DeepSeek V4.1 Flash Maya-S: 2x3090 vs 7x3090 VRAM scaling

Running the same Maya-S quant, 2x3090 with 128GB RAM and NVMe is notably slower than 7x3090 with 256GB, but 48GB of VRAM is already enough to run the model.

Axitek

@axitekArchived 6 days agoOriginally posted 2026.10.25

DeepSeek V4.1 Flash Maya-S: 2x3090 vs 7x3090 VRAM scaling

Axitek @axitekOriginally posted 2026.10.25

Open original on X

It helps to have more VRAM - but don't discount RAM and NVME support. Same model, DeepSeek-v4.1-Flash-1017:Maya-S on Maya: 🟠 Orange: 2x3090+128gb+nvme 🔵 Blue 7x3090+256gb+nvme Speed difference is significant, but even 48gb vram is enough to run DSF. https://github.com/mw00/project-maya

x.com/axitek/status/2114163754950902093Archived 2026.10.25 11:59

Open original on X