{"id":{"repo_id":"vt","oai_identifier":"oai:vtechworks.lib.vt.edu:10919/78130"},"canonical_url":"https://search.dev.ndltd.org/etd/vt/oai:vtechworks.lib.vt.edu:10919/78130","repository":{"repo_id":"vt","name":"Virginia Tech","base_url":"https://vtechworks.lib.vt.edu/oai/request"},"display":{"title":"Automatic Scheduling of Compute Kernels Across Heterogeneous Architectures","abstract":"The world of high-performance computing has shifted from increasing single-core performance to extracting performance from heterogeneous multi- and many-core processors due to the power, memory and instruction-level parallelism walls. All trends point towards increased processor heterogeneity as a means for increasing application performance, from smartphones to servers. These various architectures are designed for different types of applications — traditional \"big\" CPUs (like the Intel Xeon) are optimized for low latency while other architectures (such as the NVidia Tesla K20x) are optimized for high-throughput. These architectures have different tradeoffs and different performance profiles, meaning fantastic performance gains for the right types of applications. However applications that are ill-suited for a given architecture may experience significant slowdown; therefore, it is imperative that applications are scheduled onto the correct processor. In order to perform this scheduling, applications must be analyzed to determine their execution characteristics. Traditionally this application-to-hardware mapping was determined statically by the programmer. However, this requires intimate knowledge of the application and underlying architecture, and precludes load-balancing by the system. We demonstrate and empirically evaluate a system for automatically scheduling compute kernels by extracting program characteristics and applying machine learning techniques. We develop a machine learning process that is system-agnostic, and works for a variety of contexts (e.g. embedded, desktop/workstation, server). Finally, we perform scheduling in a workload-aware and workload-adaptive manner for these compute kernels.","abstract_html":"The world of high-performance computing has shifted from increasing single-core performance to extracting performance from heterogeneous multi- and many-core processors due to the power, memory and instruction-level parallelism walls. All trends point towards increased processor heterogeneity as a means for increasing application performance, from smartphones to servers. These various architectures are designed for different types of applications — traditional "big" CPUs (like the Intel Xeon) are optimized for low latency while other architectures (such as the NVidia Tesla K20x) are optimized for high-throughput. These architectures have different tradeoffs and different performance profiles, meaning fantastic performance gains for the right types of applications. However applications that are ill-suited for a given architecture may experience significant slowdown; therefore, it is imperative that applications are scheduled onto the correct processor. In order to perform this scheduling, applications must be analyzed to determine their execution characteristics. Traditionally this application-to-hardware mapping was determined statically by the programmer. However, this requires intimate knowledge of the application and underlying architecture, and precludes load-balancing by the system. We demonstrate and empirically evaluate a system for automatically scheduling compute kernels by extracting program characteristics and applying machine learning techniques. We develop a machine learning process that is system-agnostic, and works for a variety of contexts (e.g. embedded, desktop/workstation, server). Finally, we perform scheduling in a workload-aware and workload-adaptive manner for these compute kernels.","abstract_has_math":false,"creators":["Lyerly, Robert Frantz"],"institution":"Virginia Tech","degree_name":"Master of Science","degree_level":"masters","degree_discipline":"Electrical and Computer Engineering","degree_department":"Electrical and Computer Engineering","school":null,"contributors":[],"advisors":[],"committee_chairs":["Ravindran, Binoy"],"committee_members":["Plassmann, Paul","Patterson, Cameron D."],"year":2014,"date_issued":"2014-05-07","date_published":"2014-05-07","updated_at":"2026-07-22T22:18:49Z","subjects":["High-Performance Computing","Runtime Systems","Heterogeneous Architectures","Compilers","Scheduling"],"languages":["en_US"],"rights":["In Copyright"],"rights_urls":["http://rightsstatements.org/vocab/InC/1.0/"],"identifier_entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["etd-05202014-193503"],"render_values":[{"text":"etd-05202014-193503","href":null,"code":true}]}]},"links":{"outbound_url":"http://hdl.handle.net/10919/78130","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.committeechair","label":"Committee Chair","values":["Ravindran, Binoy"]},{"key":"dc:contributor.committeemember","label":"Committee Member","values":["Plassmann, Paul","Patterson, Cameron D."]