.. meta::
  :description: Index of CVS inference test configuration schemas, covering ATOM, vLLM, SGLang, and xDiT benchmark suites for AMD GPU clusters.
  :keywords: CVS, inference, vLLM, ATOM, SGLang, xDiT, ROCm, GPU, AMD Instinct, LLM, benchmark, JSON, configuration

************************************************************
Cluster Validation Suite (CVS) inference test configuration schemas
************************************************************

JSON configuration schemas for inference and serving benchmarks under ``cvs/input/config_file/inference/``.

- :doc:`ATOM (vLLM Benchmarking) </reference/configuration-files/inference/atom>` — ATOM inference benchmarks
- :doc:`vLLM Inference </reference/configuration-files/inference/vllm>` — vLLM serving throughput and latency
- :doc:`SGLang Inference </reference/configuration-files/inference/sglang>` — single-node, distributed, and disaggregated LLM serving
- :doc:`xDiT Inference </reference/configuration-files/inference/xdit>` — FLUX.1/FLUX.2 text-to-image and WAN 2.2 image-to-video

How to run these suites: :doc:`/how-to/test-suites/inference/index`.