In the era of dramatic growth of AI, massive investments in large-scale data-driven AI systems demand highperformance computing, which in turn consume tremendous amounts of energy and resources. This trend raises new challenges in optimizing sustainability without sacrificing scalability or performance. Among the energy-efficient alternatives of the traditional Von Neumann architecture, neuromorphic computing and its Spiking Neural Networks (SNNs) are a promising choice due to their inherent energy efficiency. However, in some realworld application scenarios such as complex continuous control tasks, SNNs often lack the performance optimizations that traditional artificial neural networks have. Researchers have addressed this by combining SNNs with Deep Reinforcement Learning (DeepRL), yet scalability remains unexplored. In this paper, we extend our previous work on SpikeRL, which is a scalable and energy efficient framework for DeepRL-based SNNs for continuous control. In our initial implementation of SpikeRL framework, we depended on the population encoding from the Population-coded Spiking Actor Network (PopSAN) method for our SNN model and implemented distributed training with Message Passing Interface (MPI) through mpi4py. We also optimized our model training by using mixed-precision for parameter updates. In our new SpikeRL framework, we have implemented our own DeepRL-SNN component with population encoding, and distributed training with PyTorch Distributed package with NCCL backend while still optimizing with mixed precision training. Our new SpikeRL implementation is 4.26x faster and 2.25x more energy efficient than state-of-the-art DeepRL-SNN methods. Our proposed SpikeRL framework demonstrates a truly scalable and sustainable solution for complex continuous control tasks in real-world applications.

We introduce an algorithmic framework for performing QR factorization with column pivoting (QRCP) on general matrices. The framework enables the design of practical QRCP algorithms through user-controlled choices for the core subroutines. We provide a comprehensive overview of how to navigate these choices on modern hardware platforms, offering detailed descriptions of alternative methods for both CPUs and GPUs. The practical QRCP algorithms developed within this framework are implemented as part of the open-source RandLAPACK library. Our empirical evaluation demonstrates that, on a dual AMD EPYC 9734 system, the proposed method achieves performance improvements of up to two orders of magnitude over LAPACK's standard QRCP routine and greatly surpasses the performance of the current state-of-the-art randomized QRCP algorithm. Additionally, on an NVIDIA H100 GPU, our method attains approximately 65 percent of the performance of cuSOLVER's unpivoted QR factorization.

SLATE (Software for Linear Algebra Targeting Exascale) is a distributed, dense linear algebra library targeting both CPU-only and GPU-accelerated systems, developed over the course of the Exascale Computing Project (ECP). While it began with several documents setting out its initial design, significant design changes occurred throughout its development. In some cases, these were anticipated: an early version used a simple consistency flag that was later replaced with a full-featured consistency protocol. In other cases, performance limitations and software and hardware changes prompted a redesign. Sequential communication tasks were parallelized; host-to-host MPI calls were replaced with GPU device-to-device MPI calls; more advanced algorithms such as Communication Avoiding LU and the Random Butterfly Transform (RBT) were introduced. Early choices that turned out to be cumbersome, error prone, or inflexible have been replaced with simpler, more intuitive, or more flexible designs. Applications have been a driving force, prompting a lighter weight queue class, nonuniform tile sizes, and more flexible MPI process grids. Of paramount importance has been building a portable library that works across several different GPU architectures - AMD, Intel, and NVIDIA - while keeping a clean and maintainable codebase. Here we explore the evolving design choices and their effects, both in terms of performance and software sustainability.

