{"id":967329,"date":"2023-10-16T02:14:29","date_gmt":"2023-10-16T09:14:29","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-project&#038;p=967329"},"modified":"2024-01-12T08:47:20","modified_gmt":"2024-01-12T16:47:20","slug":"domain-specialization","status":"publish","type":"msr-project","link":"https:\/\/www.microsoft.com\/en-us\/research\/project\/domain-specialization\/","title":{"rendered":"Domain Specialization"},"content":{"rendered":"<section class=\"mb-3 moray-highlight\">\n\t<div class=\"card-img-overlay mx-lg-0\">\n\t\t<div class=\"card-background  has-background-catalina-blue card-background--full-bleed\">\n\t\t\t\t\t<\/div>\n\t\t<!-- Foreground -->\n\t\t<div class=\"card-foreground d-flex mt-md-n5 my-lg-5 px-g px-lg-0\">\n\t\t\t<!-- Container -->\n\t\t\t<div class=\"container d-flex mt-md-n5 my-lg-5 \">\n\t\t\t\t<!-- Card wrapper -->\n\t\t\t\t<div class=\"w-100 w-lg-col-5\">\n\t\t\t\t\t<!-- Card -->\n\t\t\t\t\t<div class=\"card material-md-card py-5 px-md-5\">\n\t\t\t\t\t\t<div class=\"card-body \">\n\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\n\n<h1 class=\"wp-block-heading\" id=\"domain-specialization\">Domain Specialization<\/h1>\n\n\n\n<p>Scaling performance beyond Moore&#8217;s law<\/p>\n\n\t\t\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t<\/div>\n\t<\/div>\n<\/section>\n\n\n\n\n\n<p>Domain&nbsp;specialization&nbsp;is&nbsp;expected&nbsp;to&nbsp;play&nbsp;a&nbsp;big&nbsp;role&nbsp;in&nbsp;how&nbsp;computer&nbsp;systems&nbsp;evolve&nbsp;in&nbsp;future. With&nbsp;the&nbsp;end&nbsp;of&nbsp;Moore&#8217;s&nbsp;law,&nbsp;we&nbsp;are&nbsp;already&nbsp;seeing&nbsp;CPU,&nbsp;GPU&nbsp;and domain specific&nbsp;hardware&nbsp;evolving&nbsp;rapidly. The next decade&nbsp;is&nbsp;therefore&nbsp;expected&nbsp;to&nbsp;see&nbsp;big&nbsp;changes&nbsp;in&nbsp;how&nbsp;we&nbsp;develop,&nbsp;compile&nbsp;and&nbsp;run&nbsp;software. This project focuses on data systems, a class of systems where, as the data sizes grow, performance scaling is going to be of importance.<br><br>First,&nbsp;we&nbsp;believe&nbsp;that domain-specific&nbsp;compilers&nbsp;will&nbsp;play&nbsp;a&nbsp;crucial&nbsp;strategic&nbsp;role&nbsp;in&nbsp;helping&nbsp;software&nbsp;leverage&nbsp;the&nbsp;changing&nbsp;hardware&nbsp;landscape. Such compilers will be multi-layered and will progressively lower computation through multiple intermediate abstractions, performing domain specific optimizations at the higher layers and specializing code to the hardware in lower layers. We&nbsp;have&nbsp;been&nbsp;working&nbsp;on&nbsp;two&nbsp;such&nbsp;domain&nbsp;specific&nbsp;compilers&nbsp;in&nbsp;the&nbsp;data&nbsp;domain. Second, new hardware specific algorithms need to be designed to harness the hardware&#8217;s capabilities. To this end, we are designing new algorithms that can effectively harness the raw capabilities of a GPU irrespective of its architecture.<\/p>\n\n\n\n<p><\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-large is-resized is-style-spectrum\"><img loading=\"lazy\" decoding=\"async\" width=\"1024\" height=\"589\" src=\"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2023\/10\/image-652cf670dbed7-1024x589.png\" alt=\"diagram\" class=\"wp-image-976389\" style=\"width:700px;height:403px\" srcset=\"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2023\/10\/image-652cf670dbed7-1024x589.png 1024w, https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2023\/10\/image-652cf670dbed7-300x173.png 300w, https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2023\/10\/image-652cf670dbed7-768x442.png 768w, https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2023\/10\/image-652cf670dbed7-240x138.png 240w, https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2023\/10\/image-652cf670dbed7.png 1128w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption class=\"wp-element-caption\"><strong>The landscape of domain specialization.<\/strong><\/figcaption><\/figure>\n\n\n","protected":false},"excerpt":{"rendered":"<p>Scaling performance beyond Moore&#8217;s law Domain&nbsp;specialization&nbsp;is&nbsp;expected&nbsp;to&nbsp;play&nbsp;a&nbsp;big&nbsp;role&nbsp;in&nbsp;how&nbsp;computer&nbsp;systems&nbsp;evolve&nbsp;in&nbsp;future. With&nbsp;the&nbsp;end&nbsp;of&nbsp;Moore&#8217;s&nbsp;law,&nbsp;we&nbsp;are&nbsp;already&nbsp;seeing&nbsp;CPU,&nbsp;GPU&nbsp;and domain specific&nbsp;hardware&nbsp;evolving&nbsp;rapidly. The next decade&nbsp;is&nbsp;therefore&nbsp;expected&nbsp;to&nbsp;see&nbsp;big&nbsp;changes&nbsp;in&nbsp;how&nbsp;we&nbsp;develop,&nbsp;compile&nbsp;and&nbsp;run&nbsp;software. This project focuses on data systems, a class of systems where, as the data sizes grow, performance scaling is going to be of importance. First,&nbsp;we&nbsp;believe&nbsp;that domain-specific&nbsp;compilers&nbsp;will&nbsp;play&nbsp;a&nbsp;crucial&nbsp;strategic&nbsp;role&nbsp;in&nbsp;helping&nbsp;software&nbsp;leverage&nbsp;the&nbsp;changing&nbsp;hardware&nbsp;landscape. Such compilers will be multi-layered and will progressively lower computation through multiple intermediate abstractions, performing domain specific [&hellip;]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"research-area":[13547],"msr-locale":[268875],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-967329","msr-project","type-msr-project","status-publish","hentry","msr-research-area-systems-and-networking","msr-locale-en_us","msr-archive-status-active"],"msr_project_start":"","related-publications":[976416,980577,980586,1082997],"related-downloads":[],"related-videos":[],"related-groups":[144939],"related-events":[],"related-opportunities":[],"related-posts":[],"related-articles":[],"tab-content":[],"slides":[],"related-researchers":[{"type":"user_nicename","display_name":"Harish Doraiswamy","user_id":41051,"people_section":"Related people","alias":"hdoraiswamy"},{"type":"user_nicename","display_name":"Akash Lal","user_id":30905,"people_section":"Related people","alias":"akashl"},{"type":"user_nicename","display_name":"Sampath Rajendra","user_id":43107,"people_section":"Related people","alias":"srajendra"},{"type":"user_nicename","display_name":"Aseem Rastogi","user_id":36021,"people_section":"Related people","alias":"aseemr"}],"msr_research_lab":[199562],"msr_impact_theme":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/967329","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-project"}],"version-history":[{"count":15,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/967329\/revisions"}],"predecessor-version":[{"id":998817,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/967329\/revisions\/998817"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=967329"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=967329"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=967329"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=967329"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=967329"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}