Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “block design”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 523 records · Page 29

Developing nanometer scale, atomically precise metallo-catalysts with Molecular Lego

We propose to create atomically precise, highly robust, nanometer scale macromolecules that organize biomimetic metal binding groups in designed three-dimensional pockets to accelerate specific polymerization reactions. These atomically precise catalysts will create atomically precise polymers with improved mechanical and environmental properties from renewable building blocks. We seek to create and study nanometer scale catalysts that mimic metallo-enzymes in their activity and selectivity because these larger catalysts will be able to make more extensive non-covalent contacts with the transition state of the growing polymer. These catalysts will have extended lifetimes because they will better protect the metals and higher activity because additional reactive groups can be brought in close contact with the active site to enhance reactivity. They will be far more tolerant to extremes of temperature and will not denature in non-aqueous solvents because they are held together by multiple, strong, covalent bonds. Our approach to catalysts could be used to develop catalysts to replace noble metal-based catalysts when targeting other reactions, because the geometry of the active site enforced by the Molecular Lego scaffolding will access new reactivity, mimicking how nature uses earth abundant metals for much of its catalysis. We will initially target Lewis acid-based catalysts that assemble aliphatic polyesters with improved stereoselectivity, tacticity and complex alternation of monomers. Aliphatic polyesters have received growing attention as attractive, environmentally benign, and sustainable alternatives to polymers developed from petroleum feedstocks. Aliphatic polyesters are environmentally friendly because they undergo facile hydrolytic degradation to benign products, and they have excellent properties and high biocompatibility. We will assemble these catalysts using unique “Spiroligomers” (aka Molecular Lego) chemistry developed in the Schafmeister group combined with the organometallic chemistry expertise of the Dobereiner group. Polymers are an attractive synthetic target of nanoscale atomically precise catalysts because their resulting structure, stereochemistry and tacticity has profound impact on their properties (crystalline vs amorphous) and provides a readout of the catalytic mechanism. Polymer synthesis also serves as an example of “atomically precise manufacturing” where atomically precise nanoscale catalysts construct polymers with atomic precision at increasing length scales.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

MPACT Safeguards Modeling: FY25 Update

Sandia National Laboratories develops and maintains several open-source software packages to support material accountancy analyses. This includes the Material Accountancy Performance Indicator Toolkit (MAPIT), the Fissile Facility Flow Modeler (F3M) and the Separation and Safeguards Performance Model Library (SSPM-L). MAPIT is responsible for performing statistical safeguards analyses on bulk and itemized data from nuclear fuel cycle facilities and can operate on real or synthetic data. MAPIT is the only open-source software for such analyses. F3M is a library of modules, built in MATLAB Simulink, that contain pre made blocks to represent different generic fuel cycle processes. These blocks can be used together in a modular fashion to represent and simulate nuclear fuel cycle processes with the goal of improving facility-level accountancy during the design phase. F3M is also an open-source library. Finally, the SSPM-L library is a series of completed models built from F3M. The library includes facility models such as a generic PUREX facility and a fuel fabrication facility. The SSPM-L library is not open source, but is available to collaborators with a relevant use case. These tools include modeling and simulation pipelines to simulate nuclear fuel cycle facilities and the underlying software needed to simulate measurement uncertainty and perform statistical analyses. Together, these tools can perform end-to-end nuclear material accountancy analyses. This report documents the various improvements made to these tools in FY25. Specifically, we added new statistical test, new statistical modeling capabilities, new fuel cycle facility models, and launched a new open-source model component library.