By some metrics, spiking neural networks (SNNs) are still far from performing equally well as some of the other artificial neural networks for continuous control tasks. Even though SNNs are efficient by their design and structure, they lack many of the optimizations known from deep reinforcement learning (DeepRL) algorithms. Hence, researchers have combined SNNs with DeepRL algorithms to tackle this challenge. However, the question remains as to how scalable these DeepRL-based SNNs are in practice. In this paper, we introduce SpikeRL, a scalable and efficient framework for DeepRL-based SNNs for continuous control. In the SpikeRL framework, we first incor-porate population encoding from the Population-coded Spiking Actor Network (PopSAN) method for our SNN model. Then, we implement Message Passing Interface (MPI) through its widely used Python bindings in mpi4py in order to achieve distributed training across models and environments. We further optimize our model training by using mixed-precision for parameter updates. Our research findings demonstrate a scalability and efficiency potential for spiking reinforcement learning methods for continuous control environments.

The standardization of an interface for dense linear algebra operations in the BLAS standard has enabled interoperability between different linear algebra libraries, thereby boosting the success of scientific computing, in particular in scientific HPC. Despite numerous efforts in the past, the community has not yet agreed on a standardization for sparse linear algebra operations due to numerous reasons. One is the fact that sparse linear algebra objects allow for many different storage formats, and different hardware may favor different storage formats. This makes the definition of a FORTRAN-style all-circumventing interface extremely challenging. Another reason is that opposed to dense linear algebra functionality, in sparse linear algebra, the size of the sparse data structure for the operation result is not always known prior to the information. Furthermore, as opposed to the standardization effort for dense linear algebra, we are late in the technology readiness cycle, and many production-ready software libraries using sparse linear algebra routines have implemented and committed to their own sparse BLAS interface. At the same time, there exists a demand for standardization that would improve interoperability, and sustainability, and allow for easier integration of building blocks. In an inclusive, cross-institutional effort involving numerous academic institutions, US National Labs, and industry, we spent two years designing a hardware-portable interface for basic sparse linear algebra functionality that serves the user needs and is compatible with the different interfaces currently used by different vendors. In this paper, we present a C++ API for sparse linear algebra functionality, discuss the design choices, and detail how software developers preserve a lot of freedom in terms of how to implement functionality behind this API.

We investigate a new task-based implementation of the polar decomposition on massively parallel systems augmented with multiple GPUs using SLATE. We implement the iterative QR Dynamically-Weighted Halley (QDWH) algorithm, whose building blocks mainly consist of compute-bound matrix operations, allowing for high levels of parallelism to be exploited on various hardware architectures, such as NVIDIA, AMD, and Intel GPU-based systems. To achieve both performance and portability, we implement our QDWH-based polar decomposition in the SLATE library, which uses efficient techniques in dense linear algebra, such as 2D block cyclic data distribution and communication-avoiding algorithms, as well as modern parallel programming approaches, such as dynamic scheduling and communication overlapping, and uses OpenMP tasks to track data dependencies. We report numerical accuracy and performance results. The benchmarking campaign reveals up to an 18-fold performance speedup of the GPU accelerated implementation compared to the existing state-of-the-art implementation for the polar decomposition.

The solution of linear least-squares problems is at the heart of many scientific and engineering applications. While any method able to minimize the backward error of such problems is considered numerically stable, the theory states that the forward error depends on the condition number of the matrix in the system of equations. On the one hand, the QR factorization is an efficient method to solve such problems, but the solutions it produces may have large forward errors when the matrix is rank deficient. On the other hand, rank-revealing QR (RRQR) is able to produce smaller forward errors on rank deficient matrices, but its cost is prohibitive compared to QR due to memory-inefficient operations. The aim of this paper is to propose PAQR for the solution of rank-deficient linear least-squares problems as an alternative solution method. It has the same (or smaller) cost as QR and is as accurate as QR with column pivoting in many practical cases. In addition to presenting the algorithm and its implementations on different hardware architectures, we compare its accuracy and performance results on a variety of application-derived problems.

