BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20191206T142451Z
LOCATION:Analog 1\, 2
DTSTART;TZID=Europe/Stockholm:20180625T131300
DTEND;TZID=Europe/Stockholm:20180625T131700
UID:isc_hpc_ISC High Performance 2018_sess103_phd127@linklings.com
SUMMARY:(PhD03) Accelerators in a Hybrid HPC World: How Can Applications B
 enefit?
DESCRIPTION:PhD Forum\n\n(PhD03) Accelerators in a Hybrid HPC World: How C
 an Applications Benefit?\n\nMaffia\n\nFor quite some time, so-called accel
 erators are considered to be the core components of a successful high-perf
 ormance computing systems. While the power of GPU integration paved the wa
 y, today’s approaches based on FPGA, ASIC, and even Quantum Computing tech
 nology become more and more critical. The result is a powerful but complex
  and hybrid HPC world. As a consequence, programming such systems is THE c
 hallenge. The de-facto HPC programming standards (OpenMP and MPI) that com
 puter scientists still use are not appropriate for researchers coming from
  a different field. Life science researchers will prefer high-level suppor
 t such as given by Python and R environments, while in the industry we can
  still find demands for application-specific Java interfaces. Our approach
  bridges user needs from all three communities by providing tailor-made in
 terface layers that share commonly needed system components.\n\nResearch o
 bjectives can be summarized as follows:\n\n- Allow scientists to exploit t
 he real computational capabilities of an HPC System in a transparent manne
 r hiding the system/environment configuration complexity.\n\n- Empower use
 rs from different communities to access HPC resources without changing the
  way they carry on their experiments.\n\n- Achieve reproducibility proposi
 ng a systematic approach to the experiments.\n\n- Using a modular architec
 ture for better interfacing the diverse systems and technologies currently
  used and reduce the effort needed for including emerging ones.\n\nThe pro
 posed approach is based on designing and implementing an architecture comp
 osed of three main parts: An “interface” for defining an experiment and se
 nding the execution request to a “web server” which then establishes and m
 anages the communication with the “remote HPC system” that will run the ex
 periment. Re-using the components provided by PROVA! (prova.io) for managi
 ng HTTP requests from the interface (PROVA! Web Server) and SSH connection
 s to the remote system (PROVA! Back-End), it is possible to implement diff
 erent interfaces creating more execution workflows.\n\nIn order to evaluat
 e the Ph.D. work, a set of experiments/applications from both academics an
 d industry has been selected. The prototype system has already been used f
 or running a stencil compilers benchmark: using PROVA!, the experiment was
  reproduced on remote HPC systems from different universities comparing th
 e results obtained. Other applications from various fields such as Life Sc
 ience (Deep Neural Networks for Anomaly Detection in Lung Imaging), Financ
 e (Supervised machine learning for American Option Pricing), Business Mana
 gement (Stochastic Optimization for Supply Chain Management) will be accel
 erated and executed through the system for evaluating the other execution 
 workflows.\n\nThe results currently achieved are:\n\n- Allowing experiment
  execution and reproduction on different remote systems: it is possible to
  choose between a solution based on a software build and installation fram
 ework (EasyBuild) or one based on containers (Docker/Singularity).\n\n- Im
 plementation of a client/server extension for Jupyter Notebook interfacing
  with the PROVA! Web Server and allowing the remote execution of Jupyter N
 otebook cells.\n\nAt the end of the work, a quantitative and qualitative a
 nalysis of system performances and user acceptance will be needed.\n\nPass
 es: Conference Pass, Exhibition Pass, Containerized HPC, HPC Accelerators,
  HPC workflows, Reproducibility\n\nTag: Conference Pass, Exhibition Pass, 
 Containerized HPC, HPC Accelerators, HPC workflows, Reproducibility
URL:https://2018.isc-program.com/presentation/?id=phd127&sess=sess103
END:VEVENT
END:VCALENDAR