Chih-Chieh Yang, J. C. Pichel, Adam R. Smith, D. Padua
{"title":"层次化平铺数组作为代码集的高级抽象","authors":"Chih-Chieh Yang, J. C. Pichel, Adam R. Smith, D. Padua","doi":"10.1109/DFM.2014.11","DOIUrl":null,"url":null,"abstract":"The move from terascale to exascale systems is challenging in terms of energy and power consumption, resilience, storage, concurrency, and parallelism. These challenges require new fine-grain execution models to support the concurrent execution of millions or even billions of threads on the exascale machines. The most promising approaches are those based on the codelet execution model, which provide a flexible programming interface that allows the expression of all kinds of parallelism with fine-tuning opportunities. We propose using Hierarchically Tiled Array (HTA) as a high-level abstraction for codelets to improve the programmability and readability of programs while preserving the good performance and scalability provided by the codelet execution model.","PeriodicalId":183526,"journal":{"name":"2014 Fourth Workshop on Data-Flow Execution Models for Extreme Scale Computing","volume":"22 1 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2014-08-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"7","resultStr":"{\"title\":\"Hierarchically Tiled Array as a High-Level Abstraction for Codelets\",\"authors\":\"Chih-Chieh Yang, J. C. Pichel, Adam R. Smith, D. Padua\",\"doi\":\"10.1109/DFM.2014.11\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"The move from terascale to exascale systems is challenging in terms of energy and power consumption, resilience, storage, concurrency, and parallelism. These challenges require new fine-grain execution models to support the concurrent execution of millions or even billions of threads on the exascale machines. The most promising approaches are those based on the codelet execution model, which provide a flexible programming interface that allows the expression of all kinds of parallelism with fine-tuning opportunities. We propose using Hierarchically Tiled Array (HTA) as a high-level abstraction for codelets to improve the programmability and readability of programs while preserving the good performance and scalability provided by the codelet execution model.\",\"PeriodicalId\":183526,\"journal\":{\"name\":\"2014 Fourth Workshop on Data-Flow Execution Models for Extreme Scale Computing\",\"volume\":\"22 1 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2014-08-24\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"7\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2014 Fourth Workshop on Data-Flow Execution Models for Extreme Scale Computing\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/DFM.2014.11\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2014 Fourth Workshop on Data-Flow Execution Models for Extreme Scale Computing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/DFM.2014.11","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Hierarchically Tiled Array as a High-Level Abstraction for Codelets
The move from terascale to exascale systems is challenging in terms of energy and power consumption, resilience, storage, concurrency, and parallelism. These challenges require new fine-grain execution models to support the concurrent execution of millions or even billions of threads on the exascale machines. The most promising approaches are those based on the codelet execution model, which provide a flexible programming interface that allows the expression of all kinds of parallelism with fine-tuning opportunities. We propose using Hierarchically Tiled Array (HTA) as a high-level abstraction for codelets to improve the programmability and readability of programs while preserving the good performance and scalability provided by the codelet execution model.