Skip to main content

Installation

From a clean machine to a working LLMServingSim install. Three short pages, the first two cover everything most users need, the third is optional.

Prerequisites

Quick check before you dive in: OS, Docker, NVIDIA Container Toolkit, optional Hugging Face token.

Simulator setup

The path everyone needs. Clone the repo, launch the simulator container, build ASTRA-Sim, verify with a smoke test.

vLLM setup (optional)

Only needed if you want to profile new hardware, run vLLM benchmarks, or validate the simulator. Docker or bare-metal.

If you only plan to run pre-profiled simulations on the bundled RTXPRO6000 hardware, you can skip vLLM setup entirely.