Can It Transcode

Hardware

CPUIntelJellyfin

Intel N100

Intel N100 · Intel UHD Graphics (Alder Lake-N, 24 EU)

Another Jellyfin owner, on an N100 with 16 GB single-channel RAM and a 512 GB SSD, transcoded one 60 Mbps 4K HDR10 movie to 1080p SDR at about 15 Mbps and about 78 fps. 4K HDR to 4K SDR was about 35 fps, and Dolby Vision to 4K SDR dropped to about 28 fps. 2 other reports on this page.

Alder Lake-N chip used in a lot of cheap mini PCs (Beelink S12 Pro, Beelink EQ12, GMKtec boxes, and others). The reports here were measured on N100 machines. The write-ups did not name those retail models, so they are not listed as separate devices.

Reported results

A dash means the source did not count that kind of stream. Single-stream speeds are fps for one file, not a guess at how many clients fit.

JellyfinQuick SyncSingle-stream speed

Another Jellyfin owner, on an N100 with 16 GB single-channel RAM and a 512 GB SSD, transcoded one 60 Mbps 4K HDR10 movie to 1080p SDR at about 15 Mbps and about 78 fps. 4K HDR to 4K SDR was about 35 fps, and Dolby Vision to 4K SDR dropped to about 28 fps.

1080p
—
4K
—
4K HDR tone map
—
Power
—
Price paid
—

Not stated

Hacker News — Jellyfin on an N100 (entropicdrifter)
Notes from the source

Hacker News comment by entropicdrifter, 30 July 2024. They think Dolby Vision takes the OpenCL tone-map path rather than VPP. Hi10p anime re-encoded to H.265 with burned-in subtitles ran at 90 fps or more; they note hardware decoders do not exist for Hi10p. No simultaneous-stream count.

JellyfinQuick SyncSingle-stream speed

In Jellyfin on Linux, one 67.4 Mbps 4K HDR10 remux tone-mapped to 4K ran at about 50 fps with VPP and 57 fps without it. The same file to 1080p with subtitle burn-in ran at 82 fps. Nobody in this test stacked simultaneous streams.

1080p
—
4K
—
4K HDR tone map
—
Power
—
Price paid
—

Linux

Jellyfin Transcode Benchmarks — RandomNamer
Notes from the source

Machine: N100, 24 EU UHD graphics at 750 MHz, 8 GB DDR4. Clip JW4 is a 4K HDR10 remux. 4K-to-4K VPP tone map with burn-in was 27 fps; non-VPP with burn-in was 43 fps. A 41.5 Mbps 4K HDR Blu-ray (SW2) was about 75 fps at 4K with no subtitles and 94 fps to 1080p at 8 Mbps with burn-in. 1080p material was far above realtime (about 114–270 fps depending on the clip). The author says tone mapping plus subtitle burn-in can bottleneck the GPU 3D engine, audio transcoding capped around 5–6x, and HEVC encode was 10–20% slower than H.264. VA-API vs Quick Sync made no meaningful difference in their notes.

On the 67.4 Mbps 4K HDR remux, VPP tone mapping without subtitle burn-in was about 1 W GPU and 7 W CPU. With burn-in, about 1.8 W GPU and 10 W CPU. Those are component readings, not wall power.

JellyfinQuick SyncSingle-stream speed

The same Jellyfin clip suite on Windows 11 was slower: the 67.4 Mbps 4K HDR10 remux tone-mapped to 4K at about 30 fps, and to 1080p with subtitle burn-in at about 40 fps. A Dolby Vision clip missed realtime entirely (11 fps at 4K, 21 fps to 1080p).

1080p
—
4K
—
4K HDR tone map
—
Power
—
Price paid
—

Windows 11 23H2

Jellyfin Transcode Benchmarks — RandomNamer (Windows N100 section)
Notes from the source

Fresh Windows 11 23H2. Newer Intel drivers broke tone mapping, so the author stayed on driver 5085. 4K-to-4K with burn-in was 17 fps. Their summary line for the N100 on Windows: not able to handle 4K. SMB file pulls alone could saturate the CPU. This is still one stream at a time.