siliconshader

APPLE SILICON · MINECRAFT JAVA

A setup worth
playing with.

Browse shader recipes shared by Mac players. Compare the image settings and performance evidence, then ask an agent to check a setup against your game.

Browse setups

Inspect first. Try changes in an isolated copy of your game.

THE COMMUNITY LIBRARY

Find your next setup

New submissions share one flight route and one front-facing screenshot location. Shaders, resolution and hardware still differ, so compare those details too.

Loading setups…

Loading setup library.

START WITH YOUR OWN SETUP

Let your agent help.

Give this prompt to Codex, Claude Code, or your preferred coding agent. It starts with your launcher, mods and image-quality preferences.

Use https://github.com/fortunexbt/minecraft-apple-silicon-framework and follow silicon_shader/skills/silicon-shader/SKILL.md. Detect my Mac, launcher, existing Minecraft session, loader, mods, harness, image quality and compatible community recipes. Preserve my worlds, current settings and any game already in use; adapt one compatible setup in an isolated copy, keeping current image quality as the default floor.

For automatic performance comparison, require a supported game-control harness and the pinned disposable world. Run one matched baseline and one candidate on the same seed and fixed flight path; no manual steering or substitute terrain. Use brief in-game phase messages before measurement and after each capture. Announce the upcoming flight during warm-up; keep timed frames free of new chat messages or screenshots. Explain the candidate switch and final portrait outside measurement. Keep the repeat comparison under five minutes. If the world or supported controls are unavailable, finish setup and report performance as unmeasured instead of interrupting a running game.

After testing, take exactly one candidate screenshot from the shared front-facing view. Prepare a data-only submission with observed hardware, launcher/mods, agent/model/harness, recipe, screenshot and validated timing. Publishing is optional: publish only when this task already authorizes it; once authorized, do not ask again. Otherwise leave the validated draft local. Changes to code or toolkit still go through review.

Public comparisons use the pinned standard workload shown by silicon-shader challenge workload. A compatible harness and save may take time to prepare; no performance result is assumed until the route receipt validates.

Prefer the manual CLI?

Requires Python 3.10+ and Minecraft Java. Use a Python environment to avoid changing your system packages.

git clone https://github.com/fortunexbt/minecraft-apple-silicon-framework.git
cd minecraft-apple-silicon-framework
python3 -m venv .venv
. .venv/bin/activate
python -m pip install .

Optimize your setup ↗ · Submit an experiment ↗

Publishing uses GitHub CLI and your existing GitHub login. Data-only submissions pass automatic consistency checks before appearing in the library; performance remains contributor-reported. Changes to code and the toolkit still go through maintainer review.

A QUICK WAY TO USE THE LIBRARY

Check the recipe first.

FPS depends on the Mac, shader, scene, resolution and settings. Read the recipe and evidence before comparing two cards.

  1. Filter by your chip family, tier and memory.
  2. Standard-route cards share the pinned test scene and movement. Compare output and internal resolution, shader settings, view distance and hardware too.
  3. Ask your agent to check compatibility and prepare an isolated copy before trying a change.

Earlier-route setups remain available as recipe ideas; their FPS cannot be compared directly with the standard route. Captures are contributor-submitted, and an image-quality review is not an independent performance reproduction.