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:20211207T055404Z
LOCATION:Online
DTSTART;TZID=America/Chicago:20211114T090000
DTEND;TZID=America/Chicago:20211114T173000
UID:submissions.supercomputing.org_SC21_sess430@linklings.com
SUMMARY:ExaMPI: Workshop on Exascale MPI
DESCRIPTION:Workshop\n\nExaMPI:  Lunch Break (12:30-2)\n\n\n\n------------
 ---------\nExaMPI: Workshop on Exascale MPI\n\nGrant, Bangalore\n\nThe aim
  of this workshop is to bring together researchers and developers to prese
 nt and discuss innovative algorithms and concepts in the message passing p
 rogramming model and to create a forum for open and potentially controvers
 ial discussions on the future of MPI in the exascale era.\n\n-------------
 --------\nPartitioned Collective Communication\n\nHolmes, Skjellum, Jaeger
 , Grant, Bangalore...\n\nPartitioned point-to-point communication and pers
 istent collective communication were recently standardized in MPI-4.0. The
 y both offer performance and scalability advantages over MPI-3.1-based com
 munication when planned transfers are possible in an MPI application. Part
 itioned collectives will be p...\n\n---------------------\nExaMPI:  Aftern
 oon Break (3-3:30)\n\n\n\n---------------------\nLeveraging Interconnect Q
 oS Capabilities for Congestion-Aware MPI Communication\n\nKhalilov, Slinka
 , Zhang\n\nResource sharing between job allocations on production supercom
 puters often leads to traffic interference between MPI applications. Netwo
 rk congestion, a side effect of such resource sharing, results in substant
 ial MPI latency degradation, making application performance unpredictable.
  We consider th...\n\n---------------------\nToward Modern C++ Language Su
 pport for MPI\n\nGhosh, Alsobrooks, Ruefenacht, Skjellum, Bangalore...\n\n
 The C++ programming language has made significant inroads in improving per
 formance and productivity across a broad spectrum of applications and hard
 ware. The C++ language bindings to MPI had been deleted since MPI 3.0 (cir
 ca 2009) due to the rationale that it added minimal functionality over the
  exi...\n\n---------------------\nOverlapping Communication and Computatio
 n with ExaMPI's Strong Progress and Modern C++ Design\n\nSchafer, Hines, S
 uggs, Rüfenacht, Skjellum\n\nExaMPI is a modern, C++17+ MPI implementation
  designed for modularity, extensibility, and understandability. In this wo
 rk, we overview  functionality new to ExaMPI since its initial release, in
 cluding Libfabric-based network transport support. We also explain our rat
 ionale for why and how we choose ...\n\n---------------------\nExaMPI Comm
 unity Discussion\n\nDosanjh, Grant\n\n---------------------\nExaMPI Mornin
 g Break (10-10:30)\n\n\n\n---------------------\nExaMPI Invited Talk: Desi
 gning High-Performance and Scalable Middleware for HPC and AI: Challenges 
 and Opportunities\n\nPanda\n\nThis talk will focus on challenges and oppor
 tunities in designing middleware for HPC, AI (Deep/Machine Learning), and 
 Data Science for On-premise HPC and Cloud systems with advances in network
 ing and accelerator technologies. For the HPC domain, we will discuss abou
 t the challenges in designing runt...\n\n---------------------\nA Benchmar
 k to Understand Communication Performance in Hybrid MPI and GPU Applicatio
 ns\n\nHaskins, Bridges, Levy, Ferreira\n\nAnalyzing MPI communication cost
 s on extreme-scale HPC systems is critical to ensuring optimal performance
 . Factors including scalability and widespread use of GPUs complicate this
  analysis.  To address this challenge, we need benchmarks and tools that u
 se GPU and host memory in a manner similar to ...\n\n---------------------
 \nExaMPI Panel\n\nSkjellum\n\n---------------------\nA FACT-Based Approach
 : Making ML Collective Autotuning Feasible on Exascale Systems\n\nWilkins,
  Guo, Thakur, Hardavellas, Dinda...\n\nMachine learning (ML) autotuners us
 e supervised learning to select MPI collective algorithms, significantly i
 mproving collective performance. However, a user may find it difficult to 
 understand the benefit of autotuners because we lack a methodology to quan
 tify their performance. Additionally, to ob...\n\n---------------------\nA
 ccelerating Multi-Process Communication for Parallel 3-D FFT\n\nAyala, Tom
 ov, Stoyanov, Haidar, Dongarra\n\nToday's largest and most powerful superc
 omputers in the world are built on heterogeneous platforms; and using the 
 combined power of multi-core CPUs and GPUs, has had a great impact acceler
 ating large-scale applications. However, on these architectures, parallel 
 algorithms, such as the Fast Fourier T...\n\n\nTag: Online Only, Extreme S
 cale Computing, Parallel Programming Languages and Models, Performance\n\n
 Registration Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR