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:20181221T160904Z
LOCATION:C2/3/4 Ballroom
DTSTART;TZID=America/Chicago:20181113T083000
DTEND;TZID=America/Chicago:20181113T170000
UID:submissions.supercomputing.org_SC18_sess325@linklings.com
SUMMARY:ACM Student Research Competition Posters
DESCRIPTION:ACM Student Research Competition, Poster\nTech Program Reg Pas
 s, Exhibits Reg Pass\n\nRecursive Algebraic Coloring Engine\n\nAlappat\n\n
 Many iterative numerical methods for sparse systems and building blocks of
  sparse linear algebra are difficult to parallelize due to data dependenci
 es. These may be loop-carried dependencies as they occur in solvers like G
 auss-Seidel or write conflicts as in symmetric sparse matrix vector.  Most
  of ...\n\n---------------------\nNautDB: Toward a Hybrid Runtime for Proc
 essing Compiled Queries\n\nGrayson\n\nGeneral purpose operating and databa
 se system suffer under the load of their generality which makes achieving 
 optimal performance extremely hard, especially on modern hardware. The goa
 l of this research is to integrate, for the first time, specialization tec
 hniques from the OS community (hybrid runt...\n\n---------------------\nMi
 tigating Performance and Progress Variability in Iterative Asynchronous Al
 gorithms\n\nZarins\n\nLarge HPC machines are susceptible to irregular perf
 ormance. Factors like chip manufacturing differences, heat management, and
  network congestion combine to result in varying execution time for the sa
 me code and input sets. Asynchronous algorithms offer a partial solution. 
 In these algorithms, fast ...\n\n---------------------\nAccelerating DNA L
 ong Read Mapping with Emerging Technologies\n\nKaplan\n\nDNA sequencing te
 chnologies output only short fragments of a genome, called reads. New sing
 le-molecule real-time sequencing technologies can produce long reads, up t
 o tens of thousands base pairs, within minutes. However, these long reads 
 may contain up to 15% errors.\n\nTo construct a genome from DNA...\n\n----
 -----------------\nSimFS: A Simulation Data Virtualizing File System Inter
 face\n\nDi Girolamo\n\nIn the big (simulation) data era, simulations often
  produce petabytes of data to be stored in parallel filesystems or large-s
 cale databases. This data is accessed, often by thousands of analysts and 
 scientists, over the course of decades. However, storing these volumes of 
 data for long time periods ...\n\n---------------------\nHolistic Root Cau
 se Analysis of Node Failures in Production HPC\n\nDas\n\nProduction HPC cl
 usters endure failures incurring computation and resource wastage. Despite
  the presence of various failure detection and prediction schemes, a compr
 ehensive understanding of how nodes fail considering various components an
 d layers of the system is required for sustained resilience. ...\n\n------
 ---------------\nGeomancy: Automated Data Placement Optimization\n\nBel\n\
 nExascale cloud storage and High-Performance Computing Systems (HPC) deliv
 er unprecedented storage capacity and levels of computing power, though th
 e full potential of these systems remain untapped because of inefficient d
 ata placement. Changes in data access patterns can cause a system's perfor
 mance...\n\n---------------------\nNumerical Simulation of a Flue Instrume
 nt with Finite-Difference Lattice Boltzmann Method using GPGPU\n\nTabata\n
 \nIn this work, we discuss the possibility of using GPGPU techniques for A
 eroacoustic Simulation (especially for flue instruments) with the finite-d
 ifference lattice Boltzmann method (FDLBM).   Compressible flow simulation
  has been used in direct aeroacoustic simulation; however, the computation
 al cos...\n\n---------------------\nPrecomputing Outputs of Hidden Layers 
 to Speed Up Deep Neural Network Training\n\nShrestha\n\nDeep learning has 
 recently emerged as a powerful technique for many tasks including image cl
 assification. A key bottleneck of deep learning is that the training phase
  takes a lot of time, since state-of-the-art deep neural networks have mil
 lions of parameters and hundreds of hidden layers. The early...\n\n-------
 --------------\nAccelerating Microscope Data Analysis Using Parallel Compu
 ting\n\nRavi\n\nSingle-Molecule Localization Microscopy (SMLM) techniques 
 deal with the diffraction limit of fluorescent microscopy by localizing si
 ngle molecules with high precision by stochastically switching molecules o
 n and off. Thousands of camera frames containing subsets of blinking molec
 ules are recorded to...\n\n---------------------\nUsing Integrated Process
 or Graphics to Accelerate Concurrent Data and Index Structures\n\nFuentes\
 n\nWith the advent of computing systems with on-die integrated processor g
 raphics (iGPU), new programming challenges have emerged from these heterog
 eneous systems. We proposed different data and index structure algorithms 
 that can benefit from the Intel's iGPU architecture and the C for Media (C
 M) prog...\n\n---------------------\nPotC: Many-Body Potential Implementat
 ions à La Carte\n\nHöhnerbach\n\nMolecular dynamics is a valuable investig
 ation tool for simulations in computational chemistry and materials scienc
 e. In these simulations, atoms move according to so-called potentials, fun
 ctions that typically describe the distance-dependent interactions between
  pairs of atoms. For some application...\n\n---------------------\nOoO Ins
 truction Benchmarking Framework on the Back of Dragons\n\nHammer\n\nIn ord
 er to construct an accurate instruction execution model for modern out-of-
 order micro architectures, an accurate description of instruction latency,
  throughput and concurrency is indispensable. Already existing resources a
 nd vendor provided information is neither complete nor detailed enough a..
