libCEED 0.11.0 Released
December 24, 2022 (release)
Developed as part of the Center for Efficient Exascale Discretizations (CEED) within the U.S. Department of Energy, libCEED is a high-order API library that provides a common algebraic low-level operator description, allowing a wide variety of applications to take advantage of the efficient operator evaluation algorithms in the different CEED packages. This release includes improved support for H(div) bases, a new compressible Navier-Stokes mini-app, improvements to the interface, and more.
Learn more:
Conduit 0.8.5 Released
December 22, 2022 (release)
Conduit provides an intuitive model for describing hierarchical scientific data in C++, C, Fortran, and Python. It is used for data coupling between packages in-core, serialization, and I/O tasks. The latest release adds new methods, support for Wedges and Pyramids, optional support for Caliper performance annotations, and more.
Learn more:
Johannes Doerfert Wins Fellowship from Better Scientific Software Organization
December 22, 2022 (profile) (story)
Johannes Doerfert was selected as one of six 2023 Better Scientific Software fellows recognizing his leadership and advocacy of high-quality scientific software. The Better Scientific Software (BSSw) community is an international group of researchers, practitioners, and stakeholders from national laboratories, academic institutions, and industry who are dedicated to curating, creating, and disseminating information that leads to improved software for the advancement of computational science and engineering (CSE) and related technical computing areas, with a particular interest in CSE on high-performance (parallel) computers. The BSSw Fellowship Program gives recognition and funding to leaders and advocates of high-quality scientific software. Each 2023 Fellow receives up to $25,000 for an activity that promotes better scientific software. Doerfert will focus his fellowship funding on improving developer productivity by demystifying the compiler black box. His plans include making short introductory videos on compiler technology with a focus on improved interaction and available tooling. The new fellows will be recognized at the 2023 Exascale Computing Project Annual Meeting in January.
Adiak 0.2.2 Released
December 21, 2022 (release)
Adiak is a library for collecting metadata from HPC application runs, and distributing that metadata to subscriber tools. Adiak has a tool interface, which allows tools to subscribe to this metadata. Among the changes in version 0.2.2, launchdate now works on OSX.
Learn more:
hypre 2.27.0 Released
December 20, 2022 (release)
hypre is a library of high-performance preconditioners and solvers featuring multigrid methods for the solution of large, sparse linear systems of equations on massively parallel computers. Version 2.27.0 includes support for multi-component vectors, HIP support added to the MGR solver, optimizations for IJ solvers, CMake build system updates, and more.
Learn more:
OpenZFS: Improving File System Efficiency
December 15, 2022 (story)
Large-scale parallel computing systems generate massive amounts of data every second, and file systems are crucial for handling data transfer at LLNL’s HPC center. Livermore Computing’s (LC’s) Scalable Storage group helps manage the hardware and software necessary to keep file systems—and therefore supercomputers—running smoothly and efficiently. One key tool in the group’s portfolio is the ZFS (Zettabyte File System) project, which controls I/O operations and optimizes storage volume capacity. In 2013, the original ZFS project and ZFS on Linux evolved into what is now known as OpenZFS, which is maintained by a global developer community that includes LC staff. The Scalable Storage group has adapted OpenZFS for the Lab’s needs, especially as new generations of Lustre-based HPC systems—including the upcoming El Capitan exascale supercomputer—are designed and installed. Read more about OpenZFS on the LLNL Computing website.
Merlin 1.9.0 Released
December 15, 2022 (release)
Merlin is a tool for running machine learning based workflows. The goal of Merlin is to make it easy to build, run, and process the kinds of large scale HPC workflows needed for cognitive simulation. Version 1.9.0 includes support for Python 3.11, JSON schema validation for Merlin spec files, updated documentation, and more.
Learn more:
Automated Cache for Container Executables
December 15, 2022 (story)
LLNL’s Vanessa Sochat and collaborators from the Wellcome Sanger Institute, the Pawsey Supercomputing Research Institute, and the University of Texas at Dallas have written a paper about the Singularity Registry HPC (“shpc”). Sochat breaks down the team’s methods in a Twitter thread. You can also download the preprint PDF; abstract follows:
Linux container technologies such as Docker and Singularity offer encapsulated environments for easy execution of software. In high performance computing, this is especially important for evolving and complex software stacks with conflicting dependencies that must co-exist. Singularity Registry HPC (“shpc”) was created as an effort to install containers in this environment as modules, seamlessly allowing for typically hidden executables inside containers to be presented to the user as commands, and as such significantly simplifying the user experience. A remaining challenge, however, is deriving the list of important executables in the container. In this work, we present new automation and methods that allow for not only discovering new containers in large community sets, but also deriving container entries with important executables. With this work we have added over 8,000 containers from the BioContainers community that can be maintained and updated by the software automation over time. All software is publicly available on the GitHub platform, and can be beneficial to container registries and infrastructure providers for automatically generating container modules to lower the usage entry barrier and improve user experience.
YGM 0.5 Released
December 08, 2022 (release)
YGM is a general-purpose, pseudo-asynchronous communication library built on top of MPI in C++. The repo’s utility is provided through its mailbox abstractions, which are used for point-to-point and broadcast communications. Release v0.5 includes a parser for Parquet files, improved broadcasts, and new containers.
Learn more:
Best Paper Winner Improves Scientific Workflow Performance
December 07, 2022 (story)
The IEEE international eScience conference, which emphasizes compute- and data-intensive research methods, bestowed the 2022 Best Paper Award on a multidisciplinary team that includes LLNL staff and external collaborators. The paper, “Scalable Composition and Analysis Techniques for Massive Scientific Workflows,” details the optimization of a drug screening workflow for the American Heart Association (AHA). The AHA Molecule Screening (MoleS) workflow combines specialized software tools to manage HPC hardware heterogeneity. The MoleS end-to-end workflow relies on both general-purpose and domain-specific software tools, some of which are open source and/or developed at LLNL: Maestro for workflow execution, Flux for workload management and scheduling, RabbitMQ for message brokering (orchestrated by Kubernetes), ConveyorLC for docking automation tasks, Fusion machine learning algorithms for binding affinity predictions, and GMD (Generative Molecular Design) for the small-molecule discovery loop. Read more on the LLNL Computing website.
Carol Woodward Helps Scientists Solve Diverse Challenges
December 01, 2022 (profile)
Carol Woodward joined the Lab’s Center for Applied Scientific Computing (CASC) in 1996, first as a postdoc and then as a staff researcher. CASC has developed a reputation over the years, she notes, as an organization that can solve tough problems, so she and her colleagues are asked to consult on a diverse array of projects. “It’s nice because it means I can work at the same place and not just do one thing for a long time—I get to keep changing what I work on,” she says. She is the principal investigator for SUNDIALS, a package of time integrators and nonlinear solvers that garners more than 100,000 downloads annually and is used in myriad simulation-dependent applications. Her group’s technical contributions to the software have modernized it and upgraded its functionality for more than two decades, enabling it to scale to DOE’s highest-end computing systems. With its innovative solvers and flexibility for a variety of computing systems, SUNDIALS was awarded the prestigious 2023 SIAM/ACM Prize in Computational Science and Engineering. In 2022 Woodward was promoted to Distinguished Member of Technical Staff, LLNL’s highest technical job classification level. Read more about her work.
Caliper 2.9.0 Released
November 30, 2022 (release)
Caliper is a program instrumentation and performance measurement framework. It is designed as a performance analysis toolbox in a library, allowing one to bake performance analysis capabilities directly into applications and activate them at runtime. The latest release includes improved sample profiling support and event tracing support, options for filtering regions by name, and more.
Learn more:
CASC Newsletter Highlights Open Source Projects
November 21, 2022 (story)
LLNL’s Center for Applied Scientific Computing (CASC) has published a new issue of its external newsletter. Included in the issue articles featuring open source software projects:
- A New Tradition: MFEM Community Workshop: More than 150 researchers from dozens of organizations and countries attended each of the team’s one-day virtual workshops.
- Compiler Co-Design with the RAJA Team: The RAJA Performance Suite, developed during the Sierra platform procurement, is a key tool for the Lab’s interactions with compiler vendors and GPU vendors, such as NVIDIA and AMD, for current LLNL supercomputers.
Charliecloud 0.30 Released
November 18, 2022 (release)
LANL led with LLNL contributors, Charliecloud provides user-defined software stacks for HPC centers. It uses Linux user namespaces to run containers with no privileged operations or daemons and minimal configuration changes on center resources. Version 0.30 includes build cache performance improvements and many changes to ch-image.
