arXiv subjects

Andong Hu

Publications and source records attributed to Andong Hu.

10 recordsLinked to original sources

arXiv2026

Communication Offloading on SmartNIC DPUs: A Quantitative Approach

SmartNIC Data Processing Units (DPUs) offer a promising solution for saving high-end CPU resources by offloading tasks to programmable cores near the network interface. In this work, we explore the feasibility of SmartNIC DPUs in supporting an asynchronous communication model called "fire-and-forget", particularly its core message routing service. We design a communication offloading engine called Buddy that decouples communication tasks from the application process. Buddy runs flexibly on SmartNIC DPUs such as the Nvidia BlueField-3 DPU and generic x86 CPUs. Our evaluation results in five applications identify the memory-to-communication ratio as a key predictor of the offloading performance. Host-dominated workloads, such as Quicksilver and Sparse Matrix Transpose, achieved up to 1.55x speedup with communication offloaded to the DPU. We further identify a 625x increase in DRAM traffic due to the absence of Direct Cache Access support on the DPU, highlighting a critical need in future SmartNIC designs.

cs.DC↗

arXiv2026

BVH-Accelerated Ray Tracing for High-Frequency Electromagnetic Backscattering

As computational complexity in electromagnetics increases with frequency, full-wave solvers become computationally infeasible for electrically large problems. To address this limitation, we present a shooting and bouncing rays (SBR) method for efficiently modeling electromagnetic backscattering of metallic objects in the high-frequency regime. The method couples multi-reflection geometrical-optics ray transport with a physical optics surface integral discretized over ray tubes. To reduce the massive ray-surface intersection search space, we use a bounding volume hierarchy (BVH) and organize the computation as a trace-integrate pipeline. The ray tracing generates hit data, and the physical optics integral is evaluated over valid intersections only. Numerical accuracy is controlled through an incident-ray sampling rule that mitigates phase aliasing in the discretized physical optics integration. The method is accelerated on NVIDIA and AMD GPUs and parallelized with MPI. We validate against analytical Mie solutions for a perfectly electrically conducting (PEC) sphere and demonstrate applicability to a complex aircraft geometry for monostatic radar cross-section prediction.

cs.CE↗

arXiv2026

Accurate and Reliable Uncertainty Estimates for Deterministic Predictions Extensions to Under and Overpredictions

Computational models support high-stakes decisions across engineering and science, and practitioners increasingly seek probabilistic predictions to quantify uncertainty in such models. Existing approaches generate predictions either by sampling input parameter distributions or by augmenting deterministic outputs with uncertainty representations, including distribution-free and distributional methods. However, sampling-based methods are often computationally prohibitive for real-time applications, and many existing uncertainty representations either ignore input dependence or rely on restrictive Gaussian assumptions that fail to capture asymmetry and heavy-tailed behavior. Therefore, we extend the ACCurate and Reliable Uncertainty Estimate (ACCRUE) framework to learn input-dependent, non-Gaussian uncertainty distributions, specifically two-piece Gaussian and asymmetric Laplace forms, using a neural network trained with a loss function that balances predictive accuracy and reliability. Through synthetic and real-world experiments, we show that the proposed approach captures an input-dependent uncertainty structure and improves probabilistic forecasts relative to existing methods, while maintaining flexibility to model skewed and non-Gaussian errors.

cs.CE↗

arXiv2026

Making Room for AI: Multi-GPU Molecular Dynamics with Deep Potentials in GROMACS

GROMACS is a de-facto standard for classical Molecular Dynamics (MD). The rise of AI-driven interatomic potentials that pursue near-quantum accuracy at MD throughput now poses a significant challenge: embedding neural-network inference into multi-GPU simulations retaining high-performance. In this work, we integrate the MLIP framework DeePMD-kit into GROMACS, enabling domain-decomposed, GPU-accelerated inference across multi-node systems. We extend the GROMACS NNPot interface with a DeePMD backend, and we introduce a domain decomposition layer decoupled from the main simulation. The inference is executed concurrently on all processes, with two MPI collectives used each step to broadcast coordinates and to aggregate and redistribute forces. We train an in-house DPA-1 model (1.6 M parameters) on a dataset of solvated protein fragments. We validate the implementation on a small protein system, then we benchmark the GROMACS-DeePMD integration with a 15,668 atom protein on NVIDIA A100 and AMD MI250x GPUs up to 32 devices. Strong-scaling efficiency reaches 66% at 16 devices and 40% at 32; weak-scaling efficiency is 80% to 16 devices and reaches 48% (MI250x) and 40% (A100) at 32 devices. Profiling with the ROCm System profiler shows that >90% of the wall time is spent in DeePMD inference, while MPI collectives contribute <10%, primarily since they act as a global synchronization point. The principal bottlenecks are the irreducible ghost-atom cost set by the cutoff radius, confirmed by a simple throughput model, and load imbalance across ranks. These results demonstrate that production MD with near ab initio fidelity is feasible at scale in GROMACS.

cs.DC↗

arXiv2026

Enabling AI Deep Potentials for Ab Initio-quality Molecular Dynamics Simulations in GROMACS

State-of-the-art AI deep potentials provide ab initio-quality results, but at a fraction of the computational cost of first-principles quantum mechanical calculations, such as density functional theory. In this work, we bring AI deep potentials into GROMACS, a production-level Molecular Dynamics (MD) code, by integrating with DeePMD-kit that provides domain-specific deep learning (DL) models of interatomic potential energy and force fields. In particular, we enable AI deep potentials inference across multiple DP model families and DL backends by coupling GROMACS Neural Network Potentials with the C++/CUDA backend in DeePMD-kit. We evaluate two recent large-atom-model architectures, DPA2 that is based on the attention mechanism and DPA3 that is based on GNN, in GROMACS using four ab initio-quality protein-in-water benchmarks (1YRF, 1UBQ, 3LZM, 2PTC) on NVIDIA A100 and GH200 GPUs. Our results show that DPA2 delivers up to 4.23x and 3.18x higher throughput than DPA3 on A100 and GH200 GPUs, respectively. We also provide a characterization study to further contrast DPA2 and DPA3 in throughput, memory usage, and kernel-level execution on GPUs. Our findings identify kernel-launch overhead and domain-decomposed inference as the main optimization priorities for AI deep potentials in production MD simulations.

cs.DC↗

arXiv2025

Exascale Implicit Kinetic Plasma Simulations on El~Capitan for Solving the Micro-Macro Coupling in Magnetospheric Physics

Our fully kinetic, implicit Particle-in-Cell (PIC) simulations of global magnetospheres on up to 32,768 of El Capitan's AMD Instinct MI300A Accelerated Processing Units (APUs) represent an unprecedented computational capability that addresses a fundamental challenge in space physics: resolving the multi-scale coupling between microscopic (electron-scale) and macroscopic (global-scale) dynamics in planetary magnetospheres. The implicit scheme of iPIC3D supports time steps and grid spacing that are up to 10 times larger than those of explicit methods, without sacrificing physical accuracy. This enables the simulation of magnetospheres while preserving fine-scale electron physics, which is critical for key processes such as magnetic reconnection and plasma turbulence. Our algorithmic and technological innovations include GPU-optimized kernels, particle control, and physics-aware data compression using Gaussian Mixture Models. With simulation domains spanning 100-1,000 ion skin depths, we reach the global scale of small-to-medium planetary magnetospheres, such as those of Mercury and Ganymede, which supports fully kinetic treatment of global-scale dynamics in systems previously out of reach for fully kinetic PIC codes.

cs.CE↗

arXiv2025

Discovering Governing Equations of Geomagnetic Storm Dynamics with Symbolic Regression

Geomagnetic storms are large-scale disturbances of the Earth's magnetosphere driven by solar wind interactions, posing significant risks to space-based and ground-based infrastructure. The Disturbance Storm Time (Dst) index quantifies geomagnetic storm intensity by measuring global magnetic field variations. This study applies symbolic regression to derive data-driven equations describing the temporal evolution of the Dst index. We use historical data from the NASA OMNIweb database, including solar wind density, bulk velocity, convective electric field, dynamic pressure, and magnetic pressure. The PySR framework, an evolutionary algorithm-based symbolic regression library, is used to identify mathematical expressions linking dDst/dt to key solar wind. The resulting models include a hierarchy of complexity levels and enable a comparison with well-established empirical models such as the Burton-McPherron-Russell and O'Brien-McPherron models. The best-performing symbolic regression models demonstrate superior accuracy in most cases, particularly during moderate geomagnetic storms, while maintaining physical interpretability. Performance evaluation on historical storm events includes the 2003 Halloween Storm, the 2015 St. Patrick's Day Storm, and a 2017 moderate storm. The results provide interpretable, closed-form expressions that capture nonlinear dependencies and thresholding effects in Dst evolution.