LU factorization is a key approach for solving large, dense systems of linear equations. Partial row pivoting is commonly used to ensure numerical stability; however, the data movement needed for the row interchanges can reduce performance. To improve this, we propose using threshold pivoting to find pivots almost as good as those selected by partial pivoting but that result in less data movement. Our theoretical analysis bounds the element growth similarly to partial pivoting; however, it also shows that the growth of threshold pivoting for a given matrix cannot be bounded by that of partial pivoting and vice versa. Additionally, we experimentally tested the approach on the Summit supercomputer. Threshold pivoting improved performance by up to 32% without a significant effect on accuracy. For a more aggressive configuration with up to one digit of accuracy lost, the improvement was as high as 44%.

The SLATE project is implementing a distributed dense linear algebra library for highly-scalable distributed-memory accelerator-based computer systems. The goal is to provide a library that can be easily ported to different hardware (CPUs, GPUs, accelerators) and will provide high performance for machines into the future. Current ports include CPUs, CUDA, ROCm, and oneAPI. We achieve both performance and portability by leveraging several layers and abstractions, including OpenMP tasks to track data dependencies, MPI for distributed communication, and the BLAS++ and LAPACK++ libraries developed as a portable layer across vendor-optimized CPU and GPU BLAS and LAPACK functionality. We rely on the C++ standard library and templating to reduce code duplication for better maintainability. The few kernels not present in BLAS are implemented in CUDA, HIP, and OpenMP target offload, and are easily ported to new platforms.

Numerical exceptions, which may be caused by overflow, operations like division by 0 or sqrt(-1), or convergence failures, are unavoidable in many cases, in particular when software is used on unforeseen and difficult inputs. As more aspects of society become automated e.g., self-driving cars, health monitors, and cyber-physical systems more generally, it is becoming increasingly important to design software that is resilient to exceptions, and that responds to them in a consistent way. Consistency is needed to allow users to build higher-level software that is also resilient and consistent (and so on recursively). In this paper we explore the design space of consistent exception handling for the widely used BLAS and LAPACK linear algebra libraries, pointing out a variety of instances of inconsistent exception handling in the current versions, and propose a new design that balances consistency, complexity, ease of use, and performance. Some compromises are needed, because there are preexisting inconsistencies that are outside our control, including in or between existing vendor BLAS implementations, different programming languages, and even compilers for the same programming language. And user requests from our surveys are quite diverse. We also propose our design as a possible model for other numerical software, and welcome comments on our design choices.

Task graphs have been studied for decades as a foundation for scheduling irregular parallel applications and incorporated in many programming models including OpenMP. While many high-performance parallel libraries are based on task graphs, they also have additional scheduling requirements, such as synchronization within inner levels of data parallelism and internal blocking communications. In this paper, we extend task-graph scheduling to support efficient synchronization and communication within tasks. Compared to past work, our scheduler avoids deadlock and oversubscription of worker threads, and refines victim selection to increase the overlap of sibling tasks. To the best of our knowledge, our approach is the first to combine gang-scheduling and work-stealing in a single runtime. Our approach has been evaluated on the SLATE high-performance linear algebra library. Relative to the LLVM OMP runtime, our runtime demonstrates performance improvements of up to 13.82%, 15.2%, and 36.94% for LU, QR, and Cholesky, respectively, evaluated across different configurations related to matrix size, number of nodes, and use of CPUs vs GPUs.

The efficient utilization of mixed-precision numerical linear algebra algorithms can offer attractive acceleration to scientific computing applications. Especially with the hardware integration of low-precision special-function units designed for machine learning applications, the traditional numerical algorithms community urgently needs to reconsider the floating point formats used in the distinct operations to efficiently leverage the available compute power. In this work, we provide a comprehensive survey of mixed-precision numerical linear algebra routines, including the underlying concepts, theoretical background, and experimental results for both dense and sparse linear algebra problems.

This article describes a standard API for a set of Batched Basic Linear Algebra Subprograms (Batched BLAS or BBLAS). The focus is on many independent BLAS operations on small matrices that are grouped together and processed by a single routine, called a Batched BLAS routine. The matrices are grouped together in uniformly sized groups, with just one group if all the matrices are of equal size. The aim is to provide more efficient, but portable, implementations of algorithms on high-performance many-core platforms. These include multicore and many-core CPU processors, GPUs and coprocessors, and other hardware accelerators with floating-point compute facility. As well as the standard types of single and double precision, we also include half and quadruple precision in the standard. In particular, half precision is used in many very large scale applications, such as those associated with machine learning.

