SpikeCliff effect: empirical analysis of deterministic timing discontinuities in sponge-based XOF functions

This paper presents the characterization and systematic documentation of the SpikeCliff effect, a deterministic temporal discontinuity occurring in extendable-output functions (XOF) based on the sponge construction. The phenomenon manifests as regular processing time spikes at input lengths crossing the rate boundary, caused by additional Keccak-f permutation calls. While theoretically predictable from the sponge definition, this effect has not been previously measured, documented, or analyzed in the context of XOF functions. We introduce boundary-aware profiling, a methodology for per-byte timing analysis, and demonstrate that SpikeCliff is fully deterministic, algorithmic in nature, and reproducible across platforms when dynamic frequency scaling is disabled. BLAKE3 serves as a negative control, confirming the specificity of the effect to sponge-based architectures. Practical implications for IoT systems, benchmarking methodology, and side-channel analysis are discussed.