Magnitude open-source inference engine beats Ollama by over 2x with auto-tuned kernels
nickbaumann_ · x · 2026-10-15
Magnitude is a new open-source inference engine that tunes its kernels on your actual machine to run open models as fast as your hardware allows—no new hardware needed.
- When you download a model, it spends up to a minute on your chip trying different ways to split the math (GPU thread counts, chunk sizes, data prefetch distance) and keeps the fastest
- Averaged across 11 popular local models: Mac M4 Pro vs Ollama (MLX) is 1.32x faster prefill and 2.19x faster decode; DGX Spark vs Ollama (llama.cpp) is 1.34x/2.17x
- E.g., Qwen 3.6 35B jumps from 47 to 122 tok/s on M4 Pro
- Works on Apple Silicon, NVIDIA, AMD, or plain CPU. Existing tools ship one-size-fits-all kernels because hand-tuning per chip is impractical
More from Infra
- How Microsoft optimized Windows for Nvidia's Arm-based RTX Spark AI laptops — tomwarren · 2026-10-15
- How Microsoft optimized Windows for Nvidia's Arm-based RTX Spark laptops — The Verge AI · 2026-10-15
- Global smartphone shipments plunge 22% in worst quarterly decline on record as AI demand fuels memory costs — Polymarket · 2026-10-15
- SK Hynix drops 11% again, now below 3x FY2027 earnings as AI memory selloff deepens — JOBhakdi · 2026-10-15
- ASML Sells Zero Lithography Machines into Europe as US Hosts 75%+ of AI Compute — NinaDSchick · 2026-10-15
- Nvidia quietly reclassified SPCX as 'Hyperscale', signaling deeper neocloud reliance — RihardJarc · 2026-10-15