Over the last year, the ECP xSDK-multiprecision effort has made tremendous progress in developing and deploying new mixed precision technology and customizing the algorithms for the hardware deployed in the ECP flagship supercomputers. The effort also has succeeded in creating a cross-laboratory community of scientists interested in mixed precision technology and now working together in deploying this technology for ECP applications. In this report, we highlight some of the most promising and impactful achievements of the last year. Among the highlights we present are: Mixed precision IR using a dense LU factorization and achieving a 1.8× speedup on Spock; results and strategies for mixed precision IR using a sparse LU factorization; a mixed precision eigenvalue solver; Mixed Precision GMRES-IR being deployed in Trilinos, and achieving a speedup of 1.4× over standard GMRES; compressed Basis (CB) GMRES being deployed in Ginkgo and achieving an average 1.4× speedup over standard GMRES; preparing hypre for mixed precision execution; mixed precision sparse approximate inverse preconditioners achieving an average speedup of 1.2×; and detailed description of the memory accessor separating the arithmetic precision from the memory precision, and enabling memory-bound low precision BLAS 1/2 operations to increase the accuracy by using high precision in the computations without degrading the performance. We emphasize that many of the highlights presented here have also been submitted to peer-reviewed journals or established conferences, and are under peer-review or have already been published.

Within the past years, hardware vendors have started designing low precision special function units in response to the demand of the Machine Learning community and their demand for high compute power in low precision formats. Also the server-line products are increasingly featuring low-precision special function units, such as the NVIDIA tensor cores in ORNL's Summit supercomputer providing more than an order of magnitude higher performance than what is available in IEEE double precision. At the same time, the gap between the compute power on the one hand and the memory bandwidth on the other hand keeps increasing, making data access and communication prohibitively expensive compared to arithmetic operations. To start the multiprecision focus effort, we survey the numerical linear algebra community and summarize all existing multiprecision knowledge, expertise, and software capabilities in this landscape analysis report. We also include current efforts and preliminary results that may not yet be considered "mature technology," but have the potential to grow into production quality within the multiprecision focus effort. As we expect the reader to be familiar with the basics of numerical linear algebra, we refrain from providing a detailed background on the algorithms themselves but focus on how mixed- and multiprecision technology can help improving the performance of these methods and present highlights of application significantly outperforming the traditional fixed precision methods.

Each successive generation of computer architecture has brought new challenges to achieving high performance mathematical solvers, necessitating development and analysis of new algorithms, which are then embodied in software libraries. These libraries hide architectural details from applications, allowing them to achieve a level of portability across platforms from desktops to world-class high performance computing (HPC) systems. Thus there has been an informal translational computer science process of developing algorithms and distributing them in open source software libraries for adoption by applications and vendors. With the move to exascale, increasing intentionality about this process will benefit the long-term sustainability of the scientific software stack.

With the acquisition and widespread use of more resources that rely on accelerator/wide vector–based computing, there has been a strong demand for science and engineering applications to take advantage of these latest assets. This, however, has been extremely challenging due to the diversity of systems to support their extreme concurrency, complex memory hierarchies, costly data movement, and heterogeneous node architectures. To address these challenges, we design a programming model and describe its ease of use in the development of a new MAGMA Templates library that delivers high-performance scalable linear algebra portable on current and emerging architectures. MAGMA Templates derives its performance and portability by (1) building on existing state-of-the-art linear algebra libraries, like MAGMA, SLATE, Trilinos, and vendor-optimized math libraries, and (2) providing access (seamlessly to the users) to the latest algorithms and architecture-specific optimizations through a single, easy-to-use C++-based API.