26 July 2026 to 1 August 2026
University of Maryland, College Park
US/Eastern timezone

PerfAdvisor: An LLM-based Agent for Diagnosing Bottlenecks from GPU Systems Profiles

27 Jul 2026, 15:00
20m
Crossland (Adele H. Stamp Student Union)

Crossland

Adele H. Stamp Student Union

3972 Campus Dr, College Park, MD 20742
Contributed talk Software development and machines Software development and machines

Speaker

Leon Hostetler (Indiana University)

Description

We introduce PerfAdvisor, a lightweight Python package that uses an LLM-driven agent loop to diagnose GPU performance bottlenecks from supercomputer profiles. Modern HPC profiles contain millions of events across kernels, MPI calls, memory transfers, and idle gaps, making manual diagnosis slow and expert-intensive. PerfAdvisor automates this process by producing ranked, evidence-grounded bottleneck hypotheses and concrete mitigation suggestions in minutes. PerfAdvisor supports profiles from AMD ROCm Systems Profiler and NVIDIA Nsight Systems. It segments each profile into execution phases, computes summary metrics, and gives an LLM agent access to local, read-only tools for top-kernel summaries, idle-gap histograms, MPI and memory-transfer summaries, NVTX/ROCTx summaries, per-rank imbalance scoring, and arbitrary SQL queries. The agent iteratively gathers evidence until it returns ranked diagnoses with suggested fixes. Multiple LLM backends are supported, and prompt caching reduces token cost by up to 80%. We validate PerfAdvisor on NERSC Perlmutter using a CUDA benchmark suite with eight known performance pathologies and on OLCF Frontier using an analogous HIP benchmark suite. PerfAdvisor identifies the primary bottleneck in all eight cases with strong coverage of ground-truth mitigation strategies. In a MILC/QUDA lattice-QCD case study on NERSC Perlmutter, PerfAdvisor correctly diagnosed host-staged halo exchange and recommended GPUDirect RDMA, which reduced wall time by 43%.

Author

Leon Hostetler (Indiana University)

Presentation materials