cs.CE↗

arXiv2025

Physics-Aware Compression of Plasma Distribution Functions with GPU-Accelerated Gaussian Mixture Models

Data compression is a critical technology for large-scale plasma simulations. Storing complete particle information requires Terabyte-scale data storage, and analysis requires ad-hoc scalable post-processing tools. We propose a physics-aware in-situ compression method using Gaussian Mixture Models (GMMs) to approximate electron and ion velocity distribution functions with a number of Gaussian components. This GMM-based method allows us to capture plasma features such as mean velocity and temperature, and it enables us to identify heating processes and generate beams. We first construct a histogram to reduce computational overhead and apply GPU-accelerated, in-situ GMM fitting within iPIC3D, a large-scale implicit Particle-in-Cell simulator, ensuring real-time compression. The compressed representation is stored using the ADIOS 2 library, thus optimizing the I/O process. The GPU and histogramming implementation provides a significant speed-up with respect to GMM on particles (both in time and required memory at run-time), enabling real-time compression. Compared to algorithms like SZ, MGARD, and BLOSC2, our GMM-based method has a physics-based approach, retaining the physical interpretation of plasma phenomena such as beam formation, acceleration, and heating mechanisms. Our GMM algorithm achieves a compression ratio of up to $10^4$, requiring a processing time comparable to, or even lower than, standard compression engines.

cs.CE↗

arXiv2025

Apple vs. Oranges: Evaluating the Apple Silicon M-Series SoCs for HPC Performance and Efficiency

This paper investigates the architectural features and performance potential of the Apple Silicon M-Series SoCs (M1, M2, M3, and M4) for HPC. We provide a detailed review of the CPU and GPU designs, the unified memory architecture, and coprocessors such as Advanced Matrix Extensions (AMX). We design and develop benchmarks in the Metal Shading Language and Objective-C++ to assess FP32 computational and memory performance. We also measure power consumption and efficiency using Apple's powermetrics tool. Our results show that the M-Series chips offer up to 100 GB/s memory bandwidth, and significant generational improvements in computational performance, with up to 2.9 FP32 TFLOPS on the M4. Power consumption varies from a few Watts to 10-20 Watts, with more than 200 GFLOPS per Watt efficiency of GPU and accelerator reached by all four chips. Despite limitations in FP64 support on the GPU, the M-Series chips demonstrate strong potential for energy-efficient HPC applications. While existing HPC solutions such as the Nvidia Grace-Hopper superchip outperform Apple Silicon in both memory bandwidth and computational performance, we see that the M-Series provides a competitive power-efficient alternative to traditional HPC architectures and represents a distinct category altogether -- forming an apples-to-oranges comparison.

cs.AR↗

arXiv2021

A Machine-Learning-Ready Dataset Prepared from the Solar and Heliospheric Observatory Mission

We present a Python tool to generate a standard dataset from solar images that allows for user-defined selection criteria and a range of pre-processing steps. Our Python tool works with all image products from both the Solar and Heliospheric Observatory (SoHO) and Solar Dynamics Observatory (SDO) missions. We discuss a dataset produced from the SoHO mission's multi-spectral images which is free of missing or corrupt data as well as planetary transits in coronagraph images, and is temporally synced making it ready for input to a machine learning system. Machine-learning-ready images are a valuable resource for the community because they can be used, for example, for forecasting space weather parameters. We illustrate the use of this data with a 3-5 day-ahead forecast of the north-south component of the interplanetary magnetic field (IMF) observed at Lagrange point one (L1). For this use case, we apply a deep convolutional neural network (CNN) to a subset of the full SoHO dataset and compare with baseline results from a Gaussian Naive Bayes classifier.

astro-ph.SR↗