Thicket: Seeing the Performance Experiment Forest for the Individual Run Trees

Brink, Stephanie; McKinsey, Michael; Boehme, David; Scully-Allison, Connor; Lumsden, Ian; Hawkins, Daryl; Burgess, Treece; Lama, Vanessa; Lüttgau, Jakob; Isaacs, Katherine E.; Taufer, Michela; Pearce, Olga

doi:10.1145/3588195.3592989

Cited by 3 publications

(1 citation statement)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Profiling tools classification: Profiling tools collect finegrained execution profiles and generate calling contexts to help facilitate several performance analysis tasks, such as load balancing analysis detection [22], roofline analysis [23] or variability analysis [3], [24]. Numerous profiling tools have been developed for performance profiling of parallel applications [12], [10], [9], [25], [26]. Broadly, profiling tools can be classified into two categories: (1) statistical sampling, where system-generated interrupts are triggered using either time-based or event-based sampling, and (2) instrumentation, where measurement probes are placed directly either in the source code or by modifying the generated binary program.…”

Section: Related Workmentioning

confidence: 99%

TinyProf: Towards Continuous Performance Introspection through Scalable Parallel I/O

Fan,

Kesavan,

Petruzza

et al. 2024

ISC High Performance 2024 Research Paper Proceedings (39th International Conference)

View full text Add to dashboard Cite

Performance profiling tools are crucial for HPC specialists to identify performance bottlenecks in parallel codes at various levels of granularity (i.e., across nodes, ranks, and threads). Although numerous sophisticated profiling tools have been developed, achieving scalable performance introspection on large scales remains a challenge. This is particularly evident in efficiently writing profiles to disk during runtime and subsequently reading them with constrained computing resources for posthoc analysis. In this paper, we present TinyProf , a performance introspection framework that tackles I/O-related challenges in profiling performance data at scale. TinyProf 's scalability is attributed to an optimal runtime that consists of three key components: (1) an efficient in-memory data structure that minimizes memory consumption and decreases communication overhead during parallel file I/O; (2) a customizable threephase I/O scheme that generates optimal I/O patterns capable of scaling with high core counts; and (3) a streamlined data format for profiles, which guarantees minimal sizes for profile files. These three techniques instill scalability into the profiler, making it low overhead, even at high process counts (less than 5%). This low overhead makes it possible for the profiler to be run with an application as a default (whenever the application is running)-enabling continuous introspection of performance. We demonstrate the efficiency of our framework using large-scale parallel applications and perform a thorough evaluation against existing systems up to 32k processes.

show abstract