BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:America/Chicago X-LIC-LOCATION:America/Chicago BEGIN:DAYLIGHT TZOFFSETFROM:-0600 TZOFFSETTO:-0500 TZNAME:CDT DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0500 TZOFFSETTO:-0600 TZNAME:CST DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260522T150153Z LOCATION:C2/3/4 Ballroom DTSTART;TZID=America/Chicago:20181113T083000 DTEND;TZID=America/Chicago:20181113T170000 UID:submissions.supercomputing.org_SC18_sess340@linklings.com SUMMARY:Exhibition of Doctoral Showcase Posters DESCRIPTION:Fast and Generic Concurrent Message-Passing\n\nCommunication h ardware and software have a significant impact on the performance of clust ers and supercomputers. Message-passing model and the Message-Passing Inte rface (MPI) is a widely used model of communications in the High-Performan ce
Computing (HPC) community. However, MPI has r...\n\n\nHoang-Vu Dan g and Marc Snir (University of Illinois)\n---------------------\nDesigning High-Performance, Resilient, and Heterogeneity-Aware Key-Value Storage fo r Modern HPC Clusters\n\nDistributed key-value stores are being increasing ly used to accelerate Big Data workloads on modern HPC clusters. The advan ces in HPC technologies (e.g., RDMA, SSDs) has directed several efforts to wards employing hybrid storage with RDMA, for designing high- performance key-value stores. With this a...\n\n\nDipti Shankar, Dhabaleswar K. Panda, and Xiaoyi Lu (Ohio State University)\n---------------------\nScalable No n-Blocking Krylov Solvers for Extreme-Scale Computing\n\nThis study invest igates preconditioned conjugate gradient method variations designed to red uce communication costs by decreasing the number of allreduces and overlap ping communication with computation using a non-blocking allreduce. Experi ments show scalable PCG methods can outperform standard PCG a...\n\n\nPaul R. Eller and William Gropp (University of Illinois)\n-------------------- -\nParallel and Scalable Combinatorial String and Graph Algorithms on Dist ributed Memory Systems\n\nMethods for processing and analyzing DNA and gen omic data are built upon combinatorial graph and string algorithms. The ad vent of high-throughput DNA sequencing is enabling the generation of billi ons of reads per experiment. Classical and sequential algorithms can no lo nger deal with these growing d...\n\n\nPatrick Flick and Srinivas Aluru (G eorgia Institute of Technology)\n---------------------\nScalable Methods f or Genome Assembly\n\nGenome assembly is a fundamental problem in the fiel d of bioinformatics wherein the goal lies in the reconstruction of an unkn own genome from short DNA fragments obtained from it. With the advent of h igh-throughput sequencing technologies, billions of reads can be generated in a few hours. My reseac...\n\n\nPriyanka Ghosh and Ananth Kalyanaraman (Washington State University)\n---------------------\nHigh Performance Mid dlewares for Next Generation Architectures: Challenges and Solutions\n\nTh e emergence of modern multi-/many-core architectures and high-performance interconnects have fueled the growth of large-scale supercomputing cluster s. Due to this unprecedented growth in scale and compute density, high per formance computing (HPC) middlewares now face a plethora of new challenges t...\n\n\nSourav Chakraborty and Dhabaleswar K. Panda (Ohio State Univers ity)\n---------------------\nProductive Data Locality Optimizations in Dis tributed Memory\n\nWith deepening memory hierarchies in HPC systems, the c hallenge of managing data locality gains more importance. Coincidentally, increasing ubiquity of HPC systems and wider range of disciplines utilizin g HPC introduce more programmers to the HPC community. Given these two tre nds, it is imperative t...\n\n\nEngin Kayraklioglu and Tarek El-Ghazawi (G eorge Washington University)\n---------------------\nPattern Matching on M assive Metadata Graphs at Scale\n\nPattern matching is a powerful graph an alysis tool. Unfortunately, existing solutions have limited scalability, s upport only a limited set of patterns, and/or focus on only a subset of th e real-world problems associated with pattern matching. First, we present a new algorithmic pipeline based on gra...