Learn more:
HPCwire Award for Applying Cognitive Simulation to Inertial Confinement Fusion
November 17, 2022 (story)
The high-performance computing publication HPCwire announced LLNL as the winner of its Editor’s Choice award for Best Use of HPC in Energy for applying cognitive simulation (CogSim) methods to inertial confinement fusion (ICF) research. The award recognizes the team for progress in their machine learning-based approach to modeling ICF experiments performed at the National Ignition Facility (NIF) and elsewhere, which has led to the creation of faster and more accurate models of ICF implosions. The CogSim work addresses the need for better models that can fully utilize available datasets, can accurately estimate uncertainty, and can improve with additional data. Much of the CogSim work has been done on HPC machines including Sierra, Lassen, and Corona, using the open source projects Merlin, a custom deep-learning workflow tool, and the Livermore Big Artificial Neural Network toolkit (LBANN), a deep-learning training framework optimized for HPC.
Recap of MFEM's Second Community Workshop
November 16, 2022 (event-report) (multimedia)
The MFEM team hosted the second annual MFEM Community Workshop on October 25, 2025. The goal of the workshop was to foster collaboration among all MFEM users and developers, share the latest MFEM features with the broader community, deepen application engagements, and solicit feedback to guide future development directions for the project. If you missed the workshop, check out these resources: 2022 agenda (with speakers’ slides linked as PDFs) and news article. Videos of the talks are in production and will be posted soon.
SC22 Twitter Space: Open Source for HPC
November 15, 2022 (event-report) (multimedia)
The @Livermore_Comp account hosted a Twitter Space during the 34th annual Supercomputing Conference (SC22) on November 15. LLNL’s Meg Epperly moderated a panel featuring Elsa Gonsiorowski (SCR and other projects), Greg Becker (Spack), and David Gardner (SUNDIALS). The panel discussed answered questions about why open source software is important for HPC centers, who benefits from using and contributing to it, and why LLNL actively develops and nurtures open source software. Listen to the recording.
Spack 0.19.0 Released
November 11, 2022 (release)
Spack is a flexible, configurable, Python-based, and open-source HPC package manager. Spack automates the installation and fine-tuning of simulations and libraries, operating on a wide variety of HPC platforms and enabling users to build many code configurations. Version 0.19.0 includes:
- environment UI improvements
- installation logic for different build systems
- enable/disable inheritance for both variants and compiler flags
- ability to use git commits as versions
- improvements to Cray EX programming
- and much more
Learn more:
Optimizing Workflow with Flux
November 10, 2022 (story)
The latest issue of LLNL’s Science & Technology Review magazine showcases the R&D 100 award–winning Flux software framework. Honored with a 2021 R&D 100 Award, Flux is a scalable, flexible next-generation workload management framework that meets this need—maximizing resource utilization while allowing scientific applications and workflows to run faster and more efficiently. Developed in collaboration with university partners, Flux also enables new resource types, schedulers, and services to be deployed at data centers as they continue to evolve.
Spack: Sustaining the HPC Software Ecosystem
November 09, 2022 (event-report) (multimedia)
Todd Gamblin, an LLNL Distinguished Member of Technical Staff, gave a presentation on November 9 for the Dell Technologies HPC Community. His talk, “Sustaining the HPC Software Ecosystem,” described how HPC software can be managed more easily for all customers and users with Spack and included an overview of recent developments in the Spack community such as a partnership with AWS to provide infrastructure for a worldwide binary cache, a recent machine learning special interest group within Spack, and work to handle the complexities of installing software for GPUs. Slides can be downloaded, and a recording is available with free registration.
SUNDIALS Wins 2023 SIAM/ACM Prize in Computational Science and Engineering
November 07, 2022 (story)
The Society for Industrial and Applied Mathematics (SIAM) and Association for Computing Machinery (ACM) announced that they have awarded the 2023 SIAM/ACM Prize in Computational Science and Engineering to the team behind the LLNL-developed SUNDIALS software suite. The prestigious award is handed out every two years and recognizes outstanding contributions to the development and use of mathematical and computational tools and methods for the solution of science and engineering problems. It is one of SIAM’s most significant awards and will be presented to the team at the 2023 SIAM Conference on Computational Science and Engineering in Amsterdam next February. Because of its ease, flexibility and extensive documentation, SUNDIALS has become internationally recognized as one of the most effective and efficient time integration libraries. It is widely used by government laboratories and in academic institutions and industry, leading to advances in a variety of applications including fusion device modeling, watershed modeling, and reacting flow simulations.
Team members are Distinguished Member of the Technical Staff Carol Woodward, computational scientists Cody Balos and David Gardner, senior scientist Peter Brown, LLNL guest scholar and retiree Alan Hindmarsh, former LLNL postdoc Daniel Reynolds, and former LLNL scientist Radu Serban. Read more about the award at LLNL News and the SUNDIALS: SUite of Nonlinear and DIfferential/ALgebraic Equation Solvers at LLNL Computing.
Flux 0.45.0 Released
November 01, 2022 (release)
Flux is a flexible framework for resource management consisting of a suite of projects, tools, and libraries which may be used to build site-custom resource managers for HPC centers. This version includes new functionality for job-list, broker, and other functions, along with updates to the test suite and CI.
Learn more:
RAJA 2022.10.0 Released
October 28, 2022 (release)
RAJA is a software abstraction framework that systematically encapsulates platform-specific code to enable applications to be portable across diverse hardware architectures without major source code disruption. This release includes:
- new
RAJA::forall and reduction interface
- support for run time execution policy selection for
RAJA::forall kernels
- special case implementations to CUDA
atomicInc and atomicDec functions
- user guide updates
- and more
Learn more:
Umpire 2022.10.0 Released
October 27, 2022 (release)
Umpire is a resource management library that allows the discovery, provision, and management of memory on next-generation architectures. This release adds new HIP Advise operations, methods, event tracking, and UMAP allocation resource.
Learn more:
MFEM 4.5 Released
October 24, 2022 (release)
MFEM is a lightweight, general, scalable C++ library for finite element methods. It enables high-performance scalable finite element discretization research and application development on a wide variety of platforms, from laptops to supercomputers. The v4.5 release includes:
- meshing improvements including submesh extraction
- container + cloud support (see the new repo for Docker containers)
- GPU kernels for LOR (low order refined), DG (discontinuous Galerkin), and linear forms
- new and updated examples and miniapps
- Windows 2022 CI testing with GitHub actions
- and much more
Learn more:
SUNDIALS 6.4.0 Released
October 21, 2022 (release)
SUNDIALS is a SUite of Nonlinear and DIfferential/ALgebraic equation Solvers. This minor release adds:
- support for GPU enabled SuperLU_DIST and SuperLU_DIST v8.x.x
- support for the Ginkgo linear algebra library
- performance portability improvements
- new functions
- and more
Learn more:
Todd Gamblin Enables Developers
October 19, 2022 (profile)
As the creator of Spack, Todd Gamblin states, “Open source tools developed at LLNL enable people around the world to use HPC resources more effectively and to do better science.” In turn, these tools provide a framework for an entire community to maintain software needed by LLNL and its programs—for instance, LLNL could never maintain Spack’s thousands of software packages alone, and it benefits from the work of Spack’s enthusiastic contributors. Ultimately, he notes, “Sustaining open source communities is about finding leverage. It’s worthwhile for LLNL to put in the effort to build and maintain something like Spack if it incites a community of thousands to work together for the benefit of all.” In 2022 Gamblin was promoted to Distinguished Member of Technical Staff, LLNL’s highest technical job classification level. Read more about his recent work.
Celebrate Exascale Day 2022
October 18, 2022 (event) (multimedia)
The Exascale Computing Project (ECP) celebrates a new era of scientific discovery with Exascale Day on October 18, or “10^18” to represent the exascale threshold of floating-point operations per second. The event runs all week and provides multimedia and articles that educate explain the impact areas of exascale computing from the Department of Energy national laboratories (including LLNL), HPC manufacturers, and universities and industrial organizations. Much of the ECP’s software stack is open source.
New Repo: MDAS
October 14, 2022 (new-repo)
The MSU Disentanglement Analysis Software (MDAS) is used to disentangle the forced and unforced components of tropospheric temperature change over the satellite era (after 1979) using maps of surface temperature change as a predictor. Input data is available via Zenodo (doi: 10.5281/zenodo.7199961). An accompanying publication is forthcoming from the Proceedings of the National Academy of Sciences.
GIDIplus 3.25 Released
October 14, 2022 (release)
GIDIplus is a suite of C++ libraries for accessing nuclear data from the Generalized Nuclear Database Structure (GNDS). The latest version includes full GNDS-2.0 support, a new RISI module for reading .ris files, and more.
