Hey guys! Another new X570d4u-2L2T owner here. My board came with a pretty old BIOS (1.4) and while I can update to 1.7 using ASRock’s own website, the BIOS/BMC/etc. versions there look quote old still. Could someone share the credentials to ASRock’s FTP? DM is fine if you don’t want to post.
I emailed their support but they haven’t been replying. 
Same here, built over the weekend.
Started on 1.20 and updated to 1.70
I’d say it’s probably safer to carefully pass a small zip-tie between the plastic pushpin and the chipset die.
I wish somebody made a chipset cooler that used heatpipes to stick out the side of the board.
Hello All,
I’ve just bought one of these boards, the 2L2T variant, and am learning my way around it at the moment, I’ve bought it to build a virtualisation and SAN server with using Proxmox and TrueNAS. My main question is has anyone been able to get PCI pass through working using this board on Proxmox 8.3.1 with an Intel Arc GPU? My card is an A310 and I just keep getting error 43 under Windows 11 in the VM, any pointers greatly appreciated.
Here is how I keep my X550 chip cool, I have another identical fan on the X570 heatsink and just plug them into fan headers on the board (fan_1 and fan_2):
I have another ASRock board on Intel chipset, but I followed official documentation: Proxmox VE Administration Guide
It was enough to enable IOMMU and add vfio kernel modules, I didn’t blacklist anything. Also I didn’t enable ‘allow_unsafe_interrupts’ option.
And then I tick “All functions” for both GPU and GPU Audio:
It worked and after installing Arc GPU drivers it even updated it’s firmware (I was shocked tbh).
psst psst
this is where I bring up that you can, as of a few months ago, manage vfio drivers and pci device stubbing/driver overrides with the lae.proxmox ansible role. testers using intel hardware (and reports) would very much be appreciated.
Did anyone get deep C-states on those boards? I own W680-based board and am going to assume that Rack boards are not capable of achieving deep C-states at all (see).
came here looking for that, I tried everything, but never seen the platform hitting C3, literally never.
Can’t find an ASPM option in bios either.
I am not an owner of X570D4U board, but I have Intel based one. I wouldn’t achieve anything deeper than PC2 if I didn’t use hidden BIOS (Ctrl+Alt+F3 instead of F10/Del during POST). But there is still a ton of caveats.
I have a Ryzen 9 5950X and a Ryzen 7 PRO 5750G - which one would people choose to go with on this board?
APUs like 5750G have less PCIe lanes available. This is the main thing you should consider. If you don’t need iGPU, I would go with 5950X.
5750g though would consume less, also thanks to its monolithic die design
5750G is also still PCIe 3.0 not only 4 lanes less than the 5950x
Hello, just recently purchased this regular D4U version of this board and am having a frustrating problem. Is there anything on the bios end that needs to be done for gpu passthrough to function for a docker container in truenas? Currently my plex media server I can see my arc a310, but transcoding always fails with this error:
[Req#243c/Transcode] Codecs: testing hevc (decoder) with hwdevice vaapi
[Req#f6a/Transcode] Codecs: hardware transcoding: testing API vaapi for device ‘/dev/dri/renderD128’ (Intel DG2 [Arc A310])
[Req#19d6/Transcode] [FFMPEG] - Failed to initialise VAAPI connection: -1 (unknown libva error).
[Req#19d6/Transcode] Codecs: hardware transcoding: opening hw device failed - probably not supported by this system, error: I/O error.
I have tried everything under the sun with truenas and plex and all that is left to explore is bios options.
any error looking at Truenas dmesg? I’ve never used docker on Truenas Scale, but it shouldn’t be much different.
I guess you have iommu enabled, what driver are you using?
there is no such thing as gpu passthrough for docker or containers in general. that functionality is strictly for virtual machines. BIOS options (with regards to passthrough at least)/IOMMU shouldn’t be relevant here as there’s no virtualization going on.
i’m not familiar with truenas but the only way I can interpret this is that plex is running in docker on truenas, so the gpu should be loaded/bound to a gpu driver on truenas which is then shared to your docker container. the two main things I can think of are 1) the drivers in the container need to match the ones used on the host at the very least and 2) proper permissions…I’m assuming that’s along the lines of things you’ve tried?
@Cupola I don’t see any errors in the logs, though I confess I am not knowledgeable enough of truenas to know exactly what I am looking at in the logs to know if there was a clue there. As for driver version, it’s bundled into Electric Eel so I did not install them.
@lae Ahh a poor choice of technical words on my part. I am throwing stuff at the wall at this point as I don’t know why no containers can access this gpu in truenas for me. They all get this same essentially access denied error. User perms I have gone over atleast a dozen times looking for mistakes. I even went as far as to give the apps user ownership of the gpu drivers directory then mounting the drivers into the container, but I still get the exact same error.
have you tried running ffmpeg on the host/truenas itself to see if it’s at least functional outside of containers? there’s an example over at FFMpeg AV1 Encoding Using Intel Arc GPU tips? you could maybe use. that would at least rule out any motherboard/device-related issues.
there is no such thing as gpu passthrough for docker
Strictly saying this is still relevant in case of Windows. By default Docker there uses WSL2 which is uses Hyper-V as backend. So basically if you are not using Windows native containers, you end up using Virtual Machine.
Popped in an nvidia card and everything just worked. So, still not sure what the problem was with the Arc card, but just going to go with what works. Thanks for the help.
