# SGLang splits sparse attention memory across 8 GPUs with no accuracy loss

> Original title: Sparse attention only reads the 2,048 most relevant tokens, but it still has to keep the whole history in memory.

- Source: SGLang
- Published: 2026-11-04T16:26:53.000Z
- Bad Influence: https://badinfluence.news/items/navqwc6wpvqw5u2zqkv6481xa
- Original: https://x.com/sgl_project/status/2118017147544945124

## Summary

SGLang says it splits sparse attention's full-history memory across 8 GPUs using decode context parallelism while still selecting the same 2,048 tokens. The post says this allows up to 8x more room for history with no accuracy loss.

## Post

Sparse attention only reads the 2,048 most relevant tokens, but it still has to keep the whole history in memory. 

We split that memory across 8 GPUs with decode context parallelism, and still find the exact same 2,048 tokens. That's up to 8x more room for the history, with no accuracy loss.