Would there be a significant difference between a Intel Arc A310 and Nvidia RTX for transcoding?

[Betonhaus 18]

I'm currently using an Arc A310 but I might need to play musical chairs with my server GPUs and wanted to know how a Nvidia 3060 or similar compared for transcoding performance 

[Luke 42077]

For high-volume or high-resolution transcoding, the Intel Arc A310 is superior due to its dedicated media engine, which can handle more simultaneous transcodes, especially with codecs like AV1. In contrast, the Nvidia RTX 3060, while a more powerful overall GPU, is less efficient for this specific task because its media encoder (NVENC) is less optimized for high-stream scenarios and it consumes more power. 

Which one should you choose?

• Choose the Arc A310 if: Your primary use is a media server (like Plex or Jellyfin) and you need to handle multiple 4K transcodes or want to transcode to AV1 to save space.
• Choose the RTX 3060 if: You need a card for a mix of gaming and transcoding, or for professional applications like video editing where its raw power and broader software support are more beneficial. 

[Betonhaus 18]

fair enough, i have room for two GPUs in my system and I'm trying to juggle three things:

1. video transcoding for Emby and so on

2. a Windows VM that needs 3d video acceleration to run Windows specific apps like SolidWorks

3. a Games on Whales instance that lets me play games over the local network to computers that do not have enough power to run games locally (or cannot due to being a Mac or something)

right now Windows VMs cannot share a GPU with the host, it needs it's own. virtio-gpu driver support is still being worked on. that leaves my most GPU intensive task and least GPU intensive task sharing a GPU.

Right now I have an Intel Arc A310 and a Nvidia Quadro K6000, with the latter likely being the one i replace with a Nvidia 3060 or something equivalent. I could problably find an Intel Arc B580 to replace it and give Windows the arc A310...

[Luke 42077]

Please keep us posted on what you do. Thanks.