},{"key":"dc:contributor.department","label":"Department","values":["Electrical and Computer Engineering"]},{"key":"dc:creator","label":"Author","values":["Lyerly, Robert Frantz"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2017-06-13T19:44:10Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2017-06-13T19:44:10Z","2014-06-24"]},{"key":"dc:date.issued","label":"Date","values":["2014-05-07"]},{"key":"dc:publisher","label":"Institution","values":["Virginia Tech"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"dc:type.dcmitype","label":"Dc Type Dcmitype","values":["Text"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Electrical and Computer Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["masters"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Master of Science"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["Virginia Polytechnic Institute and State University"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["High-Performance Computing","Runtime Systems","Heterogeneous Architectures","Compilers","Scheduling"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en_US"]},{"key":"dc:rights","label":"Dc Rights","values":["In Copyright"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://rightsstatements.org/vocab/InC/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.other","label":"Dc Identifier Other","values":["etd-05202014-193503"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["http://hdl.handle.net/10919/78130"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["The world of high-performance computing has shifted from increasing single-core performance to extracting performance from heterogeneous multi- and many-core processors due to the power, memory and instruction-level parallelism walls. All trends point towards increased processor heterogeneity as a means for increasing application performance, from smartphones to servers. These various architectures are designed for different types of applications — traditional \"big\" CPUs (like the Intel Xeon) are optimized for low latency while other architectures (such as the NVidia Tesla K20x) are optimized for high-throughput. These architectures have different tradeoffs and different performance profiles, meaning fantastic performance gains for the right types of applications. However applications that are ill-suited for a given architecture may experience significant slowdown; therefore, it is imperative that applications are scheduled onto the correct processor. In order to perform this scheduling, applications must be analyzed to determine their execution characteristics. Traditionally this application-to-hardware mapping was determined statically by the programmer. However, this requires intimate knowledge of the application and underlying architecture, and precludes load-balancing by the system. We demonstrate and empirically evaluate a system for automatically scheduling compute kernels by extracting program characteristics and applying machine learning techniques. We develop a machine learning process that is system-agnostic, and works for a variety of contexts (e.g. embedded, desktop/workstation, server). Finally, we perform scheduling in a workload-aware and workload-adaptive manner for these compute kernels."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Master of Science"]},{"key":"dc:title","label":"Title","values":["Automatic Scheduling of Compute Kernels Across Heterogeneous Architectures"]}]}],"canonical_facts":{"dc:contributor.committeechair":["Ravindran, Binoy"],"dc:contributor.committeemember":["Plassmann, Paul","Patterson, Cameron D."],"dc:contributor.department":["Electrical and Computer Engineering"],"dc:creator":["Lyerly, Robert Frantz"],"dc:date.accessioned":["2017-06-13T19:44:10Z"],"dc:date.available":["2017-06-13T19:44:10Z","2014-06-24"],"dc:date.issued":["2014-05-07"],"dc:description.abstract":["The world of high-performance computing has shifted from increasing single-core performance to extracting performance from heterogeneous multi- and many-core processors due to the power, memory and instruction-level parallelism walls. All trends point towards increased processor heterogeneity as a means for increasing application performance, from smartphones to servers. These various architectures are designed for different types of applications — traditional \"big\" CPUs (like the Intel Xeon) are optimized for low latency while other architectures (such as the NVidia Tesla K20x) are optimized for high-throughput. These architectures have different tradeoffs and different performance profiles, meaning fantastic performance gains for the right types of applications. However applications that are ill-suited for a given architecture may experience significant slowdown; therefore, it is imperative that applications are scheduled onto the correct processor. In order to perform this scheduling, applications must be analyzed to determine their execution characteristics. Traditionally this application-to-hardware mapping was determined statically by the programmer. However, this requires intimate knowledge of the application and underlying architecture, and precludes load-balancing by the system. We demonstrate and empirically evaluate a system for automatically scheduling compute kernels by extracting program characteristics and applying machine learning techniques. We develop a machine learning process that is system-agnostic, and works for a variety of contexts (e.g. embedded, desktop/workstation, server). Finally, we perform scheduling in a workload-aware and workload-adaptive manner for these compute kernels."],"dc:description.degree":["Master of Science"],"dc:identifier.other":["etd-05202014-193503"],"dc:identifier.uri":["http://hdl.handle.net/10919/78130"],"dc:language.iso":["en_US"],"dc:publisher":["Virginia Tech"],"dc:rights":["In Copyright"],"dc:rights.uri":["http://rightsstatements.org/vocab/InC/1.0/"],"dc:subject":["High-Performance Computing","Runtime Systems","Heterogeneous Architectures","Compilers","Scheduling"],"dc:title":["Automatic Scheduling of Compute Kernels Across Heterogeneous Architectures"],"dc:type":["Thesis"],"dc:type.dcmitype":["Text"],"thesis:degree_discipline":["Electrical and Computer Engineering"],"thesis:degree_level":["masters"],"thesis:degree_name":["Master of Science"],"thesis:institution_name":["Virginia Polytechnic Institute and State University"]},"updated_at":"2026-07-22T22:18:49Z"}