Abstra t The Distributed ASCI Super omputer (DAS) is a homogeneous wide-area distributed system o... more Abstra t The Distributed ASCI Super omputer (DAS) is a homogeneous wide-area distributed system onsisting of four luster omputers at di erent lo ations. DAS has been used for resear h on ommuni ation software, parallel languages and programming systems, s hedulers, parallel appli ations, and distributed appli ations. The paper gives a preview of the most interesting resear h results obtained so far in the DAS proje t. 1 1 More information about the DAS proje t an be found on http://www. s.vu.nl/das/
Recent superscalar processors highly depend on ecient branch prediction to exploit instruction le... more Recent superscalar processors highly depend on ecient branch prediction to exploit instruction level parallelism. Many strategies to improve branch prediction accuracy have been proposed, the most successful ones can adapt predictions dynamically based on the outcomes of previous branches. In this paper we present a dierent strategy, that is partially non{adaptive. We will show that patterns in the outcomes of individual branch instructions are more important than the global pattern of consecutive branch instructions and we will show that some patterns in the outcomes of branch instructions are much more frequent than other patterns. We will propose a classication of patterns in outcomes of individual branch instructions, based on their frequentness and how well they can be predicted by Two Level Adaptive Branch Predictors. We will analyze six traces of benchmark applications with respect to the occurrence of these patterns. We will present a mechanism (HW, Hard Wired) that dynamically discriminates between instructions based on this categorization. This mechanism can predict trivial and loop patterns almost perfectly with a hard wired look{up table. We will discuss the usage of HW as a prediction strategy in its own and its usage as a pre{stage to other branch prediction strategies. \Predicting is hard, especially the future." Winston Churchill Chapter 1
The Distributed ASCI Supercomputer (DAS) is a homogeneous wide-area distributed system consisting... more The Distributed ASCI Supercomputer (DAS) is a homogeneous wide-area distributed system consisting of four cluster computers at different locations. DAS has been used for research on communication software, parallel languages and programming systems, schedulers, parallel applications, and distributed applications. The paper gives a preview of the most interesting research results obtained so far in the DAS project.
Abstra t The Distributed ASCI Super omputer (DAS) is a homogeneous wide-area distributed system o... more Abstra t The Distributed ASCI Super omputer (DAS) is a homogeneous wide-area distributed system onsisting of four luster omputers at di erent lo ations. DAS has been used for resear h on ommuni ation software, parallel languages and programming systems, s hedulers, parallel appli ations, and distributed appli ations. The paper gives a preview of the most interesting resear h results obtained so far in the DAS proje t. 1 1 More information about the DAS proje t an be found on http://www. s.vu.nl/das/
Recent superscalar processors highly depend on ecient branch prediction to exploit instruction le... more Recent superscalar processors highly depend on ecient branch prediction to exploit instruction level parallelism. Many strategies to improve branch prediction accuracy have been proposed, the most successful ones can adapt predictions dynamically based on the outcomes of previous branches. In this paper we present a dierent strategy, that is partially non{adaptive. We will show that patterns in the outcomes of individual branch instructions are more important than the global pattern of consecutive branch instructions and we will show that some patterns in the outcomes of branch instructions are much more frequent than other patterns. We will propose a classication of patterns in outcomes of individual branch instructions, based on their frequentness and how well they can be predicted by Two Level Adaptive Branch Predictors. We will analyze six traces of benchmark applications with respect to the occurrence of these patterns. We will present a mechanism (HW, Hard Wired) that dynamically discriminates between instructions based on this categorization. This mechanism can predict trivial and loop patterns almost perfectly with a hard wired look{up table. We will discuss the usage of HW as a prediction strategy in its own and its usage as a pre{stage to other branch prediction strategies. \Predicting is hard, especially the future." Winston Churchill Chapter 1
The Distributed ASCI Supercomputer (DAS) is a homogeneous wide-area distributed system consisting... more The Distributed ASCI Supercomputer (DAS) is a homogeneous wide-area distributed system consisting of four cluster computers at different locations. DAS has been used for research on communication software, parallel languages and programming systems, schedulers, parallel applications, and distributed applications. The paper gives a preview of the most interesting research results obtained so far in the DAS project.
Uploads
Papers by Floris Sluiter