{"65678":{"#nid":"65678","#data":{"type":"event","title":"Thesis Proposal: Jaswanth Sreeram","body":[{"value":"\u003Cp\u003E\u003Cstrong\u003EOptimistic\nSemantic Synchronization\u003C\/strong\u003E\u003C\/p\u003E\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\u003Cp\u003E\u003Cstrong\u003EJaswanth Sreeram\u003C\/strong\u003E\u003Cbr \/\u003ESchool of Computer Science\u003Cbr \/\u003ECollege of Computing\u003Cbr \/\u003EGeorgia Institute of Technology\u003C\/p\u003E\n\n\u003Cp\u003E\u003Cstrong\u003ECommittee:\u003C\/strong\u003E\u003C\/p\u003E\n\n\u003Cp\u003ESantosh Pande (Advisor, College of Computing, Georgia\nTech)\u003Cbr \/\u003EHyesoon Kim (College of Computing, Georgia Tech)\u003Cbr \/\u003EKarsten Schwan (College\nof Computing, Georgia Tech)\u003Cbr \/\u003ESudhakar Yalamanchili (School of Electrical and\nComputer Engineering, Georgia Tech)\u003C\/p\u003E\n\n\n\n\u003Cp\u003E\u003Cstrong\u003ESummary\u003C\/strong\u003E\u003C\/p\u003E\n\n\n\n\u003Cp\u003EWithin the last decade multi-core processors have become\nincreasingly commonplace with the power and performance demands of modern\nreal-world programs acting to accelerate this trend. The rapid advancements in\ndesigning and adoption of such architectures mean that there is a serious need\nfor programming models that allow the development of correct parallel programs\nthat execute efficiently on these processors. A principle problem in this\nregard is that of efficiently synchronizing concurrent accesses to shared\nmemory. Traditional solutions to this problem are either inefficient but\nprovide programmability (coarse-grained locks) or are efficient but are not\ncomposable and very hard to program and verify (fine-grained locks).\nTransactional Memory Systems modeled on database transactions are being\nproposed as a solution for achieving thread synchronization in parallel\napplications. While optimistic Transactional Memory systems provide many of the\ncomposability and programmability advantages of coarse-grained locks and good theoretical\nscaling, several studies have found that their performance in practice for many\nprograms remains quite poor. Moreover since they are modeled on database\ntransactions, current transactional memory models remain rigid - they are not\nsuited for expressing some of the complex thread interactions that are\nprevalent in modern parallel programs. Moreover, the synchronization achieved\nby these transactional memory systems is at the physical or memory level.\u003C\/p\u003E\n\n\u003Cp\u003EThis thesis proposal advocates a position that the memory\nsynchronization problem for threads should be modeled and solved in terms of\nsynchronization of underlying program values which have semantics associated\nwith them and it presents optimistic synchronization techniques that address\nthese semantic synchronization requirements.\u003C\/p\u003E\u003Cp\u003EThese techniques range from methods to enable optimistic\ntransactions to recover from expensive sharing conflicts without discarding all\nthe work made possible by the optimism to mechanisms for enabling finer grained\nconsistency rules (than allowed by traditional optimistic TM models) therefore\navoiding conflicts that do not enforce any semantic property required by the\nprogram. In addition to improving the expressibility of specific\nsynchronization idioms these techniques are also effective in improving\nparallel performance. This thesis discusses these techniques in terms of their\npurpose and the extensions to the language, the compiler as well as to the\nconcurrency control runtime necessary to implement them. It also presents an\nexperimental evaluation of each of them on a variety of modern parallel\nworkloads. These experiments show that these techniques significantly improve\nparallel performance and scalability over programs using state-of-the-art\noptimistic synchronization methods.\u003C\/p\u003E","summary":null,"format":"limited_html"}],"field_subtitle":"","field_summary":"","field_summary_sentence":[{"value":"Optimistic Semantic Synchronization"}],"uid":"27466","created_gmt":"2011-04-20 12:18:49","changed_gmt":"2016-10-08 01:54:50","author":"Dani Denton","boilerplate_text":"","field_publication":"","field_article_url":"","field_event_time":{"event_time_start":"2011-04-26T14:00:00-04:00","event_time_end":"2011-04-26T16:00:00-04:00","event_time_end_last":"2011-04-26T16:00:00-04:00","gmt_time_start":"2011-04-26 18:00:00","gmt_time_end":"2011-04-26 20:00:00","gmt_time_end_last":"2011-04-26 20:00:00","rrule":null,"timezone":"America\/New_York"},"extras":[],"groups":[{"id":"47223","name":"College of Computing"},{"id":"50875","name":"School of Computer Science"}],"categories":[],"keywords":[{"id":"12900","name":"Ph.D. Thesis Proposal Announcement"}],"core_research_areas":[],"news_room_topics":[],"event_categories":[],"invited_audience":[],"affiliations":[],"classification":[],"areas_of_expertise":[],"news_and_recent_appearances":[],"phone":[],"contact":[],"email":[],"slides":[],"orientation":[],"userdata":""}}}