Performance analysis challenges and framework for high-performance reconfigurable computing

Abstract

Reconfigurable computing (RC) applications employing both microprocessors and FPGAs have potential for large speedup when compared with traditional (software) parallel applications. However, this potential is marred by the additional complexity of these dual-paradigm systems, making it difficult to identify performance bottlenecks and achieve desired performance. Performance analysis concepts and tools are well researched and widely available for traditional parallel applications but are lacking in RC, despite being of great importance due to the applications' increased complexity. In this paper we explore challenges and present new techniques in automated instrumentation, runtime measurement, and visualization of RC application behavior. We also present ideas for integration with conventional performance analysis tools to create a unified tool for RC applications as well as our initial framework for FPGA instrumentation and measurement. Results from a case study are provided using a prototype of this new tool.

Crossing Boundaries in TimeTrial: Monitoring Communications across Architecturally Diverse Computing Platforms

“…There has been a great deal of work in monitoring the performance of executing applications, from gprof [9] and TAU [18] for platforms using traditional processor cores to the tools of Koehler et al [12] aimed at FPGA designs. Each of the above systems is tunable with respect to the impact it can have on the performance of the application (i.e., the tradeoff between performance impact and measurement fidelity is controlled by the user).…”

Performance Visualization for Large-Scale Computing Systems: A Literature Review

“…According to literature on performance analysis and visualization [1,3,4,11,12], the process of performance visualization generally consists of four major steps: instrumentation (enabling access to performance data to be measured), measurement (recording selected data during the run-time of the program), data analysis (analyzing data for performance visualization), and visualization (mapping performance characteristics to proper visual representations and interactions).…”

“…The data to be collected should reflect application performance as closely as possible while minimizing perturbation of that behavior as much as possible (tradeoff between fidelity and perturbation) [12]. The instrumentation may reside in hardware or in software.…”

“…Colors are used to indicate the computational job and associated performance data, and thick line segments are used to show routing information once the route source and destination have been selected. Information typography can also be used to integrate logical architectures, as in [12]. Information landscape.…”

Understanding Resource Provisioning for ClimatePrediction.net

“…In order to conduct a reliable performance analysis of the ClimatePrediction.net host resources, we appeal to the fivestage decomposition of this topic as proposed in [15].…”