Skip to content
sbc.compare

Jellyfin Transcoding SBC Benchmark Rankings

How many people can stream from each Single Board Computer through Jellyfin at once

10 boards tested · results from 1 to 40 (Streams, 1080p HEVC → 720p)

14 results

Top Picks

Fastest OverallOrange Pi 5B 4GBRK3588S40Streams#1 of 10 · $203Best ValueCool Pi 4B 4GBRK3588S0.22per $ (31 at $140)#2 of 10Fastest Under $100Radxa ROCK 3B 2GBRK35687Streams#3 of 10 · $51
Scenario

Full Results

02040
  1. 1
    Fastest Overall: Orange Pi 5B 4GBRK3588S · $203
    40
  2. 2
    Best Value: Cool Pi 4B 4GBRK3588S · $140
    31
  3. 3
    Fastest Under $100: Radxa ROCK 3B 2GBRK3568 · $51
    7
  4. 4
    Orange Pi 6 16GBP1 · $359
    6
  5. 5
    Radxa ROCK 5B+ 16GBRK3588 · $385
    3
  6. 6
    Raspberry Pi Compute Module 5 4GBBCM2712 · $148
    2
  7. 7
    Banana Pi CM4 4GBA311D · $104
    1
  8. 8
    Libre Computer Alta 4GBA311D · $125
    1
  9. 9
    Orange Pi Zero 3W 4GBA733 · $50
    1
  10. 10
    Orange Pi Zero 4 4GBA733 · $68
    1
02040
Simultaneous Streams, higher is better
Showing 10 boards · median 3 (4 without a Streams result not shown)Top Picks from above

Frequently Asked Questions

What is the fastest single board computer in Jellyfin Transcoding (1080p HEVC → 720p)?

The Orange Pi 5B 4GB currently tops our Jellyfin Transcoding (1080p HEVC → 720p) ranking with 40 (Streams), measured on real hardware with our standardised test setup.

What is a good Jellyfin Transcoding (1080p HEVC → 720p) score for an SBC?

Across the 10 boards we have tested, Jellyfin Transcoding (1080p HEVC → 720p) results range from 1 to 40 (Streams), with a median of 3. Anything above the median puts a board in the faster half of the market.

More benchmarks:Geekbench 6Geekbench 7Linpack7-ZipTinyMemBenchglmark2 (OpenGL)Fastest SBC Storagevkmark (Vulkan)vkpeak (Vulkan Compute)SuperTuxKartUnixBenchPassMarkcpuminerMAMEVideo DecodingOllama LLMPower ConsumptionTest methodology

About Jellyfin Transcoding SBC Benchmark Rankings testing

We run Jellyfin's own FFmpeg build (jellyfin-ffmpeg) on four transcodes, adding simultaneous streams until the slowest one can no longer keep up with real time. The result is how many people could watch at once. Each board is tested on its CPU (libx264 at the veryfast preset) and, where Jellyfin supports it, on its hardware video engine (Intel Quick Sync or VA-API, Rockchip MPP, V4L2) using the same decode, scale and encode pipeline Jellyfin builds for that hardware; the ranking uses whichever got more streams, and the table shows which. Clips play from the board's own boot drive, as a real server's library would. Where the board's image lacked the drivers that hardware path needs, the test installed them, and that is marked. Each scenario has its own ranking; switch between them above. More streams is better.