> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/cuvs/llms.txt.
> For full documentation content, see https://docs.nvidia.com/cuvs/llms-full.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/cuvs/_mcp/server.

# cuVS Bench

cuVS Bench is a reproducible benchmarking tool for ANN search implementations. It supports GPU-to-GPU and GPU-to-CPU comparisons, and helps capture index configurations that can be reproduced across on-prem and cloud hardware.

Use cuVS Bench to compare build time, search throughput, latency, and recall; find useful parameter settings for recall buckets; generate consistent plots; and identify optimization opportunities across index parameters, build time, and search performance.

For dataset file formats, conversion utilities, and ground-truth generation, see [Benchmark Datasets](/cuvs/user-guide/benchmarking-guide/cu-vs-bench-tool/datasets).

For custom benchmark execution paths and backend integrations, see [Backends](/cuvs/user-guide/benchmarking-guide/cu-vs-bench-tool/backends).

For setup, see [Installation](/cuvs/user-guide/benchmarking-guide/cu-vs-bench-tool/installation). To run benchmark workflows, see [Usage](/cuvs/user-guide/benchmarking-guide/cu-vs-bench-tool/usage). To compile the benchmark executables locally, see [Build from Source](/cuvs/user-guide/benchmarking-guide/cu-vs-bench-tool/installation#build-from-source).