Learn more:
MuyGPyS 0.6.0 Released
October 12, 2022 (release)
MuyGPyS is a Gaussian process estimation method that affords fast hyperparameter optimization by way of performing leave-one-out cross-validation. The latest release introduces support for distributed memory processing using MPI and now supports three distinction implementations of all of the math functions.
Learn more:
New Project to Improve Differentiation of Extreme-Scale Science Applications
October 10, 2022 (event-report)
Under a recently funded project, researchers at LLNL and the Massachusetts Institute of Technology (MIT) will address the challenge of efficiently differentiating large-scale applications for the Department of Energy by building on advances in LLNL’s MFEM finite element library and MIT’s Enzyme AD tool. The team’s project will address the challenge of efficiently differentiating large-scale DOE applications—predicting how adjustments in design parameters will impact the output of a code. Knowledge of optimal outputs is increasingly needed for complex simulation codes to be used for design optimization, machine learning, uncertainty quantification and sensitivity analysis, among other applications. While automatic differentiation (AD) has made the differentiation process easier, traditional AD tools require significant changes that are not feasible for many existing large-scale DOE applications. Read more about the project at LLNL News.
New Repo: Rollerball
October 06, 2022 (new-repo)
The Rollerball Notepad++ Plugin repo contains code for a benignware (pseudo malware) plugin for the Notepad++ editor based on this plugin template. The plugin purports to be an AutoSave feature, but in fact it exfiltrates file content based on keywords to a webserver that you specify. This code was designed as a way to test detection capabilities.
Flux 0.44.0 Released
October 05, 2022 (release)
Flux is a flexible framework for resource management consisting of a suite of projects, tools, and libraries which may be used to build site-custom resource managers for HPC centers. This version includes enhancements to inject functionality, the queue state, flux-jobs, and more.
Learn more:
BLT 0.5.2 Released
October 05, 2022 (release)
BLT (Building, Linking, and Testing) is a streamlined CMake build system foundation for developing HPC software. BLT makes it easy to get up and running on a wide range of HPC compilers, operating systems, and technologies. This release includes a new macro and variable.
Learn more:
LLNL Heads to SC22
October 03, 2022 (event)
The 34th annual Supercomputing Conference (SC22) will be held in Dallas throughout November 13–18.
Twitter accounts to follow: @Livermore_Comp, @LLNL_OpenSource, @NatLabsHPC, @Livermore_Lab.
Check out LLNL’s full lineup of events. All initial broadcast times are listed in Central Standard Time.
HiOp 0.7.0 Released
October 01, 2022 (release)
HiOp is an optimization solver for solving certain mathematical optimization problems expressed as nonlinear programming problems. This lightweight HPC solver leverages application’s existing data parallelism to parallelize the optimization iterations by using specialized linear algebra kernels. This version includes a Fortran interface and examples, implementation of CUDA CSR matrices, and improved robustness and performance of the mixed dense-sparse solver.
Learn more:
VisIt 3.3.1 Released
September 30, 2022 (release)
VisIt is an open source, interactive, scalable, visualization, animation, and analysis tool. Version 3.3.1 includes:
- color table searching
- updates to x-ray image query
- updates to color table window
- and much more
Learn more:
ExaCA 1.1 Released
September 29, 2022 (release)
ExaCA is a cellular automata (CA) code for grain growth under additive manufacturing conditions by ExaAM within the Exascale Computing Project. This release adds options to include remelting and to control the density of grains.
Learn more:
New Repo: ATD
September 23, 2022 (new-repo)
The Assured Timing Detector (ATD) software provides an implementation in C++ of an adaptable, model-based system for monitoring a timing signal for anomalies versus a reference time source ensemble. This system is designed to be model-based, adaptive, and customizable. A companion status display software package, ATD-SD, is also available for use with this software.
Variorum 0.6.0 Released
September 15, 2022 (release)
Variorum is a platform-agnostic library exposing monitor and control interfaces for several features in hardware architectures. Its general interfaces provide privileged functionality for monitoring and controlling various hardware-level features of multiple hardware architectures. The latest release includes support for new architectures, new Python wrappers, API updates, and more.
Learn more:
New Repo: Skywing
September 09, 2022 (new-repo)
Skywing is a high-reliability, real-time, decentralized platform for collaborative autonomy–focused applications. The repo includes installation instructions, examples, and a tutorial; some dependencies are managed as git submodules.
Celebrating 10 Years of Hackathons
September 07, 2022 (event-report)
After 10 years and 33 hackathons, nothing can stop this beloved tradition. “Hackathons are wildly popular not just because they allow employees to try new things and develop new skills, but also because they are so much fun!” says Computing’s associate director Bruce Hendrickson. Read an article about this milestone, and check out the LLNL Flickr album of hackathon photos throughout the years.
Flux 0.43.0 Released
September 06, 2022 (release)
Flux is a flexible framework for resource management consisting of a suite of projects, tools, and libraries which may be used to build site-custom resource managers for HPC centers. This version includes new options to flux-jobs(1) and flux-mini(1), show detailed job state in flux-jobs(1) output, automatic KVS garbage collection, and more.
Learn more:
Camp 2022.03.2 Released
September 03, 2022 (release)
CAMP collects a variety of macros and metaprogramming facilities for C++ projects. This release catches CAMP up to the version nomenclature for the RAJA Portability Suite, which has changed format to indicate year, month, and patch number. New features include support for CTAD and structured bindings as well as fixes for exported target issues.
Learn more:
Dev Day Returns for the Sixth Year
September 02, 2022 (event-report)
Held in a hybrid format for the first time, LLNL’s Developer Day 2022 convened more than 70 people for an agenda of lightning talks, a town hall discussion, and guest speakers. Dev Day provides different ways for the audience to learn about and engage in topics of interest to the developer community—such as lightning talks about small projects, summaries of literature, deep dives into project planning and implementation, discussions of career opportunities and challenges, and networking and brainstorming sessions—and the format varies every year. This year’s event spotlighted guest speakers who explained the impact of unique software projects on their organizations. For instance, Miranda Mundt, a research software engineer at Sandia National Laboratories, described her team’s tiered approach to software quality practices, while Dr. Arun Viswanathan from NASA’s Jet Propulsion Laboratory presented work that keeps space missions resilient from cyber threats.
Sandia Leverages LLNL's Open Source Software for New Website
August 31, 2022 (meta) (story)
A Sandia software development team has launched a new website based on one of LLNL’s open-source projects. The new software.sandia.gov website (repository: github.com/sandialabs/sandialabs.github.io) serves as a portal into Sandia’s GitHub repositories, providing more than a dozen browsable categories alongside data visualizations of GitHub data—repository relationships, commit and pull request activity, common licenses, and more. Forked from this very website (repository: github.com/LLNL/llnl.github.io), Sandia’s site is also built on a Jekyll template and served via GitHub pages. Read more about the implementation process and the ways the two websites vary.
LLNL ATDM Addresses Software Infrastructure Needs for Multiple Communities
August 30, 2022 (story)
The Advanced Technology Development and Mitigation (ATDM) program within the Exascale Computing Project (ECP) shows that the best way to support the mission is through open collaboration and a sustainable software infrastructure. Although ATDM primarily supports NNSA’s traditionally closed mission of national security, LLNL’s ATDM Software Technology (ST) project contributes key open source components of a full-featured, integrated, and maintainable software stack for exascale systems that will impact both the ECP and the broader HPC community. A new article on the ECP website describes the project’s goals and technical focus areas in an interview with Livermore Computing division leader Becky Springmeyer and ATDM ST deputy lead Todd Gamblin.
Materials Available from RADIUSS AWS Tutorials
August 29, 2022 (event-report) (multimedia)
This summer, LLNL’s RADIUSS team conducted a series of tutorials in collaboration with Amazon Web Services (AWS), demonstrating how to use several GPU-ready projects in the cloud and on premises. The tutorials were open to everyone and held every week throughout August. Participants followed along on their own AWS EC2 instance (provided). No previous was experience necessary. Materials are accessible on the RADIUSS website.
UnifyFS 1.0 Released
August 25, 2022 (release)
UnifyFS is a user-level burst buffer file system under active development. The repo supports scalable and efficient aggregation of I/O bandwidth from burst buffers while having the same lifecycle as a batch-submitted job. This major release includes improved file staging, updated example programs, new unit tests and CI tests, a client API for I/O middleware libraries, and more.
Learn more:
ECP Annual Meeting Videos Now Available
August 24, 2022 (event-report) (multimedia)
The Exascale Computing Project, a joint effort between the DOE Office of Science and NNSA, brings together several national laboratories to address many hardware, software, and application challenges inherent in the organizations’ scientific and national security missions. The ECP’s annual meeting was held this year on May 2–6. Each day’s sessions are available in a dedicated YouTube playlist. Individual sessions highlighted below feature LLNL staff and open source projects.
Conduit 0.8.4 Released
August 23, 2022 (release)
Conduit provides an intuitive model for describing hierarchical scientific data in C++, C, Fortran, and Python. It is used for data coupling between packages in-core, serialization, and I/O tasks. The latest release adds several methods, updates to mesh blueprint functionality, and more.
Learn more:
Hatchet 2022.2.0 Released
August 19, 2022 (release)
Hatchet is a Python-based library that allows Pandas dataframes to be indexed by structured tree and graph data. This release resolves issues with package install, modifies graphframe copy/deepcopy, and more.
Learn more:
H5Z-ZFP 1.1.0 Released
August 18, 2022 (release)
H5Z-ZFP(https://github.com/LLNL/H5Z-ZFP) is a highly flexible floating point and integer compression plugin for the HDF5 library using zfp compression. This release includes updates to the decompression compatibility check, testing against previous zfp versions, CMake build, Windows build added to CI, and much more.
Learn more:
CEED's Sixth Annual Meeting Held in August
August 11, 2022 (event-report)
As part of the Exascale Computing Project (ECP), the Center for Efficient Exascale Discretizations (CEED) is a research partnership between two U.S. Department of Energy laboratories and five universities. LLNL leads the Center. All of CEED’s software is open source.
CEED held its sixth annual meeting (CEED6AM) on August 9-11 in a hybrid format: in-person at the Siebel Center for Computer Science on the University of Illinois Urbana-Champaign campus in Urbana and virtually using ECP Zoom for videoconferencing and Slack for side discussions. Learn more about the agenda on the CEED6AM event website.
SUNDIALS 6.3.0 Released
August 10, 2022 (release)
SUNDIALS is a SUite of Nonlinear and DIfferential/ALgebraic equation Solvers. This minor release:
- adds
GetUserData functions in each package
- updates
MRIStepReset to call the corresponding MRIStepInnerResetFn appropriately
- adds a new CVODE example
- adds embedded DIRK methods
- and more!
Learn more:
HPC Tech Shorts Features Greg Becker and Spack
August 08, 2022 (multimedia) (story)
Spack has already been removing the ugly work from building HPC codes, but with the announcement of the Spack Binary Cache—which is hosted on AWS—build and deploy times for these complicated applications will drop by 95% or more in most cases. Spack core developer Greg Becker spoke with HPC Tech Shorts to explain how it works and discuss what’s behind it. The video “Get your HPC codes installed and running in minutes using Spack’s Binary Cache” runs 15:11.
Charliecloud 0.29 Released
August 04, 2022 (release)
LANL led with LLNL contributors, Charliecloud provides user-defined software stacks for HPC centers. It uses Linux user namespaces to run containers with no privileged operations or daemons and minimal configuration changes on center resources. Version 0.29 includes fully unprivileged end-to-end workflows and updates to ch-image.
Learn more:
Kosh 2.2 Released
August 03, 2022 (release)
Kosh allows codes to store, query, and share data via an easy-to-use Python API. This software aims to make data access and sharing as simple as possible. The latest release includes added support for Windows systems and improvements to importing sina json files.
Learn more:
New Repo: GridDS
August 02, 2022 (new-repo)
As the number of smart meters and the demand for energy is expected to increase by 50% by 2050, so will the amount of data those smart meters produce. While energy standards have enabled large-scale data collection and storage, maximizing this data to mitigate costs and consumer demand has been an ongoing focus of energy research. GridDS is an open source data science toolkit for power and data engineers that will provide an integrated energy data storage and augmentation infrastructure, as well as a flexible and comprehensive set of state-of-the-art machine learning models. By providing an integrative software platform to train and validate machine learning models, GridDS will help improve the efficiency of distributed energy resources, such as smart meters, batteries and solar photovoltaic units. GridDS also is designed to leverage advanced metering infrastructure, outage management systems data, supervisory control data acquisition, and geographic information systems to forecast energy demands and detect incipient grid failures. Read more about GridDS at LLNL News.
GEOSX Simulates Carbon Dioxide Storage
August 02, 2022 (story)
GEOSX and its predecessor, GEOS, were developed from the ground up with the help of experts from across LLNL, combining a range of disciplines including engineering, seismology, hydrology, computational geoscience, and oil- and gas-industry expertise to build a tool that can take advantage of advanced computing platforms. A Research Highlight article in the latest issue of Science & Technology Review describes how GEOSX will improve the management and security of geological repositories and support planning for the widespread implementation of CO2 storage at an industrial scale by simulating how fluids flow and rocks break deep underground. Read “GEOSX Simulates Carbon Dioxide Storage” on the S&TR website.
zfp 1.0.0 Released
August 01, 2022 (release)
zfp is a library for compressed numerical arrays that support high-throughput read and write random access. zfp also supports streaming compression of integer and floating-point data, e.g., for applications that read and write large data sets to and from disk. zfp is primarily written in C and C++ but also includes Python and Fortran bindings. This major release includes numerous changes to function signatures and data structures, a more complete API for pointers and iterators, rounding modes for reducing bias in compression errors, support for pointers and iterators into array views, refactored compressed-array C++ implementation, and much more.
Learn more:
New Repo: PDBspheres
July 29, 2022 (new-repo)
PDBspheres is a structure-based method for finding and evaluating structural similarities in protein regions relevant to ligand binding. PDBspheres comprises an exhaustive library of protein structure regions (“spheres”) adjacent to complexed ligands derived from the Protein Data Bank (PDB), along with methods to find and evaluate structural matches between a protein of interest and spheres in the library.
ExaCA 1.0 Released
July 26, 2022 (release)
ExaCA is a cellular automata (CA) code for grain growth under additive manufacturing conditions by ExaAM within the Exascale Computing Project. ExaCA runs with the default enabled Kokkos backend and has been tested with Serial, OpenMP, Pthreads, CUDA, and HIP backends. The first major release includes simulations of directional solidification, grain structure output and analysis routines, and more. Installation instructions are in the ExaCA README.
Learn more:
New Repo: Wintap
July 22, 2022 (new-repo)
Wintap is an extensible host-based agent for Windows. Wintap provides a singular and extensible service-based runtime environment, a unified data model, API abstraction, and data discovery for an integrated, locally hosted web-based analytic “workbench” from where real-time event streams can be queried and explored.
New Repo: DFTT
July 21, 2022 (new-repo)
DFTT (Detection Framework Testbed and Toolkit) is a database and associated Java programs intended to facilitate the development and testing of algorithms for operating suites of correlation and subspace detectors. This framework is a generalization of the system described in Harris and Dodge (2011). It allows retrospective processing of sequences of data using various system configurations. Results are saved in a database, so it is easy to compare the results obtained using different configurations of the system.
CCT 1.0.18 Released
July 20, 2022 (release)
The Coda Calibration Tool (CCT) calculates reliable moment magnitudes for small- to moderate-sized seismic events. This release includes updates to zooming functionality, user adjustable start times, new REST endpoint, updates to uncertainty quantification metrics, and more.
Learn more:
NAHOMCon22 Explores High Order Methods for PDEs
July 19, 2022 (event-report)
After a few years off, the North American High Order Methods Conference (NAHOMCon) returned on July 18-19 in San Diego. NAHOMCon provides a North American forum for computational scientists, mathematicians, scientists, and engineers to share ideas and techniques on, and further the state of the art of, high order methods for the solution of partial differential equations with applications to a broad range of scientific and engineering applications. The DOE co-design Center for Efficient Exascale Discretizations (CEED) participates in this conference. CEED is a partnership between two U.S. DOE laboratories (Livermore & Argonne) and five universities in support of the Exascale Computing Project.
