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:20230124T170828Z
LOCATION:C146
DTSTART;TZID=America/Chicago:20221113T083000
DTEND;TZID=America/Chicago:20221113T170000
UID:submissions.supercomputing.org_SC22_sess424@linklings.com
SUMMARY:Workshop on Latest Advances in Scalable Algorithms for Large-Scale
Heterogeneous Systems (ScalAH'22)
DESCRIPTION:Workshop\n\nNovel hybrid scalable scientific algorithms are ne
eded with the advent of variety of novel accelerators including GPUs, FPGA
s as well as with the growth of the size of the Quantum Computing Devices
and neuromorphic chips and various AI specific processors. This myriad of
devices requires a unified approach that allows efficient and scalable hyb
rid approaches combining classical and novel computing paradigms to be imp
lemented at scale. These extreme-scale heterogeneous systems require these
novel scientific algorithms to hide the complexity, hide network and memo
ry latency, have advanced communication, and have no synchronization point
s where possible. With the advent of AI in the past few years, the need of
such scalable mathematical methods and algorithms for such hybrid archite
ctures that are able to handle data and compute intensive applications at
scale becomes even more important.\n\nWorkshop Website\n\nThreshold Pivoti
ng for Dense LU Factorization\n\nLindquist, Gates, Luszczek, Dongarra\n\nL
U factorization is a key approach for solving large, dense systems of line
ar equations. Partial row pivoting is commonly used to ensure numerical st
ability; however, the data movement needed for the row interchanges can re
duce performance. To improve this, we propose using threshold pivoting to
fin...\n\n---------------------\nInvited Talk: Scalable Deep Learning Algo
rithms for Scientific Applications on Leadership Class Computing Systems\n
\nVan Essen\n\nIn this talk, we will discuss the challenges and opportunit
ies for implementing deep learning algorithms at scale for scientific appl
ications on leadership class HPC systems. Using examples drawn from multi
ple application areas we will see how challenges created by algorithmic co
mplexity as well as...\n\n---------------------\nScalAH – Afternoon Break\
n\n\n\n---------------------\nMARs: Memory Access Rearrangements in Open M
PI\n\nLi, Schuchart, Bosilca\n\nThe datatype engine in Message Passing Int
erface (MPI) libraries supports the communication layer by handling the tr
ansfer of non-contiguous datatypes. Basic datatypes (integer, float, etc.)
serve as building blocks for more complex, and potentially non-contiguous
, derived datatypes. In this context...\n\n---------------------\nInvited
Talk: Opportunities for Neuromorphic Computing Co-Processors\n\nSchuman\n\
nNeuromorphic computing is a popular technology for the future of computin
g. Much of the focus in neuromorphic computing research and development h
as focused on new architectures, devices, and materials, rather than in th
e software, algorithms, and applications of these systems. In this talk,
I wil...\n\n---------------------\nGPU Optimization of Lattice Boltzmann M
ethod with Local Ensemble Transform Kalman Filter\n\nHasegawa, Imamura, In
a, Onodera, Asahi...\n\nThe ensemble data assimilation of computational fl
uid dynamics simulations based on the lattice Boltzmann method (LBM) and t
he local ensemble transform Kalman filter (LETKF) is implemented and optim
ized on a GPU supercomputer based on NVIDIA A100 GPUs. To connect the LBM
and LETKF parts, data transp...\n\n---------------------\nInvited Talk: Hy
brid AI/HPC Approaches for Next Generation Multi-Trillion-Parameter Models
\n\nBrown\n\nWe live in a world where large-scale systems for machine inte
lligence are increasingly being used to solve complex problems in scientif
ic research. A convergence of machine learning model adoption alongside cl
assical algorithms, purpose-built scale-out systems availability in the cl
oud and maturing ...\n\n---------------------\nMixed-Precision Algorithm f
or Finding Selected Eigenvalues and Eigenvectors of Symmetric and Hermitia
n Matrices\n\nTsai, Luszczek, Dongarra\n\nThe multi-precision methods comm
only follow approximate-iterate scheme by first obtaining the approximate
solution from a low-precision factorization and solve. Then, they iterativ
ely refine the solution to the desired accuracy that is often as high as w
hat is possible with traditional approaches. W...\n\n---------------------
\nInvited Talk: Heterogeneous Computing Challenges and Opportunities\n\nJo
rdan\n\nThis is a particularly disruptive time for the development of futu
re computing systems, and the next 10 years will see some very fundamental
shifts in how those systems are architected and deployed. With the emerge
nce of Heterogeneous Computing including AI and Quantum, we are seeing an
explosive gr...\n\n---------------------\nImplementing Asynchronous Jacobi
Iteration on GPUs\n\nTsai, Nayak, Chow, Anzt\n\nComputation on architectu
res that feature fine-grained parallelism requires algorithms that overcom
e load imbalance, inefficient memory accesses, serialization, and excessiv
e synchronization. In this paper, we explore an algorithm that completely
removes the need for synchronization but allows for a...\n\n--------------
-------\nInvited Talk: Hybrid AI/HPC Approaches and Linear Algebra\n\nEmad
\n\nWe present a brief overview of machine learning techniques and show th
at certain methods of linear algebra such as the eigenvalue problem or mor
e generally singular value decomposition constitute the foundations of the
se techniques. We consider some examples of applications by highlighting t
he essen...\n\n---------------------\nScalAH – Morning Break\n\n\n\n------
---------------\nScalAH – Lunch Break\n\n\n\n---------------------\nScalab
le Finite-Element Viscoelastic Crustal Deformation Analysis Accelerated wi
th Data-Driven Method\n\nFujita, Murakami, Ichimura, Hori, Hori...\n\nTarg
eting viscoelastic crustal deformation analysis, we develop a scalable uns
tructured implicit finite-element solver accelerated by a data-driven meth
od. Here, we combine a data-driven predictor, that uses past time step dat
a for estimating high-accuracy initial solutions, and a multi-grid based c
...\n\n\nSession Format: Recorded\n\nTag: Algorithms, Exascale Computing,
Extreme Scale Computing, Heterogeneous Systems, Post-Moore Computing, Quan
tum Computing\n\nRegistration Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR