# NVIDIA engineers explain optimizing SGLang on Vera Rubin from day zero

> Original title: @nvidia's Jingfan Sun and Yang Xu share what it took to get SGLang running fast on Vera Rubin from day zero: finding bottlenecks and opti...

- Source: SGLang
- Published: 2026-11-07T00:28:50.000Z
- Bad Influence: https://badinfluence.news/items/wzsgj21nd823hxh58gfv25ow3
- Original: https://x.com/sgl_project/status/2118863208304037917

## Summary

NVIDIA's Jingfan Sun and Yang Xu describe how they got SGLang running fast on Vera Rubin from day zero. They cover finding bottlenecks and optimizing kernels and attention for large MoE models.

## Post

@nvidia's Jingfan Sun and Yang Xu share what it took to get SGLang running fast on Vera Rubin from day zero: finding bottlenecks and optimizing kernels and attention for large MoE models.

Practical lessons for anyone optimizing SGLang on new hardware.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWeyUXVacAAb3BE.jpg&mode=thumb&exp=1794268800&sig=713c4bcc85a55d29)