[Bug] difference of kv-cache-prefixing between vLLM and sglang

[chenchunhui97]

Checklist

• 1. I have searched related issues but cannot get the expected help.
• 2. The bug has not been fixed in the latest version.
• 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.
• 4. If the issue you raised is not a bug but a question, please raise a discussion at https://github.com/sgl-project/sglang/discussions/new/choose Otherwise, it will be closed.
• 5. Please use English, otherwise it will be closed.

Describe the bug

no bug. I am just wondering the difference of kv-cache-prefixing between vLLM impletention and SGLang implementation.

vLLM use hash to store and verify cached token:

SGLang uses RadixAttention, so what is the difference? I found SGLang is faster than vLLM, why SGLang RadixAttention is faster than vLLM KV-Cache-prefixing?

Reproduction

not available

Environment

//

[zhyncs]

vllm-project/vllm#2614