ECP @ SIAM PP22

Virtual Conference | February 23–26, 2022

The SIAM Conference on Parallel Processing for Scientific Computing (SIAM-PP22) showcases a broad range of work by the ECP community:

  • Software technology: Building software technologies that power HPC discoveries, including scalable math libraries, open-source programming models and runtimes, data and visualization packages, development tools, and more.
  • Application development: Developing simulation capabilities for chemistry, materials, energy, Earth and space science, data analytics, optimization, AI, national security, and more.
  • Hardware and integration: Deploying an exascale software stack, including software integration and testing across leadership computing environments. Improving software productivity and sustainability to help advance scientific productivity.

ECP community members are contributing by co-organizing minisymposia, presenting talks and posters, and speaking in plenary panel sessions. Presentations include insights about new algorithmic approaches, computing and communication protocols, data management, and more.

 

Awards at SIAM-PP22:

 

Invited Sessions at SIAM-PP22:

 

Abbreviations:

 

ECP ProjectECP Participation

Wednesday, Feb 23
Software TechnologyECP co-organized event: MS2: Progress and Challenges in Extreme Scale Computing and Big Data
ECP presenting: A Software Ecosystem Approach to Increasing Progress and Addressing Challenges in Extreme Scale Computing and Big Data
IDEAS-Productivity, ADIOSECP co-organized event: MS5: Research Challenges and Opportunities within Software Productivity, Sustainability, and Reproducibility - Part I of II 
ECP presenting: Toward a Better Ecosystem for Understanding and Reusing Scientific Workflows,
Trust Tools: Experiences in Reproducibility in Undergraduate Research
ExaGraphECP co-organized event: MS6: Graph Algorithms in the Exascale Era - Part I of II
ECP presenting: Graph Algorithms for the Exascale: A Case Study with Influence Maximization,
Scalable Preconditioner for Graph Laplacian,
Load Balancing Parallel Fock Matrix Computation in Nwchemex using Submodular B-Matching
PETSc/TAO, CEEDECP co-organized event: MS11: Versatile Solvers on GPU-Based High-Performance Computing Architectures - Part I of II
ECP presenting: Landau Collision Operator in the Cuda Programming Model Applied to Thermal Quench Plasmas,
Leveraging Mixed Precision to Accelerate High-Order Finite Element Methods on Gpus,
On Portability and Performance Versatility in Nonlinear Solid and Fluid Mechanics Using libCEED and PETSc
CEEDECP presenting: Pumipic: Infrastructure for Unstructured Mesh Pic on GPUs
ArgoECP presenting: Power-Aware Resource Management
PETSc/TAO, CEEDECP co-organized event: MS13: Versatile Solvers on GPU-Based High-Performance Computing Architectures - Part II of II
ECP presenting: A New Performance-Complexity Landscape: JIT-based Fusion for Matrix-Free Multigrid Solvers, with a Case Study in Bonded Granular Media
Combustion-PeleECP presenting: Novel Solver Algorithms for Nearly Singular Linear Systems Arising in Combustion Modelling
ExaGraphECP co-organized event: MS19: Graph Algorithms in the Exascale Era - Part II of II
ECP presenting: Sparse Matrices and High-Performance Computing Meet Biology
IDEAS-Productivity, Software Ecosystem & Delivery SDK, CEEDECP co-organized event: MS21: Research Challenges and Opportunities within Software Productivity, Sustainability, and Reproducibility - Part II of II
ECP presenting: UX-Focused Reproducible Analysis Toolkit for Mars 2020,
Impacting Software Quality and Process Through the Extreme-Scale Scientific Software Stack (E4S) and Software Development Kit (SDK) Projects,
Productive Performance Portability: Building in Rust with PETSc and libCEED
ExaWindECP presenting: Preparing Trilinos Solvers for Exascale Wind Farm Simulations

ExaStar, AMReX

ECP presenting: Scaling AMReX Applications using a Bittree Framework
PROTEAS-TUNEECP presenting: Openmp Target Task: Tasking and Target Offloading for Heterogeneous Linear Algebra Implementations
PaRSECECP presenting: Performance Optimizations of Low-Rank Matrix Computations for 3D Unstructured Mesh Deformation
ExaIOECP presenting: A Runtime System for Orchestrating Data Movement in Heterogeneous Memory and Storage Hierarchies
PROTEAS-TUNEECP presenting: Analytical, Statistical, and Machine Learning Models for Compiler Optimization of Convolutional Neural Networks
PETSc/TAOECP presenting: PETSc and Kokkos-Kernels Integration
PaRSECECP presenting: Accelerating Space and Space-Time Statistical Modeling with Mixed-Precision Arithmetic
PaRSECECP presenting: Non-Blocking Resilient Extensions in MPI
PROTEAS-TUNEECP presenting: Machine-Learning-Based Search for Autotuning
STRUMPACK/SuperLUECP presenting: Sparse Factorization-Based Solvers and Preconditioners on Modern GPUs
ALExaECP presenting: We Needed a Fast Performance-Portable Geometric Search. Here's What We Learned Along the Way
PaRSECECP presenting: Interoperability of Resilience Strategies for Simplicity and Performance

