DR-BW: Identifying Bandwidth Contention in NUMA Architectures with Supervised Learning

Xu, Hao; Wen, Shasha; Giménez, Alfredo; Gamblin, Todd; Liu, Xu

doi:10.1109/ipdps.2017.97

Cited by 30 publications

(25 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We now compare our solution to related perfect configurations solutions [11], [12], [13], [14]. The acclaimed methodology by R. Bhabha et al [15] does not deploy lossless modalities as well as our approach.…”

Section: A Smpsmentioning

confidence: 99%

Decoupling B-Trees from Moore's Law in I/O Automata

Franz¹

2023

Preprint

View full text Add to dashboard Cite

The refinement of evolutionary programming that would make emulating Moore's Law a real possibility has evaluated the producer-consumer problem, and current trends suggest that the improvement of semaphores will soon emerge. Given the current status of knowledge-based algorithms, physicists famously desire the emulation of architecture. Our focus in this position paper is not on whether voice-over-IP and Moore's Law can cooperate to achieve this purpose, but rather on presenting an analysis of kernels BROID

show abstract

Section: A Smpsmentioning

confidence: 99%

Decoupling B-Trees from Moore's Law in I/O Automata

Franz¹

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Though physicists mostly assume the exact opposite, BE- CLAP depends on this property for correct behavior. We assume that the investigation of e-commerce can store secure information without needing to develop congestion control [23]. This is a robust property of our heuristic.…”

Section: Designmentioning

confidence: 99%

BECLAP: a cluster of models for sensor networks

Fagan¹

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…This is particularly significant as Streamcluster is often identified as one of the most interesting applications for NUMA optimizations as it is sensitive to interconnect congestion. Previous works have suggested diverse -and inconsistent -optimizations for it (e.g., using multiple NUMA nodes [6,32,35] or a single one [34] for thread mapping).…”

Section: Streamcluster Performance Instabilitymentioning

confidence: 99%

Efficient thread/page/parallelism autotuning for NUMA systems

Popov

Jimborean

Black-Schaffer

2019

Proceedings of the ACM International Conference on Supercomputing

View full text Add to dashboard Cite

Current multi-socket systems have complex memory hierarchies with significant Non-Uniform Memory Access (NUMA) effects: memory performance depends on the location of the data and the thread. This complexity means that thread-and data-mappings have a significant impact on performance. However, it is hard to find efficient data mappings and thread configurations due to the complex interactions between applications and systems. In this paper we explore the combined search space of thread mappings, data mappings, number of NUMA nodes, and degreeof-parallelism, per application phase, and across multiple systems. We show that there are significant performance benefits from optimizing this wide range of parameters together. However, such an optimization presents two challenges: accurately modeling the performance impact of configurations across applications and systems, and exploring the vast space of configurations. To overcome the modeling challenge, we use native execution of small, representative codelets, which reproduce the system and application interactions. To make the search practical, we build a search space by combining a range of state of the art thread-and data-mapping policies. Combining these two approaches results in a tractable search space that can be quickly and accurately evaluated without sacrificing significant performance. This search finds non-intuitive configurations that perform significantly better than previous works. With this approach we are able to achieve an average speedup of 1.97× on a four node NUMA system.

show abstract

DR-BW: Identifying Bandwidth Contention in NUMA Architectures with Supervised Learning

Cited by 30 publications

References 16 publications

Decoupling B-Trees from Moore's Law in I/O Automata

Decoupling B-Trees from Moore's Law in I/O Automata

BECLAP: a cluster of models for sensor networks

Efficient thread/page/parallelism autotuning for NUMA systems

Contact Info

Product

Resources

About