arXiv subjects

Dong Dai

Publications and source records attributed to Dong Dai.

At least 19 recordsLinked to original sources

arXiv2026

QoSFlow: Ensuring Service Quality of Distributed Workflows Using Interpretable Sensitivity Models

With the increasing importance of distributed scientific workflows, there is a critical need to ensure Quality of Service (QoS) constraints, such as minimizing time or limiting execution to resource subsets. However, the unpredictable nature of workflow behavior, even with similar configurations, makes it difficult to provide QoS guarantees. For effective reasoning about QoS scheduling, we introduce QoSFlow, a performance modeling method that partitions a workflow's execution configuration space into regions with similar behavior. Each region groups configurations with comparable execution times according to a given statistical sensitivity, enabling efficient QoS-driven scheduling through analytical reasoning rather than exhaustive testing. Evaluation on three diverse workflows shows that QoSFlow's execution recommendations outperform the best-performing standard heuristic by 27.38%. Empirical validation confirms that QoSFlow's recommended configurations consistently match measured execution outcomes across different QoS constraints.

cs.DC↗

arXiv2026

STELLAR: Storage Tuning Engine Leveraging LLM Autonomous Reasoning for High Performance Parallel File Systems

I/O performance is crucial to efficiency in data-intensive scientific computing; but tuning large-scale storage systems is complex, costly, and notoriously manpower-intensive, making it inaccessible for most domain scientists. To address this problem, we propose STELLAR, an autonomous tuner for high-performance parallel file systems. Our evaluations show that STELLAR almost always selects near-optimal parameter configurations for parallel file systems within the first five attempts, even for previously unseen applications. STELLAR differs fundamentally from traditional autotuning methods, which often require hundreds of thousands of iterations to converge. Powered by large language models (LLMs), STELLAR enables autonomous end-to-end agentic tuning by (1) accurately extracting tunable parameters from software manuals, (2) analyzing I/O trace logs generated by applications, (3) selecting initial tuning strategies, (4) rerunning applications on real systems and collecting I/O performance feedback, (5) adjusting tuning strategies and repeating the tuning cycle, and (6) reflecting on and summarizing tuning experiences into reusable knowledge for future optimizations. STELLAR integrates retrieval-augmented generation (RAG), tool execution, LLM-based reasoning, and a multiagent design to stabilize reasoning and combat hallucinations. We evaluate the impact of each component on optimization outcomes, providing design insights for similar systems in other optimization domains. STELLAR's architecture and empirical results highlight a promising approach to complex system optimization, especially for problems with large search spaces and high exploration costs, while making I/O tuning more accessible to domain scientists with minimal added resources.

cs.DC↗

arXiv2026

CARAT: Client-Side Adaptive RPC and Cache Co-Tuning for Parallel File Systems

Tuning parallel file system in High-Performance Computing (HPC) systems remains challenging due to the complex I/O paths, diverse I/O patterns, and dynamic system conditions. While existing autotuning frameworks have shown promising results in tuning PFS parameters based on applications' I/O patterns, they lack scalability, adaptivity, and the ability to operate online. In this work, focusing on scalable online tuning, we present CARAT, an ML-guided framework to co-tune client-side RPC and caching parameters of PFS, leveraging only locally observable metrics. Unlike global or pattern-dependent approaches, CARAT enables each client to make independent and intelligent tuning decisions online, responding to real-time changes in both application I/O behaviors and system states. We then prototyped CARAT using Lustre and evaluated it extensively across dynamic I/O patterns, real-world HPC workloads, and multi-client deployments. The results demonstrated that CARAT can achieve up to 3x performance improvement over the default or static configurations, validating the effectiveness and generality of our approach. Due to its scalability and lightweight, we believe CARAT has the potential to be widely deployed into existing PFS and benefit various data-intensive applications.

cs.DC↗

arXiv2026

DIAL: Decentralized I/O AutoTuning via Learned Client-side Local Metrics for Parallel File System

