear running at the same time it seams. I just need some direction if anyone has an idea of what could be wrong, before I start pulling hardware testing by elimination

 

VM disks are created on a 2drive SSD cache pool

GTX 770 is in 2nd PCIe 16x 3.0 running as passthrough

GTX 770 shares iommu group only with HDMI audio controller, both are shared to VM

Using system temps plugin I have never seen 'physical id 0' go over 40c

Every freeze referenced below is a 100% full system lock, followed by not being able to get the system to post for 4-5 boots

SeaSonic X-1250watt Gold power supply

Cores 0-5 assigned to VM, 12800 Ram

 

Tried:

Start VM (no problem) start preclear on non-asssigned drive = freeze immediate

Started preclear (no problem), start VM = freeze immedate

Start VM (no problem), start parity check = random freeze at some point

Let the VM run with no preclear or parity = no freezing

Ran memtest for 6 hours with no errors

New VMs with both bios, and tried both machine types

 

Any help would be appreciated.

 

Thanks

[saarg]

Since you are not able to boot the machine after it freezes, I think it must be a hardware failure.

Hard to say what it could be. Might be the motherboard.

I would start with a minimum of cards in the computer and then add one by one until the freeze happens.

[bungee91]

While I know you can't capture much info prior to the hard reset, please post the output of diagnostics please.

That way we can overlook anything that may seem small, but could be causing an issue.