Bought a second RTX 3090 for local AI but see no benefit yet—what are dual-3090 setups for?
BlueSky4200 · reddit · 2026-09-27
The author runs a local AI server with 96GB DDR5 and one RTX 3090, already serving krea2, qwen image 2.1, minimax h3, ltx 2.5, qwen 3.6, qwen 3.8 q4, and even qwen 3.8 Flash next. After buying a second 3090, they as a single user see no immediate benefit and are asking the community what awesome things dual-3090 setups enable.
More from Infra
- ComfyUI user weighs zram weight offloading: zstd decompression may bottleneck the inference hot path — Agreeable_Return2632 · 2026-09-27
- Splash 1.1.0 adds GGUF quants and MLX import, runs Qwen3 27B at 50 t/s on M5 Pro — wojtek15 · 2026-09-27
- M5 Pro 64GB hits 50 t/s with Qwen3 27B as Splash 1.1.0 adds GGUF and MLX support — wojtek15 · 2026-09-27
- Databricks targets 100x cheaper warehouse AI functions for all DB vendors — sh_reya · 2026-09-27
- Ollaya runs decision models locally: 5 answers in 8–10 ms on an RTX 4090, no token generation — petrusenko_max · 2026-09-27
- DHH rewrites ttfx in asm with AI, hitting 8x mean speedup on an Intel 135U — olcan · 2026-09-27