Deep Dive: Hybrid Deployment of vLLM and Ray Across Heterogeneous Hardware

jwhh91 · reddit · 2026-07-26

A user discusses deploying large models using vLLM and Ray in a complex heterogeneous setup: two DGX Sparks and an x64 box with an RTX 5090, networked via ConnectX-7.

The goal is to run a 300B (150GB) MoE model keeping TP=2 on the Sparks, while shifting attention heads and hot experts to the 5090's GDDR7. They ask what it would take to augment Ray in vLLM to achieve this compute shifting. The author also shares positive experiences using ConnectX-7 and NVME RAID1 for model weight loading at 20 Gb/s.

Original post →

More from Infra

Infra channel →