The NAHOMCon22 program featured several LLNL speakers, whose GitHub profiles and abstracts are linked here:
Cardioid Supports Dataset of Cardiac Simulations
July 15, 2022 (story)
LLNL’s Data Science Institute launched the Open Data Initiative (ODI) a few years ago. The ODI shares LLNL’s rich, challenging, and unique datasets with the larger data science community. The goal is for these datasets to help support curriculum development, raise awareness around LLNL’s data science efforts, foster new collaborations, and be leveraged across other learning opportunities. The ODI currently has 13 publicly available datasets in its collection, the newest of which builds off the Cardioid open source code, which simulates the electrophysiology of the human heart. A research team has conducted a computational study to generate a dataset of cardiac simulations at high spatiotemporal resolutions. This dataset was built using real cardiac bi-ventricular geometries and clinically inspired endocardial activation patterns under different physiological and pathophysiological conditions. It consists of pairs of computationally simulated intracardiac transmembrane voltage recordings and electrocardiogram signals. Read more and download the data from the ODI web page.
spdlayers 0.3.0 Released
July 13, 2022 (release)
spdlayers provides symmetric positive definite (SPD) enforcement layers for PyTorch. Examples, an API, and documentation are included in the repo. This release adds the option for the Eigen layer to support semi-definite prediction.
Learn more:
New Repo: EchemFEM
July 13, 2022 (new-repo)
EchemFEM provides finite element solvers for electrochemical transport. Both continuous Galerkin (CG) and discontinuous Galerkin (DG) schemes are provided. The following transport mechanisms are available: diffusion, advection, electromigration. EchemFEM supports both non-porous and porous cases. The ionic potential can either be described using an electroneutrality constraint or a Poisson equation. In the porous case, the electronic potential can be described by a Poisson equation.
New Repo: SAC2000
July 05, 2022 (new-repo)
SAC2000 (Seismic Analysis Code for the third millennium) is a general purpose interactive program designed for the study of sequential signals, especially time-series data. Emphasis has been placed on analysis tools used by research seismologists in the detailed study of seismic events. Analysis capabilities include general arithmetic operations, Fourier transforms, three spectral estimation techniques, IIR and FIR filtering, signal stacking, decimation, interpolation, correlation, and seismic phase picking.
Caliper 2.8.0 Released
June 30, 2022 (release)
Caliper is a program instrumentation and performance measurement framework. It is designed as a performance analysis toolbox in a library, allowing one to bake performance analysis capabilities directly into applications and activate them at runtime. The latest release includes AMD ROCm/HIP activity profiling and tracing, MPI message size and count metrics, and more.
Learn more:
VisIt 3.3.0 Released
June 29, 2022 (release)
VisIt is an open source, interactive, scalable, visualization, animation, and analysis tool. Version 3.3.0 includes
- improved handling of color tables
- improved exporting to Blueprint
- updates to the French translation
- and much more!
Learn more:
New Repo: XScope
June 27, 2022 (new-repo)
XScope finds inputs that trigger floating-point exceptions, such as NaN (not a number) and infinity, in CUDA functions using Bayesian optimization (BO). XScope assumes that the CUDA functions are a black box, i.e., the source code is not available. It searches the input space using several methods to guide BO into extreme cases. When an input is found to trigger an exception in the target CUDA function, the input is shown to the user. A forthcoming paper related to this repo has been accepted to SC22.
New Repo: graphite
June 27, 2022 (new-repo)
graphite is a repository for implementing graph network models based on atomic structures. It is meant to be a general toolbox of graph model codes (e.g., helper functions, custom graph convolutions, template graph models) for atomic structures. The repository contains examples and links to implemented and replicated works.
Charliecloud 0.28 Released
June 27, 2022 (release)
LANL led with LLNL contributors, Charliecloud provides user-defined software stacks for HPC centers. It uses Linux user namespaces to run containers with no privileged operations or daemons and minimal configuration changes on center resources. Version 0.28 includes changes to ch-image and ch-run.
Learn more:
spdlayers 0.2.0 Released
June 23, 2022 (release)
spdlayers provides symmetric positive definite (SPD) enforcement layers for PyTorch. Examples, an API, and documentation are included in the repo. This release adds a positivity function and sets its default.
Learn more:
New RADIUSS Activity Portal
June 23, 2022 (this-website)
The RADIUSS project has a new activity portal (source code for this GitHub action) that displays contributors’ activities, such as pull requests, issue comments, and releases. Visitors to this page can browse the entire list or filter by activity type to learn more about who contributes to RADIUSS and how they do so. RADIUSS aims to develop and deploy a common base of foundational scientific software with opt-in adoption from LLNL applications in order to reduce long-term software costs and increase agility.
Variorum 0.5.0 Released
June 22, 2022 (release)
Variorum is a platform-agnostic library exposing monitor and control interfaces for several features in hardware architectures. Its general interfaces provide privileged functionality for monitoring and controlling various hardware-level features of multiple hardware architectures. The latest release includes support for new architectures (i.e., AMD CPUs, Intel Ice Lake), API improvements, and more.
Learn more:
New Repo: MetallData
June 16, 2022 (new-repo)
MetallData is an HPC platform for interactive data science applications at HPC-scales. It provides an ecosystem for persistent distributed data structures, including algorithms, interactivity, and storage.
@LLNL_OpenSource Account Is Twitter Verified
June 10, 2022 (this-website)
Our @LLNL_OpenSource Twitter account is now officially verified with a blue checkmark. This badge affirms the account’s authenticity and notability to users and confirms that our account is active. Read more about verified accounts on Twitter’s help website.
Flux 0.40.0 Released
June 07, 2022 (release)
Flux is a flexible framework for resource management consisting of a suite of projects, tools, and libraries which may be used to build site-custom resource managers for HPC centers. Version 0.40.0 includes upgrades to verify database integrity during module load and support new sdexec job launch plugin, as well as fixes to reconnect, job attributes, Python docstrings, and more.
Learn more:
CCT 1.0.17 Released
June 02, 2022 (release)
The Coda Calibration Tool (CCT) calculates reliable moment magnitudes for small- to moderate-sized seismic events. This release includes updates to many naming conventions, updates to the Data filter functionality, new functions to export spectra value JSON directly from the spectra plots, and much more.
Learn more:
Spack 0.18.0 Released (Including Public Binaries!)
May 28, 2022 (release)
Spack is a flexible, configurable, Python-based, and open-source HPC package manager. Spack automates the installation and fine-tuning of simulations and libraries, operating on a wide variety of HPC platforms and enabling users to build many code configurations. Version 0.18.0 includes:
- concretizer updates
- finer-grained hashes
- improved error messages
- new binary format and hardened signing
- initial Windows support
- and much more!
Learn more:
LLNL and AWS to Cooperate on Standardized HPC Software Stack
May 26, 2022 (story)
LLNL and Amazon Web Services (AWS) have signed a memorandum of understanding (MOU) to define the role of leadership-class HPC in a future where cloud HPC is ubiquitous. Under the MOU, LLNL and AWS will explore software and hardware solutions spanning cloud and on-premises HPC environments, with the goal of establishing a common stack of open source software components that can run equally well at both large HPC centers and on cloud resources. LLNL and AWS have an existing open source collaboration involving Spack; building off that collaboration, the organizations will look to better understand how HPC centers can best utilize cloud resources to support HPC and explore models for cloud-bursting, data staging, and data migration for deploying both on-site and in the cloud. Read more about the MOU at LLNL News.
GLVis 4.2 Released
May 23, 2022 (release)
GLVis is a lightweight OpenGL tool for accurate and flexible finite element visualization. Version 4.2 includes:
- 3D scene export to glTF format for raytracing and augmented reality
- Support for visualization of pyramid-shaped elements
- Pre-built Mac and Windows binaries
- Perforated faces in 3D
Learn more:
Magpie 3.0 Released
May 21, 2022 (release)
Magpie contains a number of scripts for running big data software in HPC environments. This release includes support for Python 3 as well as new versions of Spark and Hadoop. See the release notes for information on backwards compatibility.
Learn more:
New Repo: ATS
May 10, 2022 (new-repo)
ATS (Automated Testing System) is a Python-based tool for automating the running of an application’s tests across a broad range of high-performance computers. ATS features include generating and comparing plot files, testing code correctness and performance with different software libraries and compiler configurations, and archiving metrics related to each test run. The tool is flexible enough for experts and non-experts alike. ATS Documentation is available, and more information about how LLNL uses ATS can be found on the Computing website.
LLNL at ISC22
May 06, 2022 (event)
ISC High Performance Conference (ISC22) returns on May 29 through June 2, with in-person events held in Hamburg, Germany. The event brings together the HPC community—from research centers, commercial companies, academia, national laboratories, government agencies, exhibitors, and more—to share the latest technology of interest to HPC developers and users. View LLNL’s lineup of tutorials, BOFs, and workshops.
