Optimizing Synchronization in Multiprocessor Implementations of Iterative Dataflow Programs
نویسندگان
چکیده
The execution time or estimated execution time of actor. UBS Unbounded buffer synchronization. A synchronization protocol that must be used for feedforward edges of the synchronization graph. This protocol requires four synchronization accesses per iteration period. t v () v 68 Glossary Same as with the DFG understood from context. If there is no path in from to , then ; otherwise, , where is any minimum-delay path from to. Given a path , is the sum of the edge delays over all edges in. Represents an edge whose source and sink vertices are and , respectively, and whose delay is equal to. Represents the maximum cycle mean of a DFG. BBS Bounded buffer synchronization. A synchronization protocol that may be used for feedback edges in a synchronization graph. This protocol requires two synchronization accesses per schedule period. critical cycle A fundamental cycle in a DFG whose cycle mean is equal to the maximum cycle mean of the DFG. cycle mean The cycle mean of a cycle in a DFG is equal to , where is the sum of the execution times of all vertices on , and is the sum of delays of all edges in. estimated throughput Given a DFG with execution time estimates for the actors, the estimated throughput is the reciprocal of the maximum cycle mean. feedback edge An edge that is contained in at least one cycle. feedforward edge An edge that is not contained in a cycle. maximum cycle mean Given a DFG, the maximum cycle mean is the largest cycle mean over all fundamental cycles in the DFG. SCC Strongly connected component. self-timed buffer bound Given a feedback edge in a synchronization graph, the self-timed buffer bound is an upper bound on the number of tokens that can simultaneously reside on (the buffer size). synchronization access An access to shared memory that used to update or examine the status of a synchronization variable. synchronization cost The average number of synchronization accesses that must be performed per iteration period in the self timed implementation of a ρ x y , () ρ G G ρ G x y , () G x y ρ G x y , () ∞ = ρ G x y , () p () Delay = p x y p () Delay p p () Delay p d n u v , () u v n λ max C …
منابع مشابه
Optimizing synchronization in multiprocessor DSP systems
This paper is concerned with multiprocessor implementations of embedded applications specified as iterative dataflow programs, in which synchronization overhead can be significant. We develop techniques to alleviate this overhead by determining a minimal set of processor synchronizations that are essential for correct execution. Our study is based in the context of self-timed execution of itera...
متن کاملResynchronization of Multiprocessor Schedules: Part 1 — Fundamental Concepts and Unbounded-latency Analysis
1. Abstract This paper introduces a technique, called resynchronization, for reducing synchronization overhead in embedded multiprocessor implementations. The technique exploits the well-known observation [35] that in a given multiprocessor implementation, certain synchronization operations may be redundant in the sense that their associated sequencing requirements are ensured by other synchron...
متن کاملResynchronization for Multiprocessor DSP Systems
This paper introduces a technique, called resynchronization, for reducing synchronization overhead in multiprocessor implementations of digital signal processing (DSP) systems. The technique applies to arbitrary collections of dedicated, programmable or configurable processors, such as combinations of programmable DSP’s, ASICS, and FPGA subsystems. Thus, it is particularly well-suited to the ev...
متن کاملIntermediate Representations for Design Automation of Multiprocessor DSP Systems
Self-timed scheduling is an attractive implementation style for multiprocessor DSP systems due to its ability to exploit predictability in application behavior, its avoidance of over-constrained synchronization, and its simplified clocking requirements. However, analysis and optimization of selftimed systems under real-time constraints is challenging due to the complex, irregular dynamics of se...
متن کاملResynchronization for Multiprocessor Dsp Implementation — Part 1: Maximum Throughput Resynchronization1
This paper introduces a technique, called resynchronization, for reducing synchronization overhead in multiprocessor implementations of digital signal processing (DSP) systems. The technique applies to arbitrary collections of dedicated, programmable or configurable processors, such as combinations of programmable DSPs, ASICS, and FPGA subsystems. Thus, it is particularly well suited to the evo...
متن کاملResynchronization of Multiprocessor Schedules: Part 2 — Latency-constrained Resynchronization
The companion paper [7] introduced the concept of resynchronization, a post-optimization for static multiprocessor schedules in which extraneous synchronization operations are introduced in such a way that the number of original synchronizations that consequently become redundant significantly exceeds the number of additional synchronizations. Redundant synchronizations are synchronization oper...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 1995