Versions:
core-to-core-latency-plus is a lightweight Windows system utility published by KCORES that quantifies the time it takes for data to travel from one CPU core to another. By spawning two threads, pinning each to a distinct physical core, and forcing them to exchange a flurry of compare-and-swap instructions, the program exercises the processor’s cache-coherence fabric and records the round-trip delay in nanoseconds. The resulting matrix reveals intra-die topology details—such as whether pairs of cores share an L2 cache, reside on separate CCX modules, or communicate through off-chip links—information that over-clockers, firmware engineers, and performance analysts use to explain why certain thread-to-core assignments scale while others stall. Forked from nviennot’s original core-to-core-latency, this 0.1.17 release streamlines command-line switches, adds automatic CPU model detection, and exports latency heat-maps in CSV so that benchmark databases can be populated without manual reformatting. Typical use cases include validating BIOS firmware updates that alter mesh or ring-bus frequencies, diagnosing NUMA imbalances on dual-socket workstations, and comparing Zen, Alder Lake, or Sapphire Rapids dies before virtualization or gaming workloads are deployed. Because the executable is self-contained and requires no kernel driver, it fits naturally into automated CI pipelines that must flag regressions in core-to-core latency after each micro-code or board-level change. The software is available for free on get.nero.com, with downloads provided via trusted Windows package sources (e.g. winget), always delivering the latest version, and supporting batch installation of multiple applications.
Tags: