domains / families / science / nemo-mbridge-perf-moe-dispatcher-selection

nemo-mbridge-perf-moe-dispatcher-selection

separate

Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.

When to use

first correctness bring-up
small EP configurations
debugging communication regressions
Hopper or B200 deployments
cross-node EP is clearly visible in profiles
you want a mature intermediate step before testing HybridEP
GB200 or GB300 NVL72 systems
large EP degrees

Skill page

Source

Repository	Path
NVIDIA/skills	`skills/nemo-mbridge-perf-moe-dispatcher-selection/SKILL.md`