Reproduce “Exploiting What NVIDIA Profilers Hide: Fast and Accurate Analytical GPU Performance Modeling” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Ziwen Wang, Qi Luo, Kunlin Li, Changxi Liu, Dongsheng Wang, Yun Chen

ForeSight predicts execution time for GPU configurations by reusing measured fixed-latency stalls and recomputing queueing stalls from pressure on individual compute and memory units. UnitSight exposes the per-unit counters needed for this modeling. Across 50 workloads on real configurations of nine NVIDIA GPUs, ForeSight achieves 8.3% MAPE versus Habitat’s 14.1%; on RTX 4090, Habitat is more accurate.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.