Benchmark NVIDIA RAG Blueprint servers with config-driven profiling and aiperf load tests.
npx skills add https://github.com/sayalinvidia/sayali-skills-test --skill rag-perf
Please help me install this Agent Skill. Skill: rag-perf Source: https://github.com/sayalinvidia/sayali-skills-test/tree/main/skills/rag-perf Command: npx skills add https://github.com/sayalinvidia/sayali-skills-test --skill rag-perf
Benchmark a deployed NVIDIA RAG Blueprint server by driving profiling data and optional aiperf load tests via a config-driven CLI.
Run rag-perf with a YAML config to measure performance against your RAG server and generate a summarized report of TTFT and throughput.
High-intent search queries and answers about installing and using this skill.
Benchmark RAG performance by running config-driven profiling against a live RAG server to measure latency metrics like TTFT and throughput. Use YAML presets to drive load tests and generate summarized reports for retrieval tuning.
Run load tests on an NVIDIA RAG Blueprint server by using a CLI to drive optional aiperf load tests and profiling data. This config-driven approach quantifies bottlenecks across concurrency, vdb_top_k, and reranker_top_k knobs.
Sweep concurrency and reranker_top_k configurations by running grid sweeps via YAML presets against your RAG server. This identifies bottlenecks and produces reproducible results for capacity planning and SLA verification.
To run RAG latency benchmarks, you need a local RAG server running on port 8081, uv installed, Python 3.11+, and the rag-perf package installed in a virtual environment to execute the profiling scripts.
You can use synthetic data or file-based inputs for RAG throughput benchmarking. The config-driven CLI supports both input types to measure performance metrics and generate detailed reports.
RAG benchmarking tests require YAML presets to ensure reproducible results when sweeping configurations like vdb_top_k and concurrency. These presets define the testing parameters for consistent latency and throughput measurement.
rblake2320Orchestrate filesystem-based RAG benchmarking with corpus/train.json and evaluate_rag.py.
congiulucDesign and execute load tests for backend services using k6 and Locust.
rblake2320Benchmarks NVIDIA Blueprint servers with per-stage Rapid RAG performance analysis.
jmdueaIdentify and resolve performance bottlenecks in Rust applications.
