ComfyUI model loading stuck at 400 MB/s even after upgrading to NVMe
C-h_A_o-S · reddit · 2026-11-21
User on Ryzen 5 5600x / 32GB / RX 7900XTX 24GB moved from SATA (400MB/s) to a 1TB NVMe on a PCIe 2 adapter card, reaching 1600MB/s. llama.cpp loading got faster, but ComfyUI stays at 400MB/s despite a workflow that swaps models constantly — asking why it isn't the expected 3x+ faster.
More from Infra
- How do you run big batch GPU jobs — thousands of files, frames, or prompts? — mrtonymat · 2026-11-21
- Inkling 2 Mini hits ~200 tok/s on a single RTX 4090 with FP8 + NVFP4 quantization — eesahe · 2026-11-21
- Inkling 2 Mini runs at 125 tok/s on a MacBook, claims best-in-class at its size — simonguozirui · 2026-11-21
- Same RTX 5090, 4x throughput: Surogate hits 2,765 tok/s vs llama.cpp's 683 with 8 users — AIFlow_ML · 2026-11-21
- Compound's AI slowdown scenario: NVDA down ~45% if revenue growth halves and multiples revert — tengyanAI · 2026-11-21
- DatologyAI open-sources Zephon, a data loader with elastic determinism for LLM training — josh_wills · 2026-11-21