We introduce Gandiva, a new cluster scheduling framework that utilizes domain-specific knowledge to improve latency and efficiency of training deep learning models in a GPU cluster. One key characteristic of deep learning is feedback-driven exploration, where a user often runs a set of jobs (or a multi-job) to achieve the best result for a specific mission and uses early feedback on accuracy to dynamically prioritize or kill a subset of jobs; simultaneous early feedback on the entire multi-job is critical. A second characteristic is the heterogeneity of deep learning jobs in terms of resource usage, making it hard to achieve best-fit a priori. Gandiva addresses these two challenges by exploiting a third key characteristic of deep learning: intra-job predictability, as they perform numerous repetitive iterations called mini-batch iterations. Gandiva exploits intra-job predictability to time-slice GPUs efficiently across multiple jobs, thereby delivering low-latency. This predictability is also used for introspecting job performance and dynamically migrating jobs to better-fit GPUs, thereby improving cluster efficiency. We show via a prototype implementation and micro-benchmarks that Gandiva can speed up hyper-parameter searches during deep learning by up to an order of magnitude, and achieves better utilization by transparently migrating and time-slicing jobs to achieve better job-to-resource fit. We also show that, in a real workload of jobs running in a 180-GPU cluster, Gandiva improves aggregate cluster utilization by 26%, pointing to a new way of managing large GPU clusters for deep learning.<\/p>\n","protected":false},"excerpt":{"rendered":"
We introduce Gandiva, a new cluster scheduling framework that utilizes domain-specific knowledge to improve latency and efficiency of training deep learning models in a GPU cluster. One key characteristic of deep learning is feedback-driven exploration, where a user often runs a set of jobs (or a multi-job) to achieve the best result for a specific […]<\/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":""},"msr-content-type":[3],"msr-research-highlight":[],"research-area":[13547],"msr-publication-type":[193716],"msr-product-type":[],"msr-focus-area":[],"msr-platform":[],"msr-download-source":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-559971","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-systems-and-networking","msr-locale-en_us"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2018-10-1","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"USENIX","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"url","viewUrl":"false","id":"false","title":"https:\/\/www.usenix.org\/system\/files\/osdi18-xiao.pdf","label_id":"243109","label":0}],"msr_related_uploader":"","msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Wencong Xiao","user_id":0,"rest_url":false},{"type":"text","value":"Romil Bhardwaj","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Ramachandran Ramjee","user_id":33337,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Ramachandran Ramjee"},{"type":"user_nicename","value":"Muthian Sivathanu","user_id":36320,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Muthian Sivathanu"},{"type":"user_nicename","value":"Nipun Kwatra","user_id":37634,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Nipun Kwatra"},{"type":"user_nicename","value":"Zhenhua Han","user_id":39745,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Zhenhua Han"},{"type":"text","value":"Pratyush Patel","user_id":0,"rest_url":false},{"type":"text","value":"Xuan Peng","user_id":0,"rest_url":false},{"type":"text","value":"Hanyu Zhao","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Quanlu Zhang","user_id":36996,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Quanlu Zhang"},{"type":"user_nicename","value":"Fan Yang","user_id":31782,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Fan Yang"},{"type":"user_nicename","value":"Lidong Zhou","user_id":32673,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Lidong Zhou"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[920469],"msr_project":[600474,557136,968667],"publication":[],"video":[],"download":[],"msr_publication_type":"inproceedings","related_content":{"projects":[{"ID":600474,"post_title":"Micro Co-design","post_name":"micro-codesign","post_type":"msr-project","post_date":"2020-02-23 21:23:12","post_modified":"2020-02-23 21:23:13","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/micro-codesign\/","post_excerpt":"The massive scale of cloud infrastructure services enables, and often necessitates, vertical co-design of the infrastructure stack by cloud providers. Micro\u00a0co-design is a\u00a0Minimally invasive,\u00a0Cheap, and retro-fittable approach to co-design that extracts efficiency out of existing software infrastructure layers. It does this by making lightweight changes to generic software interfaces. We are exploring micro co-design in the context of four systems: Instalytics, Astra, Gandiva, and Quiver, aimed at improving the efficiency and functionality of key infrastructure…","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/600474"}]}},{"ID":557136,"post_title":"Gandiva: Scheduler for DNNs","post_name":"gandiva-scheduler-for-dnns","post_type":"msr-project","post_date":"2018-12-14 03:29:20","post_modified":"2019-11-19 09:29:06","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/gandiva-scheduler-for-dnns\/","post_excerpt":"Gandiva is a cluster scheduling framework that utilizes domain-specific knowledge of deep learning to improve the efficiency of training deep learning models in a GPU cluster.","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/557136"}]}},{"ID":968667,"post_title":"AI Infrastructure","post_name":"ai-infrastructure","post_type":"msr-project","post_date":"2023-10-04 00:44:27","post_modified":"2024-12-12 14:01:34","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/ai-infrastructure\/","post_excerpt":"Towards efficient AI\/ML deployment The AI Infrastructure team at Microsoft Research India works on cutting-edge systems optimizations for improving the efficiency of a variety of AI\/ML workloads, including an emerging class of workloads, namely, serving large language models (LLMs). AI\/ML models are expensive to train and serve at scale and therefore, systems optimizations are crucial for unlocking the true potential of AI-powered applications. The key principle behind many of our projects is co-design of Systems and…","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/968667"}]}}]},"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/559971","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/559971\/revisions"}],"predecessor-version":[{"id":559977,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/559971\/revisions\/559977"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=559971"}],"wp:term":[{"taxonomy":"msr-content-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-content-type?post=559971"},{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=559971"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=559971"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=559971"},{"taxonomy":"msr-product-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-product-type?post=559971"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=559971"},{"taxonomy":"msr-platform","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-platform?post=559971"},{"taxonomy":"msr-download-source","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-download-source?post=559971"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=559971"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=559971"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=559971"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=559971"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=559971"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=559971"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=559971"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}