97 MATHEMATICS AND COMPUTING↗

Progressing Analysis of Variable Electric Rates (PAVER) Study

The Progressing Analysis of Variable Electric Rates (PAVER) study analyzed the impact of a range of time-varying electric rates on the performance of a regional electric grid and the resulting costs for participating and non-participating customers. This analysis leveraged and extended the work of PNNL’s Distribution System Operator with Transactive (DSO+T) study. Five different rate designs were included: a flat volumetric energy charge, a typical Time of Use (TOU) rate, a dynamic energy (DE) rate (based on wholesale locational marginal prices), a dynamic energy and capacity (DE+C) rate, and, finally, a Block and Swing (B&S) rate that billed customers based on their average load profile at constant pricing, but used the DE+C dynamic price for load deviations from their average profile. These rates were analyzed in a large-scale co-simulation of an entire regional grid with a customer population representative of the current state. A large fraction (80%) of residential and commercial customers were assumed to participate in these time-varying rates with automatically controlled HVAC, water heaters, electric vehicles, and batteries. This study assumed no industrial sector participation. The DE and DE+C rates saw system peak loads reduced by 6-7%, while the large participation in the TOU rate case saw a significant rebound effect and a resulting peak load increase of >5%. The impacts to the annual and peak system demand impacted system wholesale prices and the overall grid operating costs. This cost structure determined the revenue needed to be collected from customers by each rate design. Participating customers on the DE and DE+C rates (located in one of the modeled DSOs) saw reductions in average annual electricity bills of 11-17% with average increases in monthly bill variation of no more than 13%. At such high participation levels, TOU customers saw 10% higher average annual bills (due to system-wide rebound effects) and average increased monthly bill variation of 16%. Residential owners of large flexible loads (such as electric vehicles) saw larger bill savings (17-20%) when on a fully dynamic rate. The presence of on-site generation (such as rooftop solar) did not appear to appreciably change customer outcomes. Customers on the Block and Swing rate did see 6% lower monthly bill variation (as intended) than the flat rate case, but at the expense of appreciable bill savings, which were only 3%, comparable to the savings seen by non-participants. Given this finding we recommend that additional research be conducted into how best various bill protection mechanisms can balance minimizing customer bill variation with providing financial incentives commensurate with the flexibility customers provide. We also recommend that customer outcomes be explored across a range of regions using current actual customer and system cost data.

29 ENERGY PLANNING, POLICY, AND ECONOMY↗

Open Radiation Monitoring: Histogram Builder Module Design

The Open Radiation Monitoring Project seeks to develop and demonstrate a modular radiation detection architecture designed specifically for use in arms control treaty verification (ACTV) applications that will facilitate rapid development of trusted systems to meet the needs of potential future treaties. A modular architecture can be used to reduce more complex systems to a series of single purpose building blocks, thereby facilitating equipment inspection and in turn building trust in the equipment by all treaty parties. Furthermore, a modular architecture can be used to control data flow within the measurement system, reducing the risk of "hidden switches" and constraining the amount of sensitive information that could potentially be inadvertently leaked. This report details the first revision of a prototype circuit that will convert analog pulses directly into a histogrammed data set for further processing. The circuit was designed with both spectroscopy and multiplicity analysis in mind but can, in principle, be used to reduce any raw data stream into a histogram. The number of output channels is limited, and the histogram bin ranges are user configurable to allow for non-uniform and discontinuous bins, which makes it possible to restrict the information being passed down stream if desired. Pulse processing relies entirely on analog circuitry and non- programmable logic, which enables operation without the need for a central processor or other programmable control unit. The circuit remains untested under the Open Radiation Monitoring project due to the closure of the sponsoring program. However, further development and testing is scheduled to take place in support of a purpose-built trusted verification system development effort known as COGNIZANT, which demonstrates the potential benefit of developing a suite of modular trusted system components.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Geometric control of emergent antiferromagnetic order in coupled artificial spin ices

Artificial spin ices (ASIs) composed of coupled nanomagnets offer the possibility to create designer geometrical frustration and manipulate inter-nanomagnet interactions. In particular, by using a dimer motif consisting of two strongly coupled single-domain nano magnets as a building block, we can control and realize intriguing antiferromagnetic physical states in which the magnetic charge is not conserved. Here we create a dimer kagome ASI system within which the antiferromagnetic order is controlled by tuning lattice geometry. Different disordered antiferromagnetic phases and the inter-nanomagnet interactions are visualized directly in demagnetized lattices with different lattice parameters. Monte Carlo simulations establish that the collective ground state consists of disordered antiferromagnetic dimers across the lattice when the intra-dimer interaction is dominant. However, for lattices governed by intra-triad-unit interactions, the ground state exhibits a longrange spin-ordered state in which the vertex magnetic charge is uniform across all triads.