LLNL's Spring Hackathon (and 32nd Overall) Coming Up
May 06, 2022 (event)
Held since 2012, LLNL’s hackathons are 24-hour opportunities to brainstorm, foster creativity, prototype, and explore. Participants work in groups or individually and often strive to learn new skills, programming languages, and tools in service to LLNL’s missions. This year’s spring event (May 26-27) will be held in person at the Livermore Valley Open Campus. Sponsors are two Computing divisions: Enterprise Applications Services and National Ignition Facility Computing.
Flux 0.39.0 Released
May 06, 2022 (release)
Flux is a flexible framework for resource management consisting of a suite of projects, tools, and libraries which may be used to build site-custom resource managers for HPC centers. Version 0.39.0 includes new features to get all job attributes, track protocol changes, and add color and highlight, plus testsuite updates and more.
Learn more:
New RADIUSS Catalog and Repo
May 03, 2022 (new-repo) (this-website)
The RADIUSS project has a new look including an About page and an interactive catalog of open-source products. These new web pages are managed in a dedicated repo under the LLNL GitHub organization. RADIUSS aims to develop and deploy a common base of foundational scientific software with opt-in adoption from LLNL applications in order to reduce long-term software costs and increase agility.
RAJAPerf 0.12.0 Released
May 02, 2022 (release)
The RAJA performance suite (RAJAPerf) is designed to explore performance of loop-based computational kernels found in HPC applications. It is used to assess, monitor, and compare runtime performance of kernels implemented using RAJA and variants implemented using standard or vendor-supported parallel programming models directly. The v0.12.0 release adds new command line options, basic MPI support, and new kernels, as well as updated versions of RAJA and BLT submodules.
Learn more:
Hatchet 2022.1.0 Released
April 28, 2022 (release)
Hatchet is a Python-based library that allows Pandas dataframes to be indexed by structured tree and graph data. This release catches Hatchet up to the new version nomenclature for the RAJA Portability Suite, which has changed format to indicate year, month, and patch number. New features include query language extensions, interactive visualization enhancements, and new color maps for terminal tree visualization. Additionally, the release has changes to existing APIs as well as updated tutorials and documentation.
Learn more:
Exascale Computing Project Community BOF Days
April 25, 2022 (event)
The Exascale Computing Project (ECP) 2022 Community Birds-of-a-Feather (BOF) Days will take place May 10–12 with multiple sessions per day. The BOF Days provide an opportunity for the HPC community to engage with ECP teams to discuss latest development efforts. Each BOF will be a 60- to 90-minute session on a given topic, with a brief overview followed by Q&A. All sessions will be conducted via Zoom. View the schedule; each session has its own registration link.
MFEM Named a NumFOCUS Affiliate Project
April 24, 2022 (story)
NumFOCUS is a nonprofit organization that promotes and supports open source scientific computing projects, such as Jupyter, Julia, and NumPy. NumFOCUS affiliate projects are “scientifically oriented, open, and kind,” which means they contribute to the advancement of scientific research, are published under a standard open source license, seek engagement with the public, and openly state their values and standards for the project community. MFEM, a scalable finite element methods library led by an LLNL team, was recently selected to join the list of NumFOCUS affiliate projects, broadening the project’s exposure and potential user and contributor community.
SUNDIALS 6.2.0 Released
April 22, 2022 (release)
SUNDIALS is a SUite of Nonlinear and DIfferential/ALgebraic equation Solvers. This minor release includes
SUNLogger API for logging of errors, warnings, informational output, and debugging output (many functions have been deprecated in favor of this API)- new functions to output all of the integrator, nonlinear solver, linear solver, and other statistics in one call
- new functions to change the default step size adaptivity parameters in CVODE, CVODES, IDA, and IDAS
- and many other new and updated functions
Learn more:
New Repo: AMM
April 21, 2022 (new-repo)
AMM, which stands for Adaptive Multilinear Meshes, is a new framework to represent piecewise multilinear volumetric data using mixed-precision adaptive meshes. AMM is designed to reduce in-memory and on-disk data footprint using a spatial hierarchy with “rectangular cuboidal” cells. AMM also supports mixed-precision representation of function values in byte-sized increments. The current version provides several examples of data streams that can be selected through command line. For details on AMM’s data structure and representation, see the accompanying publication, which won the Best Paper Award at the 2022 PacificVis Symposium. The abstract follows:
Adaptive representations are increasingly indispensable for reducing the in-memory and on-disk footprints of large-scale data. Usual solutions are designed broadly along two themes: reducing data precision, e.g., through compression, or adapting data resolution, e.g., using spatial hierarchies. Recent research suggests that combining the two approaches, i.e., adapting both resolution and precision simultaneously, can offer significant gains over using them individually. However, there currently exist no practical solutions to creating and evaluating such representations at scale. In this work, we present a new resolution-precision-adaptive representation to support hybrid data reduction schemes and offer an interface to existing tools and algorithms. Through novelties in spatial hierarchy, our representation, Adaptive Multilinear Meshes (AMM), provides considerable reduction in the mesh size. AMM creates a piecewise multilinear representation of uniformly sampled scalar data and can selectively relax or enforce constraints on conformity, continuity, and coverage, delivering a flexible adaptive representation. AMM also supports representing the function using mixed-precision values to further the achievable gains in data reduction. We describe a practical approach to creating AMM incrementally using arbitrary orderings of data and demonstrate AMM on six types of resolution and precision datastreams. By interfacing with state-of-the-art rendering tools through VTK, we demonstrate the practical and computational advantages of our representation for visualization techniques. With an open-source release of our tool to create AMM, we make such evaluation of data reduction accessible to the community, which we hope will foster new opportunities and future data reduction schemes.
Conduit 0.8.3 Released
April 14, 2022 (release)
Conduit provides an intuitive model for describing hierarchical scientific data in C++, C, Fortran, and Python. It is used for data coupling between packages in-core, serialization, and I/O tasks. The latest release adds new blueprint mesh examples and C/C++ version macros. In addition, the pip install logic has been improved.
Learn more:
New Repo: LOPE
April 13, 2022 (new-repo)
LOPE (Lattice Optimization for Porous Electrodes) performs an optimization over the structure of a porous electrode formed from a lattice of unit cells. This code accompanies the paper “Computational Design of Microarchitected Porous Electrodes for Redox Flow Batteries.” The abstract follows:
Porous electrodes are used as the core reactive component across electrochemical technologies. In flowing systems, controlling the fluid distribution, species transport, and reactive environment is critical to attaining high performance. However, conventional electrode materials like felts and papers provide few opportunities for precise engineering of the electrode and its microstructure. To address these limitations, architected electrodes composed of unit cells with spatially varying geometry determined via computational optimization are proposed. Resolved simulation is employed to develop a homogenized description of the constituent unit cells. These effective properties serve as inputs to a continuum model for the electrode when used in the negative half-cell of a vanadium redox flow battery. Porosity distributions minimizing power loss are then determined via computational design optimization to generate architected porosity electrodes. The architected electrodes are compared to bulk, uniform porosity electrodes and found to lead to increased power efficiency across operating flow rates and currents. The design methodology is further used to generate a scaled-up electrode with comparable power efficiency to the bench-scale systems. The variable porosity architecture and computational design methodology presented here thus offers a novel pathway for automatically generating spatially engineered electrode structures with improved power performance.
New Repo: BinFPE
April 12, 2022 (new-repo)
BinFPE detects floating-point exceptions (NaN, infinity, and subnormal quantities) in NVIDIA GPU applications using binary instrumentation. It requires no re-compilation of the application and can analyze libraries. The tool extends NVBit, which is provided by NVIDIA Labs to analyze binaries.
LLNL Launches Software Development Resource Center
April 10, 2022 (story)
The Lab’s software developers and engineers have accumulated a wealth of expertise by keeping LLNL operational while carrying out its national security mission. Launched in 2022, the Software Development Resource Center (SDRC) connects developers across LLNL through best practices in software tools, development methodologies, DevOps, security compliance, and more. Funded as an Institutional Scientific Capability Portfolio (ISCP) project, the SDRC serves as a focal point for software leadership, including coordinating working groups and providing technical advice to project teams. For instance, the SDRC committee plans to schedule seminars and invited talks as well as conduct workshops during this summer’s Developer Day. The SDRC also arises from Computing’s 10-year Strategic Plan, which prioritizes support for mission-driven programs through robust software engineering and maintenance. Read more about the SDRC.
New Repo: MTNN
April 09, 2022 (new-repo)
MTNN, which stands for Multilevel Neural Networks, is a PyTorch-based library for the application of multilevel algorithms to the training of neural networks. The algorithms behind MTNN mathematically work whenever the neural network can be decomposed into a set of operational subsets such that each subset consists of a set of neurons, channels, or other similar operational units. In this case, pairs of neurons (or channels or other operational units) can be matched up and restricted into a coarse network. Requires Python 3.6 or newer.