Enabling efficient, high-performance data access in parallel file systems (PFS) is critical for today's high-performance computing systems. PFS client-side I/O heavily impacts the final I/O performance delivered to individual applications and the entire system. Autotuning the key client-side I/O behaviors has been extensively studied and shows promising results. However, existing work has heavily relied on extensive number of global runtime metrics to monitor and accurate modeling of applications' I/O patterns. Such heavy overheads significantly limit the ability to enable fine-grained, dynamic tuning in practical systems. In this study, we propose DIAL (Decentralized I/O AutoTuning via Learned Client-side Local Metrics) which takes a drastically different approach. Instead of trying to extract the global I/O patterns of applications, DIAL takes a decentralized approach, treating each I/O client as an independent unit and tuning configurations using only its locally observable metrics. With the help of machine learning models, DIAL enables multiple tunable units to make independent but collective decisions, reacting to what is happening in the global storage systems in a timely manner and achieving better I/O performance globally for the application.

cs.DC↗

arXiv2026

IOAgent: Democratizing Trustworthy HPC I/O Performance Diagnosis Capability via LLMs

As the complexity of the HPC storage stack rapidly grows, domain scientists face increasing challenges in effectively utilizing HPC storage systems to achieve their desired I/O performance. To identify and address I/O issues, scientists largely rely on I/O experts to analyze their I/O traces and provide insights into potential problems. However, with a limited number of I/O experts and the growing demand for data-intensive applications, inaccessibility has become a major bottleneck, hindering scientists from maximizing their productivity. Rapid advances in LLMs make it possible to build an automated tool that brings trustworthy I/O performance diagnosis to domain scientists. However, key challenges remain, such as the inability to handle long context windows, a lack of accurate domain knowledge about HPC I/O, and the generation of hallucinations during complex interactions. In this work, we propose IOAgent as a systematic effort to address these challenges. IOAgent integrates a module-based pre-processor, a RAG-based domain knowledge integrator, and a tree-based merger to accurately diagnose I/O issues from a given Darshan trace file. Similar to an I/O expert, IOAgent provides detailed justifications and references for its diagnoses and offers an interactive interface for scientists to ask targeted follow-up questions. To evaluate IOAgent, we collected a diverse set of labeled job traces and released the first open diagnosis test suite, TraceBench. Using this test suite, we conducted extensive evaluations, demonstrating that IOAgent matches or outperforms state-of-the-art I/O diagnosis tools with accurate and useful diagnosis results. We also show that IOAgent is not tied to specific LLMs, performing similarly well with both proprietary and open-source LLMs. We believe IOAgent has the potential to become a powerful tool for scientists navigating complex HPC I/O subsystems in the future.

cs.DC↗

arXiv2026

AdapTBF: Decentralized Bandwidth Control via Adaptive Token Borrowing for HPC Storage

Modern high-performance computing (HPC) applications run on compute resources but share global storage systems. This design can cause problems when applications consume a disproportionate amount of storage bandwidth relative to their allocated compute resources. For example, an application running on a single compute node can issue many small, random writes and consume excessive I/O bandwidth from a storage server. This can hinder larger jobs that write to the same storage server and are allocated many compute nodes, resulting in significant resource waste. A straightforward solution is to limit each application's I/O bandwidth on storage servers in proportion to its allocated compute resources. This approach has been implemented in parallel file systems using Token Bucket Filter (TBF). However, strict proportional limits often reduce overall I/O efficiency because HPC applications generate short, bursty I/O. Limiting bandwidth can waste server capacity when applications are idle or prevent applications from temporarily using higher bandwidth during bursty phases. We argue that I/O control should maximize per-application performance and overall storage efficiency while ensuring fairness (e.g., preventing small jobs from blocking large-scale ones). We propose AdapTBF, which builds on TBF in modern parallel file systems (e.g., Lustre) and introduces a decentralized bandwidth control approach using adaptive borrowing and lending. We detail the algorithm, implement AdapTBF in Lustre, and evaluate it using synthetic workloads modeled after real-world scenarios. Results show that AdapTBF manages I/O bandwidth effectively while maintaining high storage utilization, even under extreme conditions.

cs.DC↗

arXiv2026

LLMTailor: A Layer-wise Tailoring Tool for Efficient Checkpointing of Large Language Models