36 MATERIALS SCIENCE↗

Sparsomycin inhibits translation through a conserved mechanism across all domains of life

Abstract Sparsomycin (SPA) is a broad-spectrum inhibitor of protein synthesis with activity across all three domains of life. Although SPA has long been known to target the ribosomal peptidyl transferase center (PTC), previous structural studies suggested that SPA binds differently to bacterial ribosomes compared to their archaeal and eukaryotic counterparts—an unexpected conclusion given the high evolutionary conservation of the ribosomal catalytic center. Here, we show that SPA inhibits a majority of elongation-competent bacterial ribosomal complexes and present X-ray crystal structures of Thermus thermophilus 70S ribosomes stalled by SPA at the initiation and early elongation stages of translation. These structures reveal that SPA binds to the bacterial ribosome in a manner essentially identical to that observed in archaeal and eukaryotic ribosomes, establishing a unified structural mechanism of SPA action across all domains of life. In this conserved binding mode, SPA occupies the A-site cleft of the PTC and forms an extensive network of interactions with universally conserved ribosomal RNA nucleotides and the CCA-end of the P-site transfer RNA (tRNA), thereby stabilizing the peptidyl-tRNA substrate while sterically blocking accommodation of an incoming aminoacyl-tRNA. By clarifying the mode of action of SPA on the bacterial ribosome, our work provides a structural framework for the rational design of SPA derivatives with improved potency and bacterial specificity.

Paranjpe, Madhura N [Department of Biological Scie↗

Single hepatocytes show persistence and transcriptional inactivity of hepatitis B

There is no cure for the more than 270 million people chronically infected with HBV. Nucleos(t)ide analogs (NUCs), the mainstay of anti-HBV treatment, block HBV reverse transcription. NUCs do not eliminate the intranuclear covalently closed circular DNA (cccDNA), from which viral RNAs, including pregenomic RNA (pgRNA), are transcribed. A key gap in designing a cure is understanding how NUCs affect HBV replication and transcription because serum markers yield an incomplete view of intrahepatic HBV. We applied single-cell laser capture microdissection and droplet digital PCR to paired liver biopsies collected from 5 HBV/HIV-coinfected persons who took NUCs over 2–4 years. From biopsy 1 to 2, proportions of HBV-infected hepatocytes declined with adherence to NUC treatment (P < 0.05); we extrapolated that eradication of HBV will take over 10 decades with NUCs in these participants. In individual hepatocytes, pgRNA levels diminished 28- to 73-fold during NUC treatment, corresponding with decreased tissue HBV core antigen staining (P < 0.01). In 4 out of 5 participants, hepatocytes with cccDNA but undetectable pgRNA (transcriptionally inactive) were present, and these were enriched in 3 participants during NUC treatment. Further work to unravel mechanisms of cccDNA transcriptional inactivation may lead to therapies that can achieve this in all hepatocytes, resulting in a functional cure.

59 BASIC BIOLOGICAL SCIENCES↗

AMReX: Block-structured adaptive mesh refinement for multiphysics applications

Block-structured adaptive mesh refinement (AMR) provides the basis for the temporal and spatial discretization strategy for a number of Exascale Computing Project applications in the areas of accelerator design, additive manufacturing, astrophysics, combustion, cosmology, multiphase flow, and wind plant modeling. AMReX is a software framework that provides a unified infrastructure with the functionality needed for these and other AMR applications to be able to effectively and efficiently utilize machines from laptops to exascale architectures. AMR reduces the computational cost and memory footprint compared to a uniform mesh while preserving accurate descriptions of different physical processes in complex multiphysics algorithms. AMReX supports algorithms that solve systems of partial differential equations in simple or complex geometries and those that use particles and/or particle–mesh operations to represent component physical processes. In this article, we will discuss the core elements of the AMReX framework such as data containers and iterators as well as several specialized operations to meet the needs of the application projects. In addition, we will highlight the strategy that the AMReX team is pursuing to achieve highly performant code across a range of accelerator-based architectures for a variety of different applications.

Zhang, Weiqun↗

On the prospects of optical cycling in diatomic cations: effects of transition metals, spin–orbit couplings, and multiple bonds

Molecules with optical cycling centers (OCCs) are highly desirable in the context of fundamental studies as well as applications (e.g., quantum computing) because they can be effectively cooled to very low temperatures by repeated absorption and emission (hence, cycling). Charged species offer additional advantages for experimental control and manipulation. Here, we present a systematic computational study of a series of diatomic radical-cations made of a d-block metal and a p-block ligand, that are isoelectronic (in their valence shell) to the successfully laser-cooled neutral molecules. Using high-level electronic structure methods, we characterize state and transition properties of low-lying electronic states and compute Franck-Condon factors. The computed branching ratios and radiative lifetimes reveal that the electronic transitions analogous to those successfully used in the laser cooling of neutral molecules are less than optimal in the cations. We propose alternative transitions suitable for optical cycling and highlight trends that could assist future designs of OCCs in charged or neutral molecules.