Thursday, Feb 24
Training - ECP Broader Engagement InitiativeECP co-organized event: PD1: Current and Future Trends Impacting Inclusivity in High-Performance Scientific Computing
xSDKECP presenting: The Hazards and Challenges of Low and Mixed Precision Computation
CEEDECP presenting: Tuned Preconditioners for the Spectral Element Method on GPUs
xSDKECP presenting: Mixed Precision Solver Capabilities for the Hypre Linear Solver Library
xSDKECP presenting: Mixed Precision in Linear Algebra
OMPI-XECP presenting: Extending Reinit Towards a Formal Specification and Interplay with GPUs
ExaAM, CoPAECP presenting: Architecture and Performance of Hardware-Independent Frameworks for Particle-In-Cell Methods - Part III of III
SUNDIALS/hypre
ECP presenting: Porting Hypre to Heterogeneous Computer Architectures: Strategies and Experiences
Application Integration, SOLLVEECP presenting: Testing Strategies in a Large Scale Scientific Code – Experiences and Aspirations,
Testing, Verifying, and Validating Openmp Implementations
xSDKECP presenting: Mixed-Precision Algorithm for Approximating Selected Eigenvalues and Eigenvectors of Symmetric and Hermitian Matrices
ExaGraphECP presenting: Graph Analytics in the Exascale Era
xSDKECP presenting: Mixed Precision Strategies for GMRES
WarpXECP presenting: WarpX Progress

CoPA, WDMApp

ECP presenting: Coupling Particle-in-Cell and Continuum Methods in the Global Total-f Gyrokinetic Magnetic Fusion Code XGC
CLOVER (SLATE)ECP co-organized event: MS32: Towards Exascale Eigenvalue Algorithms for Physical Simulation - Part I of II
ExaSGDECP presenting: Extreme-Scale Grid Optimization with the Exascale Grid Optimization (ExaGO) Toolkit
ExaSGDECP presenting: A Scalable Decomposition for Solving ACOPF with GPU Batch Solver
ExaSGDECP presenting: Towards Real-Time ACOPF on Emerging Architectures
xSDKECP co-organized event: MS29: Understanding and Exploiting Mixed-Precision Accelerators for High-Performance Computing - Part I of II
STRUMPACK/SuperLUECP co-organized event: CP4: Low Rank Techniques
ECP presenting: Scalable GPU H2-Matrix Solvers,
Multifrontal Solver with Low-Rank Compression
xSDKECP presenting: Lossy Compression and Mixed Precision Strategies for Memory-Bound Linear Algebra
ExaStar, ProductivityECP co-organized event: IP3: Software as an Instrument of Science
ALExaECP presenting: HDBSCAN* on GPUs: Can We Do It Better?
SUNDIALS/hypre
ECP presenting: Sundials: Suite of Nonlinear and Differential/Algebraic Equation Solvers

Friday, Feb 25
STRUMPACK/SuperLUECP co-organized event: SP1 SIAG Best Paper Prize: Optimizing Communication-Avoiding Sparse Lu Factorization on Multi-Gpu Clusters

AMREX, WarpX

ECP presenting: An Overview of Particles in Amrex
IDEAS-Productivity, Software Packaging Technologies, ExaStar ECP co-organized event: PD3: Build, Integration and Testing for Sustainable Scientific Computing Software
AMReXECP presenting: Advances in GPU Methodologies in AMReX
AMReXECP presenting: Parallel Simulation of Cancer Cell Migration
xSDKECP co-organized event: MS45: Understanding and Exploiting Mixed-Precision Accelerators for High-Performance Computing - Part II of II
ExaSGDECP presenting: A Distributed-Memory Optimization Solver for Two-Stage Stochastic Programming Problems
ExaSGDECP presenting: A Hybrid Direct-Iterative Method for Solving Kkt Linear Systems
ExaSGDECP presenting: An Inertia-Free Approach for Nonlinear Optimization on GPU
CEEDECP presenting: High-Order Matrix-Free GPU-Accelerated Linear Solvers with Applications to Fluid Flow
SLATE, CEEDECP presenting: Magma: Towards High Performance and Cross Platform Algorithms for Numerical Linear Algebra
CEEDECP presenting: Fast High Order GPU Simulation Support for Exascale Applications
SUNDIALS/hypre
ECP presenting: GPU Support in SUNDIALS Time Integrators and Nonlinear Solvers
MFIX-ExaECP presenting: CFD-Dem Simulations of Reacting Fluidized Beds with MFIX-Exa
ExagraphECP presenting: Sphynx: The First Graph Partitioner on AMD GPUs
PETSc/TAOECP presenting: Toward Performance-Portable PETSc for GPU-Based Exascale Systems

Saturday, Feb 26
PROTEAS-TUNEECP co-organized event: MS70: Targeting Future Exascale and Extreme Heterogeneity Era - Part I of II
SUNDIALS/hypre
ECP presenting: Accelerating Hypre with GPUs: Strategies and Performance
PROTEAS-TUNEECP co-organized event: MS85: Targeting Future Exascale and Extreme Heterogeneity Era - Part II of II
STRUMPACK/SuperLU, PETSc/TAOECP co-organized event: MS68: Experiences in Developing GPU Support for DOE Math Libraries - Part III of III
ECP presenting: Experiences with Quasi-Newton Methods for Numerical Optimization on GPUs,
Multi-GPU Implementation of Sparse Triangular Solve and Its Performance Analysis
CODARECP presenting: Scientific Data Refactoring and Compression on the GPU with MGARD
PaRSECECP presenting: Template Task Graph (TTG): Flow Graph Programming for Fine-Grained Distributed Memory Parallelism
National Nuclear Security Administration logo U.S. Department of Energy Office of Science logo