 .\n\n---------------------\nStudying the Impact of Power Capping on MapRed
 uce-Based, Data-Intensive Mini-Applications on Intel KNL and KNM Architect
 ures\n\nDavis\n\nIn this poster, we quantitatively measure the impacts of 
 data movement on performance in MapReduce-based applications when executed
  on HPC systems. We leverage the PAPI ‘powercap’ component to identify ide
 al conditions for execution of our applications in terms of (1) dataset ch
 aracteristics (i.e., ...\n\n---------------------\nAccelerating Big Data P
 rocessing in the Cloud with Scalable Communication and I/O Schemes\n\nGugn
 ani\n\nWith the advent of cloud computing, the field of Big Data has seen 
 rapid growth. Most cloud providers provide hardware resources such as NVMe
  SSDs, large memory nodes, and SR-IOV. This opens up the possibility of la
 rge-scale high-performance data analytics and provides opportunities to us
 e these res...\n\n---------------------\nAccelerating 2D FFT: Exploit GPU 
 Tensor Cores through Mixed-Precision\n\nCheng, Sorna\n\nThe two-dimensiona
 l Fourier Transform is a widely-used computational kernel in many HPC appl
 ications. The popular NVIDIA cuFFT library provides a simple interface to 
 compute 2D FFT on GPUs, but it's yet to utilize the recent hardware advanc
 ement in half-precision floating-point arithmetic. In this p...\n\n-------
 --------------\nMonitoring Parsl Workflows\n\nPigg\n\nAs a Python library 
 that enables workflows, Parsl gives users the ability to define complex wo
 rkflows in Python and run them in parallel on any computer system. This po
 ster describe the process of adding monitoring to Parsl. Simple and compre
 hensive monitoring of a workflow’s state and resource usag...\n\n---------
 ------------\nIdentifying Network Data Transfer Bottlenecks in HPC Systems
 \n\nTu\n\nImproving network data transfer performance is a major factor fo
 r improving high performance computing systems. Most studies analyze data 
 transfer and file system IO performance separately, but understanding the 
 relationship between the two is essential for optimizing scheduling and re
 source managem...\n\n---------------------\nSupercomputing for the Multi-D
 river Routing\n\nYe\n\nSupercomputing is essential for routing traffic by 
 providing drivers the optimal routes with minimal traveling distances or t
 ime. The unique challenges that require supercomputers to overcome are of 
 multiple folds: numerous drivers, massive simultaneous requests, multiple 
 locations, and needs of ins...\n\n---------------------\nHardware Transact
 ional Persistent Memory\n\nGiles\n\nThis research solves the problem of cr
 eating durable transactions in byte-addressable Non-Volatile Memory or Per
 sistent Memory (PM) when using Hardware Transactional Memory (HTM)-based c
 oncurrency control.  It shows how HTM transactions can be ordered correctl
 y and atomically into PM by the use of a...\n\n---------------------\nMeas
 uring Swampiness: Quantifying Chaos in Large Heterogeneous Data Repositori
 es\n\nJung, Whitaker\n\nAs scientific data repositories and filesystems gr
 ow in size and complexity, they become increasingly disorganized. The coup
 ling of massive quantities of data with poor organization makes it challen
 ging for scientists to locate and utilize relevant data, thus slowing the 
 process of analyzing data of...\n\n---------------------\nDendro-GR: Massi
 vely Parallel Simulations of Binary Black Hole Intermediate-Mass-Ratio Ins
 pirals\n\nFernando\n\nWe present a portable and highly-scalable algorithm 
 and framework that targets problems in the astrophysics and numerical rela
 tivity communities. This framework combines together a parallel octree-ref
 ined adaptive mesh with wavelet adaptive multiresolution and a physics mod
 ule to solve the Einstein ...\n\n---------------------\nDesigning Shared A
 ddress Space MPI Libraries in Many-Core Era\n\nHashmi\n\nThe emergence of 
 modern multi-/many-cores has put more emphasis on optimizing intra-node co
 mmunication. Existing designs in MPI libraries that work on the concept of
  distributed address spaces incur the overhead of intermediate memory copi
 es to stage the data between processes. This can lead to seve...\n\n------
 ---------------\nModeling Single-Source Shortest Path Algorithm Dynamics t
 o Control Performance and Power Tradeoffs\n\nKaramati, Young, Vuduc\n\nThi
 s work presents a new methodology to improve the performance of parallel a
 lgorithms by tuning the amount of available parallelism for execution thro
 ughout the runtime. As such, we expose key parameters controlling the perf
 ormance and parallelism of the algorithm and build a software-based contro
 l...\n\n---------------------\nEulerian Algorithms for the Discretization 
 of Plasma Kinetic Equations\n\nJuno\n\nWhile fluid models are common tools
  in the study of plasmas, many of these systems, whether in astrophysics o
 r the lab, are only weakly collisional and far from equilibrium, making th
 em more accurately described by kinetic equations. Kinetic equations can b
 e computationally demanding due to the need...\n
END:VEVENT
END:VCALENDAR