Metal (Apple)
Definition pending
We've cataloged "Metal (Apple)" but haven't written a full definition yet. Definitions are hand-curated rather than auto-generated, so it takes time to cover every term.
Want this one prioritized? Email us and we'll bump it.
Practical example
Metal is Apple's GPU acceleration framework — it's what makes LLMs run fast on Macs with M-series chips. Without Metal, inference runs on CPU only (or ANE, which is slower for LLMs). Metal enables the GPU cores on M1/M2/M3/M4 to run matrix multiplies at near-CUDA speeds, using the unified memory architecture.
Workflow example
Metal acceleration in Ollama: enabled by default on macOS. Check: ollama run --verbose shows "Metal" in GPU layers. Mac performance: M1 Max (32-core GPU, 400 GB/s bandwidth) ≈ RTX 3060 for LLM inference. The unified memory is the killer feature — a 64 GB MacBook can run 70B Q4 entirely in "VRAM" (unified memory shared between CPU and GPU). No CPU offload needed.