74 ATOMIC AND MOLECULAR PHYSICS↗

Modulating the Molecular Geometry and Solution Self-Assembly of Amphiphilic Polypeptoid Block Copolymers by Side Chain Branching Pattern

Solution self-assembly of coil-crystalline diblock copolypeptoids has attracted increasing attention due to its capability to form hierarchical nanostructures with tailorable morphologies and functionalities. While the N-substituent (or side chain) structures are known to affect the crystallization of polypeptoids, their roles in dictating the hierarchical solution self-assembly of diblock copolypeptoids are not fully understood. Herein, we designed and synthesized two types of diblock copolypeptoids, i.e., poly(N-methylglycine)-b-poly(N-octylglycine) (PNMG-b-PNOG) and poly(N-methylglycine)-b-poly(N-2-ethyl-1-hexylglycine) (PNMG-b-PNEHG), to investigate the influence of N-substituent structure on the crystalline packing and hierarchical self-assembly of diblock copolypeptoids in methanol. With a linear aliphatic N-substituent, the PNOG blocks pack into a highly ordered crystalline structure with a board-like molecular geometry, resulting in the self-assembly of PNMG-b-PNOG molecules into a hierarchical microflower morphology composed of radially arranged nanoribbon subunits. By contrast, the PNEHG blocks bearing bulky branched aliphatic N-substituents are rod-like and prefer to stack into a columnar hexagonal liquid crystalline mesophase, which drives PNMG-b-PNEHG molecules to self-assemble into symmetrical hexagonal nanosheets in solution. A combination of time-dependent small/wide-angle X-ray scattering and microscopic imaging analysis further revealed the self-assembly mechanisms for the formation of these microflowers and hexagonal nanosheets. These results highlight the significant impact of the N-substituent architecture (i.e., linear versus branched) on the supramolecular self-assembly of diblock copolypeptoids in solution, which can serve as an effective strategy to tune the geometry and hierarchical structure of polypeptoid-based nanomaterials.

99 GENERAL AND MISCELLANEOUS↗

Control Design of Passive Grid-Forming Inverters in Port-Hamiltonian Framework

This article presents a modified dispatchable virtual oscillator control approach for achieving the passivity of gridforming inverters (GFMs), without assuming constant voltage and constant frequency. The proposed control framework utilizes the Port-Hamiltonian (PH) based structure that mimics the behaviors of coupled harmonic oscillators, along with an energy ‘pumpingor-damping’ block and the Control by Interconnection (CbI) technique, to render the inverter passive. Once passivity is achieved, transient stability of the system will be guaranteed. The proposed control framework is composed of three loops: an outer power dispatching loop that generates the voltage and frequency references, a virtual oscillator loop that emulates the spontaneous synchronization of oscillators, and an inductor current loop that maintains lossless interconnection in PH systems. In conclusion, the study shows that the proposed control approach ensures the passivity of GFMs, facilitating the transient stability design of multi-inverter systems, as interconnections of passive systems remain passive and stable.

42 ENGINEERING↗

A Solution-Processable High-Modulus Crystalline Artificial Solid Electrolyte Interphase for Practical Lithium Metal Batteries

The solid electrolyte interphase (SEI) has been identified as a key challenge for Li metal anodes. The brittle and inhomogeneous native SEI generated by parasitic reactions between Li and liquid electrolytes can devastate battery performance; therefore, artificial SEIs (ASEIs) have been proposed as an effective strategy to replace native SEIs. Herein, as a collaboration between academia and industrial R&D teams, a multifunctional (crystalline, high modulus, and robust, Li + ion conductive, electrolyte-blocking, and solution processable) ASEI material, LiAl-FBD (where “FBD” refers to 2,2,3,3-tetra-fluoro-1,4-butanediol), for improving Li metal battery performance is designed and synthesized. The LiAl-FBD crystal structure consists of Al 3+ ions bridged by FBD 2– ligands to form anion clusters while Li + ions are loosely bound at the periphery, enabling an Li + ion conductivity of 9.4 × 10 –6 S cm –1 . Further, the fluorinated, short ligands endow LiAl-FBD with electrolyte phobicity and high modulus. The ASEI is found to prevent side reactions and extend the cycle life of Li metal electrodes. Specifically, pairing LiAl-FBD coated 50 μm thick Li with industrial 3.5 mAh cm –2 NMC811 cathode and 2.8 μL mAh –1 lean elec-trolyte, the Li metal full cells show superior cycle life compared to bare ones, achieving 250 cycles at 1 mA cm –2 .

