![]() ![]() It requires Windows which is a shame but the software works very well. In the second VM I run Blue Iris, my NVR software of choice. Once that is done, install intel-gpu-tools (assuming you're on Ubuntu this is available in the repos) and monitor the utilisation with sudo intel_gpu_top. Once you've got the container up and running you'll need to tell Plex to use hardware transcoding in settings. The key here is devices: /dev/dri:/dev/dri. version : "2" services : plex : image : plexinc/pms-docker container_name : plex network_mode : host devices : - /dev/dri:/dev/dri volumes : - /mnt/tank/appdata/plex/config:/config - /mnt/storage:/data environment : - PUID=1000 - PGID=1000 - TZ=America/New_York restart : unless-stopped Once you've enabled IOMMU verify you've done so correctly like so: I did an episode of Linux Unplugged where we went into full details on passthrough.įull instructions are provided on the to enable passthrough. For reference my hardware is an Intel i5 8500 and an AsRock Z370-I motherboard. PCI passthrough requires a very particular set of hardware although things now are easier than they used to be a few years ago. Less power draw means less heat and lower electricity bills - what's not to like?! Setting up PCI passthrough (on Proxmox) ¶ On top of that, Quick Sync can handle in excess of 20 1080p streams and about 5 4k transcodes in my testing. ![]() Quick Sync is capable of a 4k transcode at 35w! This is a 10x reduction from the Dual Xeon setup I'd been using previously. This was a legitmate "holy shit!" moment for me. If you needed even more convincing, take a look at the following numbers and see just how incredibly power efficient Quick Sync is compared to software based transcoding with Plex. Power Usage ¶įull details in episode 34 of Self-Hosted. If you're not convinced yet, take a look this video from Wendell who explains in a lot of detail why this technology is so cool! Intels rumoured upcoming XE graphics card might support it but until then, we'll have to make do with the iGPU in your CPU. In our case, we'll use the Quick Sync portion of the iGPU to transcode Plex H264 streams in one VM and Blue Iris streams in another. Take the iGPU and give a VM, or multiple VMs, a slice of that graphics chip to do with whatever it wants. That is precisely what GVT-g permits us to do! Wouldn't it be nice if we could slice up 1 GPU and use it with multiple VMs at once? Passing through an entire GPU is very useful for specific tasks but isn't a very efficient use of resources. This blog post which was helpful in my research for this topic. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |