[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84468-en":3,"doc-seo-84468-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},84468,687197100911,"Himbo","https://ap-avatar.wpscdn.com/avatar/a000239b6f1da00475?x-image-process=image/resize,m_fixed,w_180,h_180&k=1782698725881665579",6,"Technology","AlayaLaser Efficient Index Layout and Search Strategy for Large-scale High-dimensional Vector Similarity Search","On-disk graph-based approximate nearest neighbor search (ANNS) is crucial for large-scale, high-dimensional vector retrieval, yet conventional analysis emphasizes I/O costs as the dominant limitation. AlayaLaser challenges this view by observing that, as vector dimensionality increases to hundreds or thousands, on-disk graph-based index performance becomes compute-bound rather than I/O-bound. Based on roofline-model analysis, the work introduces an on-disk data layout leveraging SIMD on modern CPUs, plus cache, entry selection, and early dispatch optimizations. Extensive experiments validate superior results, matching or exceeding in-memory indices.","AlayaLaser: Efficient Index Layout and Search Strategy for Large-scale High-dimensional Vector Similarity Search  \nWeijian Chen  \nSUSTech  \n[chenwj2024@mail.sustech.edu.cn](chenwj2024@mail.sustech.edu.cn)  \nLong Xiang  \nAlayaDB AI [long.xiang@alayadb.ai](long.xiang@alayadb.ai)  \nHaotian Liu  \nAlayaDB AI [haotian.liu@alayadb.ai](haotian.liu@alayadb.ai)  \nLiang Huang  \nSUSTech  \n[huangl2025@mail.sustech.edu.cn](huangl2025@mail.sustech.edu.cn)  \nYangshen Deng University of Edinburgh [yangshen.deng@ed.ac.uk](yangshen.deng@ed.ac.uk)  \nBo Tang∗  \nSUSTech [tangb3@sustech.edu.cn](tangb3@sustech.edu.cn)  \narXiv :2602 .23342v4 [ cs .DB] 11 Jul 2026  \nAbstract  \nOn-disk graph-based approximate nearest neighbor search (ANNS) is essential for large-scale, high-dimensional vector retrieval, yet its performance is widely recognized to be limited by the prohibitive I/O costs. Interestingly, we observed that the performance of ondisk graph-based index systems is compute-bound, not I/O-bound, with the rising of the vector data dimensionality (e.g., hundreds or thousands). This insight uncovers a significant optimization opportunity: existing on-disk graph-based index systems universally target I/O reduction and largely overlook computational overhead, which leaves a substantial performance improvement space.  \nIn this work, we propose AlayaLaser, an efficient on-disk graphbased index system for large-scale high-dimensional vector similarity search. In particular, we first conduct performance analysis on existing on-disk graph-based index systems via the adapted roofline model, then we devise a novel on-disk data layout in AlayaLaser to effectively alleviate the compute-bound, which is revealed by the above roofline model analysis, by exploiting SIMD instructions on modern CPUs. We next design a suite of optimization techniques (e.g., degree-based node cache, cluster-based entry point selection, and early dispatch strategy) to further improve the performance of AlayaLaser. We last conduct extensive experimental studies on a wide range of large-scale high-dimensional vector datasets to verify the superiority of AlayaLaser. Specifically, AlayaLaser not only surpasses existing on-disk graph-based index systems but also matches or even exceeds the performance of in-memory index systems.  \n1 Introduction  \nLarge-scale, high-dimensional vector similarity search is a core subroutine in many applications, including recommendation systems [16, 29, 39, 52], computer vision [7, 23, 42], and RetrievalAugmented Generation (RAG) [17, 35, 45, 55] in Large Language Models. Due to the curse of dimensionality, exact search becomes prohibitively expensive in high-dimensional spaces. Consequently,  \n∗ Dr. Bo Tang is the corresponding author.  \nPermission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than the author(s) must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission [and/or a fee. Request permissions from permissions@acm.org](and/or a fee. Request permissions from permissions@acm.org).  \nConference’17, Washington, DC, USA  \n© 2026 Copyright held by the owner/author(s) . Publication rights licensed to ACM. ACM ISBN 978-x-xxxx-xxxx-x/YYYY/MM [https://doi.org/10.1145/nnnnnnn.nnnnnnn](https://doi.org/10.1145/nnnnnnn.nnnnnnn)  \nCPU  \nI/O  \n3.76  \n(a) DiskANN [32]  \nTime (ms)  \nCPU  \nI/O  \n(b) Starling [49]  \nCPU  \nI/O  \n 2.07  Time (ms)  \n(c) PipeANN [26]  \nCPU  \nI/O  \n0.84  Time (ms)  \n(d) AlayaLaser (this work)  \nFigure 1: Query latency profiling of on-disk graph-based index systems, Recall@10=0.90 on 960-dimensional GIST1M  \napproximate nearest neighbor search (ANNS) has em","cbCaikmz6UIMs7u0","https://ap.wps.com/l/cbCaikmz6UIMs7u0","pdf",1634679,1,15,"English","en",105,"# Introduction\n## Vector similarity search and ANNS evaluation metrics\n## Motivation: memory limits vs on-disk indexes\n# AlayaLaser overview\n## Compute-bound insight and roofline analysis\n## Proposed on-disk index layout for SIMD acceleration\n# Optimization techniques\n## Degree-based node cache\n## Cluster-based entry point selection\n## Early dispatch strategy\n# Experimental evaluation\n## Datasets and large-scale high-dimensional benchmarks\n## Comparison against prior on-disk and in-memory systems","[{\"question\":\"Why does AlayaLaser focus on compute rather than I/O costs in on-disk graph-based ANNS?\",\"answer\":\"It observes that as vector dimensionality rises to hundreds or thousands, the performance of on-disk graph-based index systems becomes compute-bound instead of I/O-bound.\"},{\"question\":\"What core method does AlayaLaser use to guide its design?\",\"answer\":\"AlayaLaser conducts performance analysis on existing on-disk graph-based indexes using an adapted roofline model, then uses the result to address the identified compute bottleneck.\"},{\"question\":\"Which optimization techniques are proposed beyond the on-disk data layout?\",\"answer\":\"The system includes degree-based node cache, cluster-based entry point selection, and an early dispatch strategy to further improve query performance.\"}]",1784195826,38,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"alayalaser-efficient-index-layout-and-search-strategy-for-large-scale-high-dimensional-vector-similarity-search","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/technology/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/alayalaser-efficient-index-layout-and-search-strategy-for-large-scale-high-dimensional-vector-similarity-search/84468/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why does AlayaLaser focus on compute rather than I/O costs in on-disk graph-based ANNS?","Question",{"text":75,"@type":76},"It observes that as vector dimensionality rises to hundreds or thousands, the performance of on-disk graph-based index systems becomes compute-bound instead of I/O-bound.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What core method does AlayaLaser use to guide its design?",{"text":80,"@type":76},"AlayaLaser conducts performance analysis on existing on-disk graph-based indexes using an adapted roofline model, then uses the result to address the identified compute bottleneck.",{"name":82,"@type":73,"acceptedAnswer":83},"Which optimization techniques are proposed beyond the on-disk data layout?",{"text":84,"@type":76},"The system includes degree-based node cache, cluster-based entry point selection, and an early dispatch strategy to further improve query performance.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,113,118,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":111,"slug":112},50,"technology",{"id":114,"doc_module":4,"doc_module_name":45,"category_name":115,"show_sort_weight":116,"slug":117},7,"Healthcare",40,"healthcare",{"id":119,"doc_module":4,"doc_module_name":45,"category_name":120,"show_sort_weight":121,"slug":122},8,"Research & Report",30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]