New Repo: TargetID
April 08, 2022 (new-repo)
TargetID is a drug target and chemotype identification pipeline that enables rapid identification and characterization of binding sites in SARS-CoV-2 proteins as well as the core chemical components with which these sites interact. The repo contains three Jupyter Notebooks and input data files.
New Repo: Pond B
April 07, 2022 (new-repo)
Pond B documents the code used to analyze the Pond B (Savannah River Site, South Carolina) microbial community for a manuscript titled “Microbial Dynamics Impact Plutonium and Iron Biogeochemical Cycles in a Seasonally Stratified Pond.” The repo includes figures and tables from the paper, which is being prepared for publication.
New Repo: XNAS
April 06, 2022 (new-repo)
XNAS, which stands for eXplainable Neural Architecture Search, provides code for the paper “Learning Interpretable Models Through Multi-Objective Neural Architecture Search.” Requires Python 3.6 or newer. The abstract follows:
Monumental advances in deep learning have led to unprecedented achievements across a multitude of domains. While the performance of deep neural networks is indubitable, the architectural design and interpretability of such models are nontrivial. Research has been introduced to automate the design of neural network architectures through neural architecture search (NAS). Recent progress has made these methods more pragmatic by exploiting distributed computation and novel optimization algorithms. However, there is little work in optimizing architectures for interpretability. To this end, we propose a multi-objective distributed NAS framework that optimizes for both task performance and introspection. We leverage the non-dominated sorting genetic algorithm (NSGA-II) and explainable AI (XAI) techniques to reward architectures that can be better comprehended by humans. The framework is evaluated on several image classification datasets. We demonstrate that jointly optimizing for introspection ability and task error leads to more disentangled architectures that perform within tolerable error.
New Repo: op
April 05, 2022 (new-repo)
op is a lightweight general optimization solver interface. Its primary purpose is to simplify the process of integrating different optimization solvers (serial or parallel) with scalable parallel physics engines. From a user perspective, optimization problems can be described abstractly, and op guarantees portability for different optimization engines and optimization problem configurations.
HiOp 0.6.0 Released
March 31, 2022 (release)
HiOp is an optimization solver for solving certain mathematical optimization problems expressed as nonlinear programming problems. This lightweight HPC solver leverages application’s existing data parallelism to parallelize the optimization iterations by using specialized linear algebra kernels. This version includes the primal decomposition (PriDec) solver for structured two-stage problems as well as improved support for NVIDIA GPUs for solving sparse optimization problems.
Learn more:
Charliecloud 0.27 Released
March 31, 2022 (release)
LANL led with LLNL contributors, Charliecloud provides user-defined software stacks for HPC centers. It uses Linux user namespaces to run containers with no privileged operations or daemons and minimal configuration changes on center resources. Version 0.27 includes new tests, examples, and FAQ; as well as additional functionality for ch-convert, ch-image, and ch-run.
Learn more:
CHAI 2022.03.0 Released
March 29, 2022 (release)
CHAI is a C++ library providing an array object that can be used transparently in multiple memory spaces. CHAI can be used standalone, but is best when paired with the RAJA library, which has built-in CHAI integration that takes care of everything. This release catches CHAI up to the new version nomenclature for the RAJA Portability Suite, which has changed format to indicate year, month, and patch number.
Learn more:
New Repo: ddd
March 24, 2022 (new-repo)
ddd, which stands for Delaunay Density Diagnostic, implements algorithms described in a numerical analysis paper. The abstract follows:
Accurate approximation of a real-valued function depends on two aspects of the available data: the density of inputs within the domain of interest and the variation of the outputs over that domain. There are few methods for assessing whether the density of inputs is \textit{sufficient} to identify the relevant variations in outputs – i.e., the “geometric scale” of the function – despite the fact that sampling density is closely tied to the success or failure of an approximation method. In this paper, we introduce a general purpose, computational approach to detecting the geometric scale of real-valued functions over a fixed domain using a deterministic interpolation technique from computational geometry. Our algorithm is based on the observation that a sequence of piecewise linear interpolants will converge to a continuous function at a quadratic rate (in L2 norm) if and only if the data are sampled densely enough to distinguish the feature from noise. We present numerical experiments demonstrating how our method can identify feature scale, estimate uncertainty in feature scale, and assess the sampling density for fixed (i.e. static) datasets of input-output pairs. In addition, we include analytical results in support of our numerical findings and will release lightweight code that can be adapted for use in a variety of data science settings.
MFEM 4.4 Released
March 22, 2022 (release)
MFEM is a lightweight, general, scalable C++ library for finite element methods. It enables high-performance scalable finite element discretization research and application development on a wide variety of platforms, from laptops to supercomputers. The v4.4 release includes:
- upgraded solvers, such as improved formatting of iterative solvers output and support for hypre on AMD GPUs
- meshing improvements, such as new TMOP-based methods and initial support for meshes with pyramidal elements
- discretization improvements, such as GPU-enabled partial and element assembly support for discontinuous Galerkin methods
- several new examples and mini-apps
- updated documentation
- and much more
Learn more:
libCEED 0.10.0 Released
March 21, 2022 (release)
Developed as part of the Center for Efficient Exascale Discretizations (CEED) within the U.S. Department of Energy, libCEED is a high-order API library that provides a common algebraic low-level operator description, allowing a wide variety of applications to take advantage of the efficient operator evaluation algorithms in the different CEED packages. This release includes single precision support, a capability to assemble operators on GPUs, performance enhancements, interface and error checking improvements, and mini-app improvements.
Learn more:
New Repo: krowkee
March 21, 2022 (new-repo)
krowkee is a toolkit for scalably and efficiently summarizing many data streams in distributed memory. It is intended for applications where one needs to summarize huge loosely structured data, such as matrices or graphs, and where individual components such as rows/columns or vertex adjacency information are impractical to store and directly inspect.
CCT 1.0.16 Released
March 18, 2022 (release)
The Coda Calibration Tool (CCT) calculates reliable moment magnitudes for small- to moderate-sized seismic events. This release includes updates to the multi-waveform display panel, term and relative site term plots, Site and Measurement results, and more.
Learn more:
New Repo: TOPE
March 17, 2022 (new-repo)
TOPE, which stands for Topology Optimization for Porous Electrodes, produced the results of a paper published in Applied Physics. The abstract follows:
Porous electrodes are an integral part of many electrochemical devices since they have high porosity to maximize electrochemical transport and high surface area to maximize activity. Traditional porous electrode materials are typically homogeneous, stochastic collections of small scale particles and offer few opportunities to engineer higher performance. Fortunately, recent breakthroughs in advanced and additive manufacturing are yielding new methods to structure and pattern porous electrodes across length scales. These architected electrodes are emerging as a promising new technology to continue to drive improvement; however, it is still unclear which structures to employ and few tools are available to guide their design. In this work we address this gap by applying topology optimization to the design of porous electrodes. We demonstrate our framework on two applications: a porous electrode driving a steady Faradaic reaction and a transiently operated electrode in a supercapacitor. We present computationally designed electrodes that minimize energy losses in a half-cell. For low conductivity materials, the optimization algorithm creates electrode designs with a hierarchy of length scales. Further, the designed electrodes are found to outperform undesigned, homogeneous electrodes. Finally, we present three-dimensional porous electrode designs. We thus establish a topology optimization framework for designing porous electrodes.
RAJA 2022.03.0 Released
March 15, 2022 (release)
RAJA is a software abstraction that systematically encapsulates platform-specific code to enable applications to be portable across diverse hardware architectures without major source code disruption.
Note that the release version nomenclature has changed format to indicate year, month, and patch number. This format applies to the coordinated release of RAJA Portability Suite components. The v2022.03.0 release includes:
- features that support SIMD/SIMT programming
- ROCTX support to enable kernel naming
- expanded CUDA execution policies
- improved CMake logic for using CUB
- updated submodules
- and more
As of this release, RAJA requires CMAKE version 3.14.5 or newer and a C++14-compliant compiler to build.
Learn more:
Kosh 2.1 Released
March 15, 2022 (release)
Kosh allows codes to store, query, and share data via an easy-to-use Python API. This software aims to make data access and sharing as simple as possible. The latest release includes updates to transformers, operators, dataset cloning, dataset objects, and more.
