Validating Distributed LLM Serving Benchmarks with NVIDIA srt-slurm, SLURM Recipes, Parameter Sweeps, and Pareto Analysis

NVIDIA srt-slurm converts declarative YAML configurations into reproducible SLURM benchmark workflows for distributed LLM serving. The framework supports parameter sweeps, typed Python API interactions, and Pareto analysis of throughput versus latency. A disaggregated prefill-and-decode deployment for DeepSeek-R1 is modeled using built-in and custom recipes.