Vectorized and Parallel Computation of Large Smooth-Degree Isogenies using Precedence-Constrained Scheduling

Strategies and their evaluations play important roles in speeding up the computation of large smooth-degree isogenies. The concept optimal strategies for such was introduced by De Feo et al., virtually all implementations isogeny-based protocols have adopted this approach, which is provably single-core platforms. In spite its inherent sequential nature, several recent works studied ways isogeny exploiting rich parallelism available vectorized multi-core One obstacle to taking full advantage parallelism, however, that al.’s are not necessarily environments. To illustrate how speed parallel can be improved at strategylevel, we present two novel software utilize a state-of-the-art evaluation technique, called precedence-constrained scheduling (PCS), presented Phalakarn with our proposed crafted these Our first implementation relies only on provided processors. second targets processors supporting latest generation Intel’s Advanced Vector eXtensions (AVX) technology, commonly known as AVX-512IFMA instructions. better handle computational concurrency associated PCS, equip both extensive synchronization techniques. outperforms Cervantes-Vázquez al. yielding 14.36% reduction execution time, when targeting platforms two- four-core implementation, equipped four cores, achieves 34.05% time compared Cheng CHES 2022.