Towards Dynamic Cache and Bandwidth Invasion

Tradowsky, Carsten; Schreiber, Martin; Vesper, Malte; Domladovec, Ivan; Braun, Michael; Bungartz, Hans‐Joachim; Becker, J.

doi:10.1007/978-3-319-05960-0_9

Cited by 1 publication

(1 citation statement)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our current work is on systems with run time configurable cache-coherency protocols and cache-levels [43] to further enhance the performance and programmability of parallel applications on such systems. With our simulations conducted on an HPC shared-memory NUMA domain which requires only 1 additional hop to each domain, thus hiding the NUMA domain effects very efficiently, we expect that evaluation of the invasive concepts leads to additional requirements on larger scale NUMA domains.…”

Section: Discussionmentioning

confidence: 99%

Invasive Compute Balancing for Applications with Shared and Hybrid Parallelization

Schreiber

Riesinger

Neckel

et al. 2014

Int J Parallel Prog

Self Cite

View full text Add to dashboard Cite

Achieving high scalability with dynamically adaptive algorithms in high-performance computing (HPC) is a non-trivial task. The invasive paradigm using compute migration represents an efficient alternative to classical data migration approaches for such algorithms in HPC. We present a coredistribution scheduler which realizes the migration of computational power by distributing the cores depending on the requirements specified by one or more parallel program instances. We validate our approach with different benchmark suites for simulations with artificial workload as well as applications based on dynamically adaptive shallow water simulations, and investigate concurrently executed adaptivity parameter studies on realistic Tsunami simulations. The invasive approach results in significantly faster overall execution times and higher hardware utilization than alternative approaches. A dynamic resource management is therefore mandatory for a more efficient execution of scenarios similar to our simulations, e.g. several Tsunami simulations in urgent computing, to overcome strong scalability challenges in the area of HPC. The optimizations obtained by invasive migration of cores can be generalized to similar classes of algorithms with dynamic resource requirements.

show abstract

Section: Discussionmentioning

confidence: 99%