Hardware Performance Monitoring for the Rest of Us: A Position and Survey

Moseley, Tipp; Vachharajani, Neil; Jalby, William

doi:10.1007/978-3-642-24403-2_23

Cited by 7 publications

(10 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Performance counters are a common profiling tool embedded in designs for post-silicon performance introspection [37]. However, since these counters are included as part of the final silicon design's area, power, and other budgets, they are generally limited in number and frequently shared amongst many events, complicating the process of extracting meaningful information from them [34].…”

Section: Hardware Profiling With Autocountermentioning

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

Section: Hardware Profiling With Autocountermentioning

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

show abstract

“…Previous works have aimed to evaluate the true accuracy of performance monitoring counters [7,21,22]. However, little work has been done to understand the accuracy and limitations of sampling techniques such as HEM across different configurations and profiling targets, pointed out as a key issue facing hardware performance monitoring by Moseley et al [12]. Below, we present what we believe is the first quantitative study on the accuracy of event counts generated from executions monitored via HEM specifically with respect to three impacting variables: |E HEM |, π , and the average task duration.…”

Section: Hardware Event Multiplexingmentioning

confidence: 99%

“…Analysis of hardware behavior using PMCs is significantly limited, however, as the number of available events is generally orders of magnitude higher than the number of PMCs, identified as a key issue facing hardware performance monitoring [12]. Hence, only a small subset of possible events can be monitored simultaneously during program execution.…”

Section: Introductionmentioning

confidence: 99%

Fuse

Neill

Drebes

Pop

2017

ACM Trans. Archit. Code Optim.

View full text Add to dashboard Cite

Collecting hardware event counts is essential to understanding program execution behavior. Contemporary systems offer few Performance Monitoring Counters (PMCs), thus only a small fraction of hardware events can be monitored simultaneously. We present new techniques to acquire counts for all available hardware events with high accuracy by multiplexing PMCs across multiple executions of the same program, then carefully reconciling and merging the multiple profiles into a single, coherent profile. We present a new metric for assessing the similarity of statistical distributions of event counts and show that our execution profiling approach performs significantly better than Hardware Event Multiplexing.

show abstract

“…that its actual behaviour matches the schedule. This will rely on observation.We implement the program tracing system into the program itself [11] [15].…”

Section: Motivation and Related Workmentioning

confidence: 99%