25 ENERGY STORAGE↗

Single vs double atom catalyst for N 2 activation in nitrogen reduction reaction: A DFT perspective

Abstract Ammonia synthesis through electrochemical reduction of nitrogen molecules is a promising strategy to significantly reduce the energy consumption in traditional industrial process. Detailed mechanism study of multistep complex nitrogen reduction reaction is prerequisite for the design of highly efficient catalyst. Stable atomically dispersed catalyst with unique geometric and electronic structure is suitable for the mechanism clarification of such a complex reaction. In this study, d ‐block transition‐metal (TM) anchored C 2 N single layer catalyst is investigated by the density functional theory (DFT) calculation. Both single TM‐anchored single atom catalyst (SAC) and double TM‐anchored double atom catalyst (DAC) exhibit good thermodynamic stability in atomically dispersed catalyst. In the case of SACs, IVB metals (Ti, Zr, Hf) exhibit the highest reactivity and lowest overpotential. While in the case of DACs, Cr─Cr system leads to the NH 3 formation, but V─V system leads to the N 2 H 4 formation. The SACs show much lower overpotential and stronger activation of N 2 molecule than the DACs due to the different activation mechanisms: traditional σ ‐donation/ π ‐backdonation N 2 activation mechanism is found in SACs, while a new π ‐donation/π‐backdonation N 2 activation mechanism is found in the DACs. The present work demonstrates that the different catalytic effect for NRR between SAC and DAC and their corresponding electronic structure origin, which gives more insight into the single atom catalyst. image

Qian, Yumin↗

Discrete fracture network model benchmarks developed and applied in a DECOVALEX-2023 repository performance assessment study

This study presents newly developed benchmarks for modeling flow and transport within discrete fracture networks (DFNs) and useful methods for analyzing the results. The new benchmarks are designed to test modeling approaches for use in probabilistic performance assessment models of deep geologic repositories in fractured rock. The benchmarks simulate flow and transport through a 1 km 3 block of fractured rock. The first simulates migration of a short pulse of tracer through a simple network of four intersecting fractures. The second adds 1089 stochastically generated fractures. The third changes the pulse to a continuous point source. Evaluation of model performance relies on moment analysis and comparison of the results of different models. The expected nondimensional first moment of the conservative tracer for each benchmark is 1. The benchmarks were simulated by teams from Canada, Czechia, Germany, Korea, Sweden, Taiwan, and the United States as part of a DECOVALEX-2023 study (decovalex.org). The teams used various approaches, including explicit DFN modeling, DFN upscaling to an equivalent continuous porous medium (ECPM), and a combination of both methods. Transport mechanisms are modeled using either the advection-dispersion equation or particle tracking. Results demonstrate strong agreement among the models in breakthrough behavior up to the 75th percentile. Significant deviations in first moments and well-clustered outputs led to the identification of inaccuracies in several models. Such findings exemplify the benefit of exercising these benchmarks and using the presented methods to test DFN flow and transport models.

Benchmark↗

Testing and signal identification for two-sample high-dimensional covariances via multi-level thresholding

The paper considers testing and signal identification for covariance matrices from two populations of marginally sub-Gaussian distributed. A multi-level thresholding procedure is proposed for testing the equality of two high-dimensional covariance matrices, which is designed to detect sparse and faint differences between the covariances. A novel -statistic composition is developed to establish the asymptotic distribution of the thresholding statistics in conjunction with the matrix blocking and the coupling techniques. It is shown that the proposed test is more powerful than the existing tests in detecting sparse and weak signals in covariances. Multiple testing procedures are constructed to discover different covariances and the sub-groups of variables with different covariance structures between the two populations. The proposed procedures are based on the multi-level thresholding test, which are able to control the false discovery proportion () with high power. Simulation experiments and a case study on the returns of the S&P 500 stocks before and after the COVID-19 pandemic are conducted to demonstrate and compare the utilities of the proposed methods.