\n\n\nTahsin Reza and Matei Ripe anu (University of British Columbia)\n---------------------\nEnabling Effi cient Data Infrastructure and Analytics on HPC Systems\n\nWe propose to le verage PGAS and one-sided communication for building data infrastructure a nd analytics frameworks on HPC systems. Specifically, we have developed SH MEMCache, a distributed in-memory key-value store and SHMEMGraph, a balanc ed graph processing framework. We have tackled unique challeng...\n\n\nHua nsong Fu and Weikuan Yu (Florida State University)\n---------------------\ nUsing Integrated Processor Graphics to Accelerate Concurrent Data and Ind ex Structures\n\nWith the advent of computing systems with on-die integrat ed processor graphics (iGPU), new programming challenges have emerged from these heterogeneous systems. We proposed different data and index structu re algorithms that can benefit from the Intel's iGPU architecture and the C for Media (CM) prog...\n\n\nJoel Fuentes and Isaac Scherson (University of California, Irvine)\n---------------------\nIn-Memory Accelerator Archi tectures for Machine Learning and Bioinformatics\n\nMost contemporary acce lerators are von Neumann machines. With the increasing sizes of gathered a nd then processed data,  memory bandwidth is the main limiting of performa nce. One approach to mitigate the bandwidth constraint is to bring the pro cessing units closer to the data. This approach is ...\n\n\nRoman Kaplan a nd Ran Ginosar (Israel Institute of Technology)\n---------------------\nLi near Algebra Is the Right Way to Think About Graphs\n\nGraph algorithms ar e challenging to implement on new accelerators such as GPUs. To address th is problem, GraphBLAS is an innovative on-going effort by the graph analyt ics community to formulate graph algorithms as sparse linear algebra, so t hat they can be expressed in a performant, succinct and in ...\n\n\nCarl Y ang (University of California, Davis; Lawrence Berkeley National Laborator y); John D. Owens (University of California, Davis); and Aydin Buluc (Lawr ence Berkeley National Laboratory; University of California, Berkeley)\n-- -------------------\nThe Algorithm and Framework Designs and Optimizations for Scalable Automata Processing on HPC Platforms\n\nAutomata processing could perform as the core of many applications in the areas such as networ k security, text mining, and bioinformatics. Achieving high-speed and scal able automata processing is exceptionally challenging. For one thing, the classic DFA representation is memory-bandwidth efficient b...\n\n\nXiaodon g Yu and Danfeng Yao (Virginia Tech)\n---------------------\nHardware Tran sactional Persistent Memory\n\nThis research solves the problem of creatin g durable transactions in byte-addressable Non-Volatile Memory or Persiste nt Memory (PM) when using Hardware Transactional Memory (HTM)-based concur rency control.  It shows how HTM transactions can be ordered correctly and atomically into PM by the use...\n\n\nEllis Giles and Peter Varman (Rice University)\n---------------------\nEfficient Deployment of Irregular Comp utations on Multi- and Many-Core Architectures\n\nMulti- and manycore proc essors have been advancing High Performance Computing with their high thro ughput and power efficiency. There has been an increasing interest in acce lerating irregular computations on these devices that offer massive parall elism. My thesis focuses on compiler techniques and co...\n\n\nHancheng Wu and Michela Becchi (North Carolina State University)\n------------------- --\nCompiler and Runtime Based Parallelization and Optimization for GPUs\n \nThis thesis targets directive-based programming models to enhance their capability for GPU programming.  It introduces a new dialect model, which is a combination of OpenMP and OmpSs. The new model allows the use of mult iple GPUs in conjunction with the heavily multithreaded capabilities in mu l...\n\n\nGuray Ozen, Jesus Labarta, and Eduard Ayguade (Barcelona Superco mputing Center, Polytechnic University of Catalonia)\n\nRegistration Categ ory: Workshop Reg Pass, Tutorial Reg Pass, Tech Program Reg Pass, Exhibits Reg Pass, Exhibits - Exhibit Hall Only Reg Pass END:VEVENT END:VCALENDAR