~ / blog / series / 洋垃圾跑大模型
❯ ls ~/blog/series/洋垃圾跑大模型
5 posts
- partdatetitle
- 12026-07-21[Junk-Tier Big Models #1] Running a 119B MoE at 74 tok/s on Three 2080 Tis
A US$1.6k junk EPYC + 3× 2080 Ti 22G box (66G VRAM) runs a quantized 119B MoE at 74 tok/s. Expert offload to RAM costs about 2.4× decode — plus a multi-card OOM gotcha.
- 22026-07-22[Junk-Tier Big Models #2] Running Poolside Laguna S 2.1, a 118B Coding MoE, on ONE 22GB 2080 Ti
Poolside Laguna S 2.1, a 118B-A8B coding MoE, on one 22GB 2080 Ti via CPU/GPU hybrid offload plus a companion DFlash speculative-decoding draft at ~29 tok/s; attention-Q8 saves ~2.45 GiB, +7% decode.
- 32026-07-29[Junk-Tier Big Models #3] A 284B MoE on ONE 2080 Ti — and it beats my DGX Spark
DeepSeek-V4-Flash is 284B. It decodes at 17.4 tok/s on a single modded 22GB 2080 Ti — faster than the DGX Spark I serve it on. Card count barely matters, and MTP speculative decoding dies on unimplemented runtime, not a missing GPU.
- 42026-08-02[Junk-Tier Big Models #4] A frontier-class open model on hardware you already own: DeepSeek-V4-Flash-0731 on one 22GB 2080 Ti
How to run DeepSeek-V4-Flash-0731 (91GB, 284B MoE) on a single modded 22GB 2080 Ti at 16.5 tok/s and 1M context — including a formula for picking -ncmoe.
- 52026-08-18[Junk-Tier Big Models #5] Qwen3.8-27B on a 2018 22GB card: 30 tok/s, and it one-shot a 3D scene
A modded 22GB RTX 2080 Ti runs Qwen3.8-27B at 30-31 tok/s with 128K context. The full llama.cpp command, the MTP flags, and the thinking-budget trap.