BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:America/New_York X-LIC-LOCATION:America/New_York BEGIN:DAYLIGHT TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:EDT DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:EST DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20210402T160557Z LOCATION:Track 7 DTSTART;TZID=America/New_York:20201113T100000 DTEND;TZID=America/New_York:20201113T164500 UID:submissions.supercomputing.org_SC20_sess226@linklings.com SUMMARY:ROSS 2020: 10th International Workshop on Runtime and Operating Sy stems for Supercomputers DESCRIPTION:Workshop\n\nROSS – Break\n\n\n\n---------------------\nROSS – Break\n\n\n\n---------------------\nROSS – Introduction: 10th Internationa l Workshop on Runtime and Operating Systems for Supercomputers\n\nIskra, H oefler, Gerofi\n\nThe complexity of node architectures in supercomputers i ncreases as we cross petaflop milestones on the way towards exascale. Inc reasing levels of parallelism in multi- and many-core chips and emerging h eterogeneity of computational resources coupled with energy and memory con straints force a re-ev...\n\n---------------------\nPinpoint the Joules: U nifying Runtime-Support for Energy Measurements on Heterogeneous Systems\n \nKöhler, Herzog, Hönig, Wenzel, Plauth...\n\nFor the design and operation of today's computer systems, power and energy requirements are highest pr iorities. Unlike performance analyses, however, power and energy measureme nts of heterogeneous systems are difficult to conduct. Especially at the s ystem-software level, performing power and energy ...\n\n----------------- ----\nImproving Job Launch Rates in the TaPaSCo FPGA Middleware by Hardwar e/Software-Co-Design\n\nHeinz, Hofmann, Sommer, Koch\n\nIn recent years, F PGAs have established themselves as important acceleration platforms next to GPUs in heterogeneous HPC systems, providing flexibility and high perfo rmance for tasks such as machine learning inference or DNA sequencing. Whi le the design of the FPGA-based accelerator cores has become...\n\n------- --------------\nLocality-Aware Scheduling for Scalable Heterogeneous Envir onments\n\nKamatar, Friese, Gioiosa\n\nHeterogeneous computing promises to boost performance of scientific applications by allowing massively parall el execution of computational tasks. Manually managing extremely heterogen eous, multi-device systems is complicated, however, and may result in sub- optimal performance. Specifically, data man...\n\n---------------------\nT owards Generalizable Models of I/O Throughput\n\nIsakov, del Rosario, Madi reddy, Balaprakash, Carns...\n\nMany modern HPC applications do not make g ood use of the limited available I/O bandwidth. Developing an understandin g of the I/O subsystem is a critical first step in order to better utilize an HPC system. While expert insight is indispensable, I/O experts are in rare supply. We seek to automate thi...\n\n---------------------\nCh’i: Sc aling Microkernel Capabilities in Cache-Incoherent Systems\n\nRen, Parmer, Milojicic\n\nHardware cache coherence limits the scalability of shared-me mory multicore and multi-processors. Recently, there has been an increasin g shift toward cache-incoherent architectures in many computing environmen ts. Supercomputers can benefit from shared memory, yet must avoid the pric e of coherency acr...\n\n---------------------\nApplication-Driven Require ments for Node Resource Management in Next-Generation Systems\n\nLeon, Ger ofi, Jaeger, Mercier, Riesen...\n\nEmerging workloads on supercomputing pl atforms are pushing the limits of traditional high-performance computing s oftware environments. Multi-physics, coupled simulations, big data process ing and machine learning frameworks, and multi-component workloads pose se rious challenges to system and applicat...\n\n---------------------\nROSS – Closing Remarks\n\n\n\n---------------------\nROSS – Break\n\n\n\n------ ---------------\nProgramming Model Developments Present Opportunities for Runtime and Operating Systems\n\nOlivier\n\nChanges in high performance co mputing systems and the workloads they support have prompted parallel prog ramming models to either evolve or risk losing relevance. System heterogen eity extends beyond manycore and GPU acceleration to complex memory and ne twork configurations. Problems of interest are n...\n\n------------------- --\nA Dataflow-Graph Partitioning Method for Training Large Deep Learning Models\n\nUnat, Qararyah\n\nLarge Deep Neural Network (DNNs) models have s ubstantial memory requirements to store the model parameters and intermedi ate results. As a result, the limited device memory becomes a bottleneck w hen training those models. We propose a deterministic, generic and efficie nt partitioning strategy for DNN...\n\n---------------------\nThe Challeng es of Hybrid Computing\n\nKissell\n\nIt has been clear for some time that the future of high-performance computing is hybrid, and "hybrid" in multip le senses of the word. Hybrid CPU/GPU, hybrid physics based/ML and hybrid on-prem/cloud are just the most obvious cross-fertilizations of technology in pursuit of higher performance, lower ...\n\n---------------------\nDat a-Centric Resource Management for Complex Memory Fabrics\n\nGavrilovska\n\ nThe continued need for scaling of performance, capacity, and cost efficie ncy for data-intensive applications, is giving rise to new system designs combining heterogeneous memory components, accelerators and accelerator-ne ar memory, interconnected with programmable high-performance interconnects . Max...\n\n\nTag: System Software and Runtime Systems\n\nRegistration Cat egory: Workshop Reg Pass END:VEVENT END:VCALENDAR