AI/ML●●●Banger
RunAnwhere – Faster AI Inference on Apple Silicon
Custom Metal shaders beat llama.cpp and MLX—1.67x faster on M4 Max.
WizardrySlickZero to One
sanchitmonga22
2401532mo ago
Back to browse
GitHub Repository
Preflight checks and collision prevention for local AI inference workloads
3 starsPython
Fleet Watch – preflight guard for local AI inference on Apple Silicon
by cjarchivist·Apr 14, 2026·2 points·1 comment
AI Analysis
●●●BangerSolve My ProblemBig BrainNiche Gem
GPU working set estimation catches memory overcommit before your 7B model swaps to SSD.
Strengths
- •Auto-discovery via lsof/ps means zero manual registration of running processes.
- •Working set calculation includes KV cache, activations, and framework overhead.
- •SQLite registry prevents port collisions and repo lock conflicts between agents.
Weaknesses
- •Apple Silicon only — Windows and Linux GPU users get nothing.
- •Niche audience: only matters if you run multiple local inference engines.
Category
Target Audience
Developers running local LLMs on Apple Silicon
Similar To
NVIDIA DCGM · Run:AI · Kubernetes device plugins
Similar Projects
Developer Tools●●●Banger
oMLX – Native Mac inference server that persists KV cache to SSD
SSD-cached KV blocks dodge re-prefill tax on context shifts—Claude Code now viable locally.
Solve My ProblemWizardryShip It
jundot
103mo ago
AI/ML●●Solid
Flotilla – An orchestrator for persistent agent fleets on Apple Silicon
File-based agent communication via MISSION_CONTROL.md is elegantly simple.
Ship ItNiche Gem
ursushoribilis
112mo ago
Developer Tools●●●Banger
GhostVM – macOS Containers on Apple Silicon
Native macOS VMs with APFS snapshots beat Docker for agent isolation.
WizardrySolve My ProblemNiche Gem
CarolineWang
702mo ago
Infrastructure●●●Banger
IEEE-754-Conformant FP64 on Metal (Apple Silicon)
Bit-exact f64 emulation on Metal GPUs where Apple's native double support is missing.
WizardryNiche Gem
guyfischman
1127d ago
AI/ML●●●Banger
Bonsai 1.7B ternary model at 442T/s on M4 Max
Autonomous agent wrote custom Metal kernels boosting decode speed 42% over upstream llama.cpp.
WizardryDark Horse
hhuytho
1331mo ago