97 MATHEMATICS AND COMPUTING↗

Polydiallylammonium-Polysulfone Multiblock Copolymers for Moisture-Swing Direct Air Capture of Carbon Dioxide

Polysulfones are an important class of materials for a variety of applications due to their excellent strength and thermal stability. Quaternary ammonium polymers are also useful materials due to their ion exchange properties, and such materials with alkaline counterions have been used to capture carbon dioxide (CO 2 ) from ambient air through a moisture-driven mechanism. Herein, we design sulfone-based multiblock copolymers containing ammonium functionalities and demonstrate their potential for CO 2 capture. Specifically, a series of multiblock copolymers containing both polydiallyldimethylammonium (PDADMA) and polysulfone (PSf) blocks were synthesized. PSf blocks provide mechanical integrity, while the PDADMA blocks enable the direct air capture of CO 2 . Multiblock copolymers were synthesized at high yield, and PDADMA(OH)-PSf copolymer films with good flexibility and strength were formed with IECs of 424 μmol/g (17 mol % PDADMA(OH)), 1407 μmol/g (44 mol % PDADMA(OH)), and 1726 μmol/g (50 mol % PDADMA(OH)). Thermogravimetric analysis was used to determine that all films are thermally stable up to 345 °C. Differential scanning calorimetry revealed that the 17 and 44 mol % functionalized PDADMA(OH)-PSf multiblock copolymers showed one T g at 185 °C and 50 mol % PDADMA(OH)-PSf polymer showed a T g at 183 °C and a second T g at 16 °C. Atomic force microscopy showed that the multiblock polymers exhibit disordered phase separation. Furthermore, the copolymer materials displayed moisture-swing direct air capture of CO 2 , opening up pathways to utilize polymer architecture and chemistry to tailor the properties of promising sorbents.

42 ENGINEERING↗

Developmental assembly of multi-component polymer systems through interconnected synthetic gene networks in vitro

Abstract Living cells regulate the dynamics of developmental events through interconnected signaling systems that activate and deactivate inert precursors. This suggests that similarly, synthetic biomaterials could be designed to develop over time by using chemical reaction networks to regulate the availability of assembling components. Here we demonstrate how the sequential activation or deactivation of distinct DNA building blocks can be modularly coordinated to form distinct populations of self-assembling polymers using a transcriptional signaling cascade of synthetic genes. Our building blocks are DNA tiles that polymerize into nanotubes, and whose assembly can be controlled by RNA molecules produced by synthetic genes that target the tile interaction domains. To achieve different RNA production rates, we use a strategy based on promoter “nicking” and strand displacement. By changing the way the genes are cascaded and the RNA levels, we demonstrate that we can obtain spatially and temporally different outcomes in nanotube assembly, including random DNA polymers, block polymers, and as well as distinct autonomous formation and dissolution of distinct polymer populations. Our work demonstrates a way to construct autonomous supramolecular materials whose properties depend on the timing of molecular instructions for self-assembly, and can be immediately extended to a variety of other nucleic acid circuits and assemblies.

Science & Technology - Other Topics↗

Unified Communication Optimization Strategies for Sparse Triangular Solver on CPU and GPU Clusters

This paper presents a unified communication optimization framework for sparse triangular solve (SpTRSV) algorithms on CPU and GPU clusters. The framework builds upon a 3D communication-avoiding (CA) layout of Px × Py × Pz processes that divides a sparse matrix into Pz submatrices, each handled by a Px × Py 2D grid with block-cyclic distribution. We propose three communication optimization strategies: First, a new 3D SpTRSV algorithm is developed, which trades the inter-grid communication and synchronization with replicated computation. This design requires only one inter-grid synchronization, and the inter-grid communication is efficiently implemented with sparse allreduce operations. Second, broadcast and reduction communication trees are used to reduce message latency of the intra-grid 2D communication on CPU clusters. Finally, we leverage GPU-initiated one-sided communication to implement the communication trees on GPU clusters. With these nested inter- and intra-grid communication optimization strategies, the proposed 3D SpTRSV algorithm can attain up to 3.45x speedups compared to the baseline 3D SpTRSV algorithm using up to 2048 Cori Haswell CPU cores. In addition, the proposed GPU 3D SpTRSV algorithm can achieve up to 6.5x speedups compared to the proposed CPU 3D SpTRSV algorithm with Pz up to 64. Finally it is remarkable that the proposed GPU 3D SpTRSV can scale to 256 GPUs using the Perlmutter system while the existing 2D SpTRSV algorithm can only scale up to 4 GPUs.

Sao, Piyush↗