I just got a used rig from my friend that has been offline for 2/3 years (he said it worked pretty fine back then). It consists of:
8x RX580 8gb PowerColor
Asus Prime Z270-AR
Intel Pentium® G4400@3.30Ghz
4gb of ram (2133mhz)
2x Corsair CX750m power supply
Sandisk 16gb flash drive
2x PSU Corsair CX750m
When I run only one gpu, HiveOS starts fine. When I plug the second GPU onward, I start to get the error “UVD not responding, trying to restart the VCPU” as can be seen on the image (https://imgur.com/a/JTGAqMw).
I tried to change the PCI-e Auto, Gen1 and Gen2 (only didn’t try Gen3). Also tried to edit /etc/default/grub to change the line to GRUB_CMDLINE_LINUX=“radeon.modeset=nomodeset” (also tried radeon.modeset=0 and just “nomodeset”).
Not sure if there is another config that can be made to solve this.
Any clued to help!? Ty a lot!
USEFULL INFO
I’m testing with only a single PSU and 2 or tops 3 gpus, so the power cables come all from the same PSU.
All gpus are with original factory BIOS.
UPDATE:
Tests already made, without managing to boot more than two gpus:
Hive OS update to latest stable version: done.
Motherboard battery change: done.
Change the riser of the second gpu for a new one: done.
Change from 4gb of ram to 8gb of ram: done.
Update the BIOS: done.
Config in the BIOS:
System Agent (SA) Configuration > VT-d > Disabled: done.
System Agent (SA) Configuration > DMI/OPI Configuration > DMI Max Link Speed ) Gen2: done
System Agent (SA) Configuration > PEG Port Configuration > PCIEX16_1 > Gen2: done.
System Agent (SA) Configuration > PEG Port Configuration > PCIEX16_2 > Gen2: done.
System Agent (SA) Configuration > PEG Port Configuration > PCIe Spread Spectrum Clocking > disabled: done.
You know what version of bios grachics cards they have? Maybe your friend have them bios flashed allready with a bios before 3 years ? in your situation i restore the default bios in one by one and test it … if you check all ready all the above maybe your problem is your bios of your cards… and check also if have the latest hive os stable version…
four gpus in one psu is not a problem (check at the wall with power meter to be sure about aberage… your connections to the risers must it with 6 pin power pin…(not sata cable) this cards with default bios take 120-130 watts …
Put one gpu in first pcix 1x not 16x (the bigger one ) flash ssd drive with the latest stable version of hive os (it has the drivers for this gpu’s) make the power connection with 6 pin to risers … after this you sould be running … then add another gpu e.tc. (put in your motherboard one by one and check if has a problem…
I am really bugged by this rig. I found out that the problem is not exactly plugging the second gpu onwards, it is actually with 7 of 8 gpus that came with the rig. I plugged other two gpus that I had and they worked fine along the only gpu of the 8 that worked.
What is bugging me is that when I got the rig, there were only one gpu plugged directly on the mobo and there were other 7 gpus plugged in risers. And these 7 gpus are the ones that are not working.
I’m still trying to understand what is happening, trying different arrangements. But I’m guessing there is a good chance that I’ll return the rig to my friend…
I second that: did you ever manage to get to the bottom of it and resolve this?
I’ve got the same problem with one of my AMD RX 580 4gb and don’t know what to do. It used to work and stopped showing this UVD error all of a sudden