An empirical characterization of stream programs and its implications for language and compiler design

Abstract

Stream programs represent an important class of high-performance computations. Defined by their regular processing of sequences of data, stream programs appear most commonly in the context of audio, video, and digital signal processing, though also in networking, encryption, and other areas. In order to develop effective compilation techniques for the streaming domain, it is important to understand the common characteristics of these programs. Prior characterizations of stream programs have examined legacy implementations in C, C++, or FORTRAN, making it difficult to extract the high-level properties of the algorithms.In this work, we characterize a large set of stream programs that was implemented directly in a stream programming language, allowing new insights into the high-level structure and behavior of the applications. We utilize the StreamIt benchmark suite, consisting of 65 programs and 33,600 lines of code. We characterize the bottlenecks to parallelism, the data reference patterns, the input/output rates, and other properties. The lessons learned have implications for the design of future architectures, languages and compilers for the streaming domain.

Hard-real-time scheduling of data-dependent tasks in embedded streaming applications

“…Typically, embedded streaming applications receive their input samples from hardware sensors/devices which generate these samples periodically. By considering acyclic CSDF graphs, our investigation findings and proofs are applicable to most streaming applications since it has been shown recently that around 90% of streaming applications can be modeled as acyclic SDF graphs [28]. Note that SDF graphs are a subset of the CSDF graphs we consider in this paper.…”

“…The first source is the StreamIt benchmark [28] which contributes 11 streaming applications. The second source is the SDF 3 benchmark [27] which contributes 5 streaming applications.…”

“…Only three applications (CD2DAT-(S,C) and Satellite) have lower throughput under our SPS. To understand the impact of the results, we introduce the concept of matched I/O applications which according to [28] is the class of applications with a small value of L. The authors in [28] reported recently an interesting finding: Neighboring actors often have matched I/O rates. This reduces the opportunity and impact of advanced scheduling strategies proposed in the literature.…”

“…This reduces the opportunity and impact of advanced scheduling strategies proposed in the literature. According to [28], the advanced scheduling strategies proposed in the literature (e.g. [26]) are suitable for mis-matched I/O rates applications (i.e.…”

A Study of Dynamic Phase Adaptation Using a Dynamic Multicore Processor

“…However this work does not involve changing the core fusion during the execution of the benchmark. Also, the machine learning model focuses on using highlevel information from StreamIt's [22] language constructs.…”