mooncake-track
/
dynamo
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
dynamo/benchmarks
History
Hongkuan Zhou f3868b1f8e
fix: support config without resource limit for profile sla script (#2201) 		2025-07-30 17:07:12 -07:00
..
data_generator feat: generate random texts from hashes using lorem ipsum (#1458) 2025-06-10 22:05:21 -07:00
llm docs: add concurrency choice to the perf.sh (#1497) 2025-06-12 16:16:42 -07:00
nixl feat: add nixl benchmark deployment instructions (#2060) 2025-07-24 01:19:27 +00:00
profiler fix: support config without resource limit for profile sla script (#2201) 2025-07-30 17:07:12 -07:00
README.md feat: generate random texts from hashes using lorem ipsum (#1458) 2025-06-10 22:05:21 -07:00
pyproject.toml feat: data synthesizer based on prefix statistics (#1087) 2025-06-05 09:58:41 +08:00

README.md

Benchmarks

This directory contains benchmarking scripts and tools for performance evaluation.

Installation

This is already included as part of the dynamo vllm image. To install locally or standalone, run:

pip install -e .

Currently, this will install lightweight tools for:

  • Analyzing prefix-structured data (datagen analyze)
  • Synthesizing structured data customizable for testing purposes (datagen synthesize) Detailed information are provided in the data_generator directory.

The benchmarking scripts for the core dynamo components are to come soon (e.g. routing, disagg, Planner).