Deploying LLMs efficiently requires testing hundreds of serving configurations, but evaluating each one on a GPU cluster takes hours and costs thousands of dollars. Discrete-event simulators are faster and cheaper, but they require re-implementing the serving system's control logic -- a burden that compounds as frameworks evolve. We present Revati, a time-warp emulator that enables performance modeling by directly executing real serving system code at simulation-like speed. The system intercepts CUDA API calls to virtualize device management, allowing serving frameworks to run without physical GPUs. Instead of executing GPU kernels, it performs time jumps -- fast-forwarding virtual time by predicted kernel durations. We propose a coordination protocol that synchronizes these jumps across distributed processes while preserving causality. On vLLM and SGLang, Revati achieves less than 5% prediction error across multiple models and parallelism configurations, while running 5-17x faster than real GPU execution.
@article{arxiv.2601.00397,
title = {Revati: Transparent GPU-Free Time-Warp Emulation for LLM Serving},
author = {Amey Agrawal and Mayank Yadav and Sukrit Kumar and Anirudha Agrawal and Garv Ghai and Souradeep Bera and Elton Pinto and Sirish Gambhira and Mohammad Adain and Kasra Sohrab and Chus Antonanzas and Alexey Tumanov},
journal= {arXiv preprint arXiv:2601.00397},
year = {2026}
}