Accurate phase-level cross-platform power and performance estimation

Zheng, Xinnian; John, Lizy K.; Gerstlauer, Andreas

doi:10.1145/2897937.2897977

Cited by 51 publications

(34 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are many diferent types of forecasting, but all rely on the same principle: constructing a forecast model of the system under study which can be used to predict how the system will behave under unobserved conditions. In the ield of real-time systems, Zheng [31] applied linear regression techniques to relate the amount of resource accesses (obtained via the PMCs) to the inlation in the execution times caused by the accesses; however, in our experience, a linear relationship does not hold for most tasks and platforms. Griin et al [9] employed forecasting to determine information about the behaviour of tasks when their execution time budgets were exceeded, by constructing a model based on the observed behaviour of the tasks' execution times.…”

Section: Related Workmentioning

confidence: 99%

Forecast-based interference

Griffin

Lesage

Bate

et al. 2017

Proceedings of the 25th International Conference on Real-Time Networks and Systems

View full text Add to dashboard Cite

While there is signiicant interest in the use of COTS multicore platforms for Real-time Systems, there has been very little in terms of practical methods to calculate the interference multiplier (i.e. the increase in execution time due to interference) between tasks on such systems. COTS multicore platforms present two distinct challenges: irstly, the variable interference between tasks competing for shared resources such as cache, and secondly the complexity of the hardware mechanisms and policies used, which may result in a system which is very diicult if not impossible to analyse; assuming that the exact details of the hardware are even disclosed! This paper proposes a new technique, Forecast-Based Interference analysis, which mitigates both of these issues by combining measurement-based techniques with statistical techniques and forecast modelling to enable the prediction of an interference multiplier for a given set of tasks, in an automated and reliable manner. The combination of execution times and interference multipliers can be used both in the design, e.g. for specifying timing watchdogs, and analysis, e.g. verifying schedulability. ACM Reference format:

show abstract

Section: Related Workmentioning

confidence: 99%

Forecast-based interference

Griffin

Lesage

Bate

et al. 2017

Proceedings of the 25th International Conference on Real-Time Networks and Systems

View full text Add to dashboard Cite

show abstract

“…Abstract system-level simulators have long been used in the architecture and design automation communities for performance estimation and analysis [11,19,27,38,39,42,44,49,52]. In particular, [12] used system simulation to evaluate the interaction between the OS and a 10 Gbit/s Ethernet NIC.…”

Section: Related Workmentioning

confidence: 99%

FirePerf

Karandikar

Amid

et al. 2020

Proceedings of the Twenty-Fifth International Conference on Architectural Support for Programming Languages and Operating Syste

View full text Add to dashboard Cite

Achieving high-performance when developing specialized hardware/software systems requires understanding and improving not only core compute kernels, but also intricate and elusive system-level bottlenecks. Profiling these bottlenecks requires both high-fidelity introspection and the ability to run sufficiently many cycles to execute complex software stacks, a challenging combination. In this work, we enable agile full-system performance optimization for hardware/ software systems with FirePerf, a set of novel out-of-band system-level performance profiling capabilities integrated into the open-source FireSim FPGA-accelerated hardware simulation platform. Using out-of-band call stack reconstruction and automatic performance counter insertion, FirePerf enables introspecting into hardware and software at appropriate abstraction levels to rapidly identify opportunities for software optimization and hardware specialization, without disrupting end-to-end system behavior like traditional profiling tools. We demonstrate the capabilities of FirePerf with a case study that optimizes the hardware/software stack of an open-source RISC-V SoC with an Ethernet NIC to achieve 8× end-to-end improvement in achievable bandwidth for networking applications running on Linux. We also deploy a RISC-V Linux kernel optimization discovered with FirePerf on commercial RISC-V silicon, resulting in up to 1.72× improvement in network performance.

show abstract

“…RAPL registers can be used to keep track of very fast programming events, as demonstrated by Hähnel et al [Hähnel et al (2012)]. Zheng et al [Zheng et al (2016)] proposes a learning algorithm that uses hardware counter information to estimate power consumption of different platforms. Along similar lines, Stokke et al [Stokke et al (2015[Stokke et al ( , 2016] have built what is possibly the most precise energy model nowadays available for the TK1 board using counters.…”

Section: Related Workmentioning

confidence: 99%

JetsonLEAP: A framework to measure power on a heterogeneous system-on-a-chip device

Bessa

Gull

Quintão

et al. 2019

Science of Computer Programming

View full text Add to dashboard Cite

Computer science marches towards energy-aware practices. This trend impacts not only the design of computer architectures, but also the design of programs. However, developers still lack affordable and accurate technology to measure energy consumption in computing systems. The goal of this paper is to mitigate such problem. To this end, we introduce JetsonLEAP, a framework that supports the implementation of energy-aware programs. Jet-sonLEAP consists of an embedded hardware, in our case, the Nvidia Tegra TK1 System-on-a-chip device, a circuit to control the flow of energy, of our own design, plus a library to instrument program parts. We discuss two different circuit setups. The most precise setup lets us reliably measure the energy spent by 225,000 instructions, the least precise, although more affordable setup, gives us a window of 975,000 instructions. To probe the precision of our system, we use it in tandem with a high-precision, high-cost acquisition system, and show that results do not differ in any significant way from those that we get using our simpler apparatus. Our entire infrastructure -board, power meter and both circuits -can be reproduced with about $500.00. To demonstrate the efficacy of our framework, we have used it to measure the energy consumed by programs running on ARM cores, on the GPU, and on a remote server. Furthermore, we have studied the impact of OpenACC directives on the energy efficiency of high-performance applications.

show abstract

Accurate phase-level cross-platform power and performance estimation

Cited by 51 publications

References 15 publications

Forecast-based interference

Forecast-based interference

FirePerf

JetsonLEAP: A framework to measure power on a heterogeneous system-on-a-chip device

Contact Info

Product

Resources

About