Trying to do a Win10 VM, passthrough of the FX 3700 GPU seems to cause an issue, when I removed a lot of pass-through features, it let the VM attempt to start up and install- when I get to the part where one inputs the virtIO drivers per the guides, it is considered a non-compatible driver package and the installation cannot go any further (it is also saying there is no disk...)

 

Some fault codes are:

internal error: early end of file from monitor: possible problem:

2016-03-12T04:07:53.343870Z qemu-system-x86_64: -device vfio-pci,host=0f:00.1,id=hostdev0,bus=pci.0,addr=0x5: vfio: failed to set iommu for container: Operation not permitted

2016-03-12T04:07:53.343915Z qemu-system-x86_64: -device vfio-pci,host=0f:00.1,id=hostdev0,bus=pci.0,addr=0x5: vfio: failed to setup container for group 14

2016-03-12T04:07:53.343923Z qemu-system-x86_64: -device vfio-pci,host=0f:00.1,id=hostdev0,bus=pci.0,addr=0x5: vfio: failed to get group 14

2016-03-12T04:07:53.343939Z qemu-system-x86_64: -device vfio-pci,host=0f:00.1,id=hostdev0,bus=pci.0,addr=0x5: Device initialization failed

2016-03-12T04:07:53.343949Z qemu-system-x86_64: -device vfio-pci,host=0f:00.1,id=hostdev0,bus=pci.0,addr=0x5: Device 'vfio-pci' could not be initialized

 

The above quote is having removed the GPU passthrough but this time around it won't even spin up the VM- and VT-x and VT-d2 are enabled.

x2 xeon 5520, 1TB HDD, 120 gig cache.  HP z800.

 

After deleting/removing the attempt and trying a new VM, when trying to passthrough the GPU

internal error: early end of file from monitor: possible problem:

2016-03-12T04:16:47.586855Z qemu-system-x86_64: -device vfio-pci,host=42:00.0,bus=root.1,addr=00.0,multifunction=on,x-vga=on: vfio: failed to set iommu for container: Operation not permitted

2016-03-12T04:16:47.586904Z qemu-system-x86_64: -device vfio-pci,host=42:00.0,bus=root.1,addr=00.0,multifunction=on,x-vga=on: vfio: failed to setup container for group 22

2016-03-12T04:16:47.586919Z qemu-system-x86_64: -device vfio-pci,host=42:00.0,bus=root.1,addr=00.0,multifunction=on,x-vga=on: vfio: failed to get group 22

2016-03-12T04:16:47.586937Z qemu-system-x86_64: -device vfio-pci,host=42:00.0,bus=root.1,addr=00.0,multifunction=on,x-vga=on: Device initialization failed

2016-03-12T04:16:47.586954Z qemu-system-x86_64: -device vfio-pci,host=42:00.0,bus=root.1,addr=00.0,multifunction=on,x-vga=on: Device 'vfio-pci' could not be initialized

 

System as follows:

Model: Custom

M/B: Hewlett-Packard - 0AECh

CPU: Intel® Xeon® CPU E5540 @ 2.53GHz

HVM: Enabled

IOMMU: Enabled

Cache: 256 kB, 1024 kB, 8192 kB

Memory: 24576 MB (max. installable capacity 96 GB)

Network: eth0: 1000Mb/s - Full Duplex

Kernel: Linux 4.1.18-unRAID x86_64

OpenSSL: 1.0.1s

 

More info, went into system and the two GPU's info are:

 

0f:00.0 VGA compatible controller: NVIDIA Corporation GK107 [GeForce GT 630 OEM] (rev a1)

/sys/kernel/iommu_groups/14/devices/0000:0f:00.0

 

 

42:00.0 VGA compatible controller: NVIDIA Corporation G92GL [Quadro FX 3700] (rev a2)

/sys/kernel/iommu_groups/22/devices/0000:42:00.0

 

Some reason the FX 3700 isn't playing nice.

 

Update, put the FX 3700 in my windows machine, its been a drama queen.  Windows tried to download drivers while I was installing the Nvidia driver.  Seems like a lot of hickups happening.  A lot of forced shut downs, either by me (holding the power down) because the screen going blank, or just solid green, or a few times Windows did a forced shutdown. 

 

The GPU is getting really hot to the touch, but the fan won't ramp up, I have to manually ramp it up with an Nvidia controller- funny thing is the temps do not go much past 50c and the card quits out on me.  To see if it was a driver issue I ran a live boot Ubuntu USB, it ran really great for about 5 minutes and started to freeze up as well.

 

Took the GPU apart and cleaned off the old gunky heat compound and put some new stuff on, doesn't seem to have helped.  After this next cool down I'm going to hold the fan speed high and see what happens.

I've learned, with Windows 10, NOT to use the nvidia install package yourself because Windows is retrieving the version IT decided you should be running 

 

That said, you've tested the card in 2 machines and had issues. Sounds hardware related. Though GPU's do run hot, so hot to the touch isn't always a sign of a problem.

Yeah, should have given Win10 more time and that driver might have done better.

 

I ramped the speed up manually and the GPU seemed to last a little longer, and degraded in a more dramatic fashion when it did start to overheat again- started to get minor artifacts, slow feedback and then blanked out.  So it is heat related for sure, but it's tolerance to heat is very very low, this is happening at ~45c vs. 90+c.  It was a "free-be" with some used hardware I bought, so go figure.

 

How is the AMD support on unRAID with passthrough.  There is a Sapphire R7 250 Ultimate that is passively cooled- something about passive cooling I really like- this would be for VMs that do not require a lot of GPU umph of course.

 

 

--- UPDATE ---

Still haven't booted up unRAID, but I took the FX3700 apart and baked the board for about 10 min, re-did the heat compound and bam, works great now on the Win10 machine.