Benchmark board

TargetBridge benchmarks: real 5K presets, HEVC modes and latency notes

This page collects the real TargetBridge presets already documented by the project, from 1440p 60 up to 5K 48 HEVC. Where hard comparison data is still missing, the page says so clearly instead of pretending the benchmark set is more complete than it really is.

Current board

What is documented today, and what still needs proper testing.

The rows below separate documented presets from numbers that still need a same-hardware comparison across TargetBridge, AirPlay and Luna Display. That keeps the page useful now without dressing up missing data as if it were already settled.

Metric TargetBridge AirPlay Luna Display Source note
Highest documented preset 5K · 5120 x 2880 @ 48 · HEVC No comparable 5K-first preset is documented here. Astropad advertises full 5K iMac support. This row shows what is publicly documented today, not the result of a like-for-like lab test.
Sharp-text preset Crisp · 3840 x 2160 @ 48 · HEVC No similar text-focused preset is documented here. No comparable preset table is published here. This is the preset most people will care about for text, coding and general desktop work.
Motion preset Smooth+ · 3200 x 1800 @ 60 No same-hardware 60 fps comparison is published here. No same-hardware 60 fps comparison is published here. This is the row to cite when talking about cursor feel, scrolling and window movement.
Lower-latency preset Smooth · 2560 x 1440 @ 60 No comparable low-latency preset is documented here. No comparable low-latency preset is documented here. Think of this as the safest public baseline before moving to higher resolutions.
Average latency The repo does not publish a same-hardware latency figure in milliseconds yet. No same-hardware latency figure is published here. No same-hardware latency figure is published here. Milliseconds only become useful when sender, receiver, cabling and macOS versions all match.
CPU / GPU load The repo does not yet include a public sender/receiver load chart. No comparable load chart is published here. No comparable load chart is published here. When this data is published, sender and receiver load should stay separate so the thermal cost is easy to read.

Public reports

What public test reports already say

These are not official lab benchmarks. They are public GitHub reports that add concrete context about 5K HEVC behavior and Network Link limits on real hardware.

GitHub issue #11

Issue #11: 5K HEVC sender FPS ranged from 2 to 16 on one public test setup

A public GitHub issue reports 5120 x 2880 @ 48 HEVC running between 2 and 16 sender FPS on a MacBook Pro M3 Pro to 2020 Intel iMac pair, with severe flicker that made the setup unusable on that hardware combination.

Read source report

GitHub issue #47

Issue #47: 5K over Network Link worked, but Wi-Fi became the bottleneck

A public GitHub issue reports 5K starting over both WiFi-to-WiFi and WiFi-to-LAN paths. The same report says the local network was not fast enough for good speed, which supports the project recommendation to prefer Thunderbolt Bridge first.

Read source report

Methodology

How to publish credible numbers

Every new benchmark drop should read like a reproducible experiment. That keeps the numbers comparable and makes public reports more useful for everyone evaluating the project.

Benchmark checklist

  • Always publish sender model, receiver model, macOS versions and cable/adapter path.
  • Run each benchmark on the same desk setup before comparing TargetBridge, AirPlay and Luna Display.
  • Report mean latency, 95th percentile latency, average FPS and dropped-frame observations.
  • Split CPU and GPU load by sender and receiver so readers can compare thermals and power draw.
  • Keep raw logs or screen recordings linked from GitHub discussions whenever possible.

Suggested public benchmark set

  • Mirror mode at a stable 1440p / 4K preset
  • Extended desktop at a stable 1440p / 4K preset
  • 5K HEVC run with sender model, receiver model and exact cable path recorded
  • CPU and GPU snapshots for sender and receiver while the same scene is running
  • AirPlay and Luna runs performed on the same pair of Macs when those paths are available

Benchmark contribution flow

Share your hardware run and help the benchmark board grow.

Include exact sender and receiver models, macOS versions, cable or adapter path, and raw observations. GitHub Discussions and Issues are the right place to share that evidence before it gets promoted on this page.