Ph.d. Dissertation Evaluation and Exploitation of Locality in the Data Driven Execution Model

OF PH.D. DISSERTATION EVALUATION AND EXPLOITATION OF LOCALITY IN THE DATA DRIVEN EXECUTION MODEL The advent of hybrid von Neumann-data driven architectures arose from a desire to combine the most salient features of coarse grain von Neumann and ne-grain data driven models. Hybrid architectures achieve high performance through concurrent execution and the exploitation of program and data locality in conjunction with a reduction in the execution overhead associated with instruction-level synchronization inherent in classical data ow processors. This is accomplished by increasing the task granularity from one to multiple instructions and executing the resulting tasks as traditional von Neumann instruction threads. The addition of coarse grain features has brought signi cant improvements to the basic data driven model. Each of these coarse grain features addresses one aspect of the tradeo between locality and ne grain parallelism. The characteristics and nature of these are relatively well understood in a qualitative way individually. However, very little research has been done to quantify their combined e ect on performance. The research presented in this dissertation involves the quanti cation of the available forms of locality in data driven execution and the resulting impact of its exploitation on a hybrid von Neumann-data driven architecture's ability to mask long communication and memory access latencies. A model for data driven vector execution is formulated to allow for the exploitation of data structure locality. The performance of this model is compared to that of a massively parallel data driven processor and a conventional tightly coupled MIMD vector supercomputer. William Marcus Miller Department of Computer Science Colorado State University Fort Collins, Colorado 80523 Fall 1993