{"id":3005,"date":"2026-10-30T15:09:03","date_gmt":"2026-10-30T15:09:03","guid":{"rendered":"https:\/\/smartcryptosecurity.com\/?p=3005"},"modified":"2026-10-30T15:09:03","modified_gmt":"2026-10-30T15:09:03","slug":"efficient-simd-shake-through-cost-aware-permutation-scheduling","status":"publish","type":"post","link":"https:\/\/smartcryptosecurity.com\/?p=3005","title":{"rendered":"Efficient SIMD SHAKE through Cost-Aware Permutation Scheduling"},"content":{"rendered":"<p>Lattice-based cryptography has rapidly emerged as a foundational pillar for next-generation security, particularly as the cryptographic community transitions toward quantum-resistant algorithms. These modern schemes heavily rely on SHAKE\u2014an extendable-output function derived from the Keccak hash function family\u2014to dynamically expand seeds into long byte strings. These expanded strings are critical for various core operations, including matrix generation and statistical sampling. Because applications frequently require multiple hash streams to be generated simultaneously, developers routinely employ SIMD implementations. SIMD architectures allow a single processor instruction to execute operations across multiple data lanes in parallel, making them theoretically ideal for handling concurrent hash streams.<\/p>\n<p>Despite their theoretical promise, standard SIMD implementations of Keccak often fail to achieve optimal resource utilization. In real-world cryptographic executions, individual hash streams frequently vary in length. Furthermore, complex algorithms introduce stringent input dependencies, meaning that certain data streams cannot be processed until intermediate results from other calculations become available. These disparities create significant operational friction. When different lanes within a SIMD register finish processing at different times or must wait for dependent inputs, parts of the parallel execution vector sit idle, leading to wasted computational cycles and degraded throughput. <\/p>\n<p>To resolve this inefficiency, the research team developed a cost-aware scheduling methodology designed to dynamically manage Keccak permutations across multiple hash streams during a single execution cycle of an overarching cryptographic algorithm. Rather than treating parallel streams as rigid, uniform blocks, the newly proposed method actively monitors stream states. It groups streams that are ready for processing while strategically dividing long streams into manageable segments. These segments can then be dynamically regrouped to ensure that the SIMD lanes remain fully saturated with active computational work as often as possible.<\/p>\n<p>The core of this new optimization framework is a comprehensive mathematical model that accounts for the fundamental costs associated with cryptographic execution. The model evaluates permutation calls, state movement across memory or registers, initialization overhead, and input preparation costs. By continuously weighing these factors, the scheduling algorithm makes precise, cost-informed decisions regarding the selection of stream groups, the optimal length of individual segments, and the exact placement of cryptographic states within the processing hardware. This holistic view ensures that the overhead introduced by managing the schedule itself does not outweigh the performance benefits gained from parallelization.<\/p>\n<p>To validate their theoretical model, the researchers conducted extensive empirical evaluations using two prominent post-quantum schemes: ML-KEM and HAETAE. The tests were executed across a diverse array of modern computing architectures to demonstrate hardware-agnostic versatility. The evaluation platform included an Apple M5 processor, an ARM-based Cortex-A72 core, and an Intel Core i5-11500 desktop processor. The experiments utilized specialized Keccak kernels configured to process two, four, or eight concurrent data streams per single instruction call, mapping directly to varying SIMD register widths available on the test hardware.<\/p>\n<p>The performance improvements demonstrated by the cost-aware scheduling approach are substantial. In evaluations involving ML-KEM, the optimization yielded speedups of up to 1.31 times for encapsulation operations and up to 1.44 times for decapsulation operations. Similarly, testing on the signature scheme HAETAE revealed significant efficiency gains, achieving speedups of up to 1.28 times for signing processes and up to 1.26 times for verification routines. These speedups are achieved entirely through software-level scheduling intelligence, requiring no specialized hardware modifications.<\/p>\n<p>The implications of this research extend directly to the practical deployment of post-quantum cryptography. As organizations worldwide prepare to transition their digital infrastructure to withstand potential threats from quantum computing, the computational overhead introduced by lattice-based schemes remains a central engineering challenge. Software implementations that can squeeze higher performance out of existing hardware reduce CPU bottlenecks, lower latency, and decrease energy consumption\u2014factors of critical importance for everything from resource-constrained edge devices to high-throughput cloud servers. By addressing the subtle execution inefficiencies inherent in parallel hash generation, the work by Zhang, Shi, Liu, and Lu provides a valuable tool for software developers working to make post-quantum security standards practical and efficient at scale.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Lattice-based cryptography has rapidly emerged as a foundational pillar for next-generation security, particularly as the&#8230;<\/p>\n","protected":false},"author":20,"featured_media":3004,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[676],"tags":[678,1653,1762,607,685,677,2540,214,2541,463,2539],"class_list":["post-3005","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-cryptography-math","tag-algorithms","tag-aware","tag-cost","tag-cryptography","tag-efficient","tag-math","tag-permutation","tag-research","tag-scheduling","tag-shake","tag-simd"],"_links":{"self":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/posts\/3005","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/users\/20"}],"replies":[{"embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=3005"}],"version-history":[{"count":0,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/posts\/3005\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=\/wp\/v2\/media\/3004"}],"wp:attachment":[{"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3005"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3005"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/smartcryptosecurity.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3005"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}