Multi-GPU & Clusters
Splitting a model across cards, boxes and idle machines.
Past one card the problems change: layer splits, tensor parallelism, interconnect bandwidth, and the economics of buying two cheaper cards instead of one large one. These pieces cover dual-GPU rigs, home clusters built from idle hardware, and the cases where a cluster is the wrong answer.
NVIDIA PAIR: Idle Home PCs as a Local AI Cluster
NVIDIA PAIR explained: what the open-source personal AI router does, which devices work, and how much time it saves when your home PCs share AI work.
- Sep 4, 2026 · 8 min readLocal LLM GPU Buying Guide (2026)What to buy to run local LLMs in September 2026, chosen by VRAM tier with current US prices, used-market alternatives and honest tradeoffs.
- Sep 4, 2026 · 3 min readMac Mini M4 Clusters for Local LLMsThe Apple-silicon route to local inference: why M4 Macs grew 10%+ in 2026, what the multi-unit "cluster" story is, and when it beats a GPU box.
- Sep 6, 2026 · 3 min readDual RTX 3090 70B Build (48GB, 2026)Two used 24GB cards give 48GB for 70B-class local inference. What it costs at the 2026-09-28 snapshot, and sourced tok/s.
- Sep 4, 2026 · 2 min readRun 70B Models on Two Used RTX 3090sThe 48GB route: running 70B-class models at usable speeds on a pair of used RTX 3090s for around $2,600. Pricing as of 2026-09-28.