Learn more:
Julian Andrej Applies Mathematics and Engineering to Support LLNL Missions
March 09, 2022 (profile)
Computational mathematician Julian Andrej began using LLNL-developed, open source software while in Germany. Now at Livermore, he lends his expertise to the Center for Applied Scientific Computing, developing code for next-generation computing hardware. “I know that every part of my work is contributing to a mission, and I can see a clear traceable path of its impact,” he says. Andrej contributes to MFEM and SUNDIALS projects. Read more about his work.
BLT 0.5.0 Released
March 07, 2022 (release)
BLT (Building, Linking, and Testing) is a streamlined CMake build system foundation for developing HPC software. BLT makes it easy to get up and running on a wide range of HPC compilers, operating systems, and technologies. The v0.5.0 release includes added support for IntelLLVM compilers and hip targets, along with changes to the CMAKE_HIP_ARCHITECTURES variable and other modifications.
Learn more:
MuyGPyS 0.5.0 Released
March 01, 2022 (release)
MuyGPyS is a Gaussian process estimation method that affords fast hyperparameter optimization by way of performing leave-one-out cross-validation. The latest release introduces just-in-time compilation and GPU support using JAX. This change allows for the acceleration of workflows on CPU (1-2.5x) and NVidia GPU (30-60x).
Learn more:
The Flux Supercomputing Workload Manager: Improving on Innovation and Planning for the Future
February 24, 2022 (story)
The LLNL-developed Flux project addresses challenges posed by complex scientific research supercomputing workflows, and the team has played a major role in the Exascale Computing Project’s ExaWorks effort. A recent ECP article describes how Flux works, its impact, and what’s next for the project.
Camp 0.5.0 Released
February 23, 2022 (release)
CAMP collects a variety of macros and metaprogramming facilities for C++ projects. This version incorporates BLT updates to better support rocm builds, and uses the cmake setup to include a binary component that lets CAMP avoid pulling in expensive standard library headers at include time.
Learn more:
Spack's Long-Term Roadmap
February 22, 2022 (story)
The Spack team presented a long-term roadmap at the Networking and Information Technology Research and Development (NITRD) Program’s recent meeting on software sustainability. NITRD coordinates federal research and development to identify, develop, and transition into use the Unites States’ IT, HPC, networking, and software capabilities. With more than 6,000 software packages and nearly 1,000 contributors, Spack’s long-term strategy is based around broad adoption and collaboration. Current R&D efforts include a multi-year strategic initiative aimed at reducing the human maintenance burden, as well as improvements to Spack’s continuous integration workflow and sustainability plans after the conclusion of the Exascale Computing Project.
SCR 3.0 Released
February 16, 2022 (release)
The Scalable Checkpoint/Restart (SCR) library enables MPI applications to utilize distributed storage on Linux clusters to attain high file I/O bandwidth for checkpointing, restarting, and writing large datasets. This release includes:
- added Python bindings for the SCR library
- improved support for large datasets and shared access to files
- watchdog support on SLURM systems
- new API calls
- new redundancy scheme
- and more
Learn more:
New Repo: SYNDATA
February 15, 2022 (new-repo)
SYNDATA includes a suite of statistical/machine learning models to generate discrete/categorical synthetic data. To train each model, the user must provide the input data from which the model parameters will be inferred. Once the models are trained, they can be used to generate entirely synthetic data. In addition to the actual models, SYNDATA includes code to process data, evaluate results (based on cross validation), and create a PDF report.
Merlin Integrates Machine Learning into Scientific HPC Workflows
February 15, 2022 (story)
The Merlin team has published a paper in Future Generation Computer Systems that describes Merlin’s machine learning–integrated workflow system and the considerations driving its design. The authors detail Merlin’s performance results on LLNL’s Pascal and Sierra supercomputers. For example, to demonstrate Merlin’s scalability, researchers created an unprecedentedly large fusion simulation dataset consisting of the multivariate results of approximately 100 million individual simulations on Sierra. Other case studies show Merlin’s flexibility in cascading and iterative scientific workflows.
New Repo: PyMFEM
February 14, 2022 (new-repo)
PyMFEM provides Python binding for MFEM, a high-performance parallel finite element method (FEM) library. Installer (setup.py) builds both MFEM and binding together. By default, pip install mfem downloads and builds the serial version of MFEM and PyMFEM. Additionally, the installer supports building MFEM with specific options together with other external libraries, including MPI version.
Hatchet 1.3.0 Released
February 07, 2022 (release)
Hatchet is a Python-based library that allows Pandas dataframes to be indexed by structured tree and graph data. Version 1.3.0 includes a new interactive tree visualization in Jupyter, new APIs, new graph output formats, an updated tutorial, and more.
Learn more:
VisIt 3.2.2 Released
January 31, 2022 (release)
VisIt is an open source, interactive, scalable, visualization, animation, and analysis tool. Version 3.2.2 includes
- expansion to VTK reader
- improved installation of Python endpoint scripts
- enhanced
build_visit features
Learn more:
Quandary 2.1 Released
January 26, 2022 (release)
Quandary provides optimal control for open quantum systems via an optimization solver. The underlying quantum dynamics model open quantum systems, using the Lindblad master equation to evolve a density matrix in time. With this latest release, Quandary can solve either Lindblad’s master equation for the density matrix (open quantum systems), or Schrödinger’s equation for the state vector if no collapse operators are present (closed quantum systems). The latter drastically reduces the computational complexity; however, system interactions with the environment are not considered in this model.
Learn more:
Peter Lindstrom: Then and Now
January 25, 2022 (profile)
The Department of Energy’s (DOE’s) Office of Science interviewed LLNL computer scientist Peter Lindstrom about his work since receiving the 2011 Early Career Award. Lindstrom leads the zfp project, which provides a compressed format for representing multidimensional floating-point and integer arrays. zfp is currently used in the DOE’s Exascale Computing Project.
libyogrt 1.27 Released
January 24, 2022 (release)
libyogrt (Your One Get Remaining Time library) provides functions to query a resource manager for the time remaining in a job. It supports LCRM, LSF, MOAB, SLURM, and AIX with SLURM. The project has been around for many years and just recently added support for the Flux resource management framework.
Learn more:
Charliecloud 0.26 Released
January 20, 2022 (release)
LANL led with LLNL contributors, Charliecloud provides user-defined software stacks for HPC centers. It uses Linux user namespaces to run containers with no privileged operations or daemons and minimal configuration changes on center resources. Version 0.26 includes added functionality for several ch- scripts and an expanded test suite.
Learn more:
Supplemental Repositories
January 19, 2022 (new-repo)
Did you know that the LLNL GitHub catalog contains several supplemental repos? They are “companions” to primary repos and provide additional data or functionality, such as examples or test cases. For instance, and this is not an exhaustive list:
MuyGPyS 0.4.1 Released
January 19, 2022 (release)
MuyGPyS is a Gaussian process estimation method that affords fast hyperparameter optimization by way of performing leave-one-out cross-validation. The latest release streamlines the codebase by moving all of the computation inside of pure functions.
Learn more:
Magpie 2.6 Released
January 18, 2022 (release)
Magpie contains a number of scripts for running big data software in HPC environments. This release includes support for multiple versions of Spark and Hadoop.
Learn more:
Kosh 2.0 Released
January 10, 2022 (release)
Kosh allows codes to store, query, and share data via an easy-to-use Python API. This software aims to make data access and sharing as simple as possible. The latest release aligns Kosh with Sina and makes it the only backend. Other new features include support for ensembles, ability to clone datasets, and additional command line options.
Learn more:
New Repo: UMT
January 07, 2022 (new-repo)
UMT, which stands for Unstructured Mesh Transport, is an LLNL ASC proxy application (mini-app) that solves a thermal radiative transport equation using discrete ordinates (Sn). It utilizes an upstream corner balance method to compute the solution to the Boltzmann transport equation on unstructured spatial grids. For additional information, the repo’s README includes references to relevant radiation transport publications.
FEM@LLNL Seminar Series
January 05, 2022 (multimedia) (story)
The MFEM team has announced a new FEM@LLNL seminar series focusing on finite element research and applications talks of interest to the MFEM community. Visit the MFEM website to see the full lineup of speakers. Seminars will be hosted and recorded via WebEx; videos of the recordings will be available from the MFEM website.
New Repo: Abmarl
January 03, 2022 (new-repo)
Abmarl is a package for developing Agent-Based Simulations and training them with MultiAgent Reinforcement Learning (MARL). It provides an intuitive CLI for engaging with the full workflow of MARL experimentation: training, visualizing, and analyzing agent behavior. Abmarl leverages RLlib’s framework for reinforcement learning and extends it to more easily support custom simulations, algorithms, and policies. Abmarl documentation is also available.
