Reproduce “Pickle: Precise, Flexible Cross-Core Last-level Cache Data Prefetching for Irregular Memory Accesses” 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
Hoa Nguyen, Pongstorn Maidee, Jason Lowe-Power, Alireza Kaviani
Pickle is a hardware-managed prefetching system for irregular memory accesses. It lets programmers supply kernels and hints describing access chains, which a separate engine follows to fetch data into the last-level cache. In cycle-level simulation of graph workloads, Pickle achieved a 1.49× geometric-mean speedup over no prefetching, with 2% additional DRAM traffic; results are simulated and workload-dependent.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.