Is an $800 128GB RAM Rig Worth It for Running MoE LLMs? Reddit Debates
Juan_Phoenix7 · reddit · 2026-11-07
A Redditor asks: with a 3090 (24GB VRAM) + 32GB RAM, should they spend $800 on a used Xeon E5-2680 v4 + 128GB DDR4 machine to slot the 3090 in and run bigger MoE models via RAM offload (e.g. Qwen Flash 3/4-bit + Strata)? The alternative is $350 to bump the current machine to 64GB.
Key concerns: the Xeon is dual-channel only (not quad), a likely bandwidth bottleneck; the board caps at 128GB; an 850W PSU adds cost; future expansion would mean a second-hand 3060 12GB. Use cases span image/video generation, transcription, translation, and coding. The core question: is running MoE in RAM actually worth it, or is there a better path.
More from Infra
- Intel CEO: Agentic workloads could 5x server CPU TAM to $200B in 5 years — Beth_Kindig · 2026-11-07
- Personal AI agent rig: local 40 TOPS Arduino board + Tailscale/Cloudflare/Azure — unixterminal · 2026-11-07
- Big Tech capex up 5-8x since Q1 2023 while Apple spends 20% less; only Mag7 name beating Nasdaq — a16z · 2026-11-07
- Prefill-dominated endpoints coming for extraction and summarization workloads, researcher predicts — sh_reya · 2026-11-07
- SurrealDB 4 ships as 'the database for every agent' with PostgreSQL/MySQL support — mattturck · 2026-11-07
- India pledges indigenous GPU in 24 months; analyst says EUV access is the real clock — SumitGup · 2026-11-07