Checkpointing is essential for fault tolerance in training large language models (LLMs). However, existing methods, regardless of their I/O strategies, periodically store the entire model and optimizer states, incurring substantial storage overhead and resource contention. Recent studies reveal that updates across LLM layers are highly non-uniform. Across training steps, some layers may undergo more significant changes, while others remain relatively stable or even unchanged. This suggests that selectively checkpointing only layers with significant updates could reduce overhead without harming training. Implementing such selective strategies requires fine-grained control over both weights and optimizer states, which no current tool provides. To address this gap, we propose \texttt{LLMTailor}, a checkpoint-merging framework that filters and assembles layers from different checkpoints to form a composite checkpoint. Our evaluation indicates that LLMTailor can work with different selective checkpointing strategies and effectively reduce checkpoint size (e.g., 4.3 times smaller for Llama3.1-8B) and checkpoint time (e.g., 2.8 times faster for Qwen2.5-7B) while maintaining model quality.

cs.DC↗

arXiv2025

DeepCQ: General-Purpose Deep-Surrogate Framework for Lossy Compression Quality Prediction

Error-bounded lossy compression techniques have become vital for scientific data management and analytics, given the ever-increasing volume of data generated by modern scientific simulations and instruments. Nevertheless, assessing data quality post-compression remains computationally expensive due to the intensive nature of metric calculations. In this work, we present a general-purpose deep-surrogate framework for lossy compression quality prediction (DeepCQ), with the following key contributions: 1) We develop a surrogate model for compression quality prediction that is generalizable to different error-bounded lossy compressors, quality metrics, and input datasets; 2) We adopt a novel two-stage design that decouples the computationally expensive feature-extraction stage from the light-weight metrics prediction, enabling efficient training and modular inference; 3) We optimize the model performance on time-evolving data using a mixture-of-experts design. Such a design enhances the robustness when predicting across simulation timesteps, especially when the training and test data exhibit significant variation. We validate the effectiveness of DeepCQ on four real-world scientific applications. Our results highlight the framework's exceptional predictive accuracy, with prediction errors generally under 10\% across most settings, significantly outperforming existing methods. Our framework empowers scientific users to make informed decisions about data compression based on their preferred data quality, thereby significantly reducing I/O and computational overhead in scientific data analysis.

cs.LG↗

arXiv2025

Improving SpGEMM Performance Through Matrix Reordering and Cluster-wise Computation

Sparse matrix-sparse matrix multiplication (SpGEMM) is a key kernel in many scientific applications and graph workloads. Unfortunately, SpGEMM is bottlenecked by data movement due to its irregular memory access patterns. Significant work has been devoted to developing row reordering schemes towards improving locality in sparse operations, but prior studies mostly focus on the case of sparse-matrix vector multiplication (SpMV). In this paper, we address these issues with hierarchical clustering for SpGEMM that leverages both row reordering and cluster-wise computation to improve reuse in the second input (B) matrix with a novel row-clustered matrix format and access pattern in the first input (A) matrix. We find that hierarchical clustering can speed up SpGEMM by 1.39x on average with low preprocessing cost (less than 20x the cost of a single SpGEMM on about 90% of inputs). Furthermore, we decouple the reordering algorithm from the clustered matrix format so they can be applied as independent optimizations. Additionally, this paper sheds light on the role of both row reordering and clustering independently and together for SpGEMM with a comprehensive empirical study of the effect of 10 different reordering algorithms and 3 clustering schemes on SpGEMM performance on a suite of 110 matrices. We find that reordering based on graph partitioning provides better SpGEMM performance than existing alternatives at the cost of high preprocessing time. The evaluation demonstrates that the proposed hierarchical clustering method achieves greater average speedup compared to other reordering schemes with similar preprocessing times.

cs.DC↗

arXiv2025

AI-Assisted Rapid Crystal Structure Generation Towards a Target Local Environment

In the field of material design, traditional crystal structure prediction approaches require extensive structural sampling through computationally expensive energy minimization methods using either force fields or quantum mechanical simulations. While emerging artificial intelligence (AI) generative models have shown great promise in generating realistic crystal structures more rapidly, most existing models fail to account for the unique symmetries and periodicity of crystalline materials, and they are limited to handling structures with only a few tens of atoms per unit cell. Here, we present a symmetry-informed AI generative approach called Local Environment Geometry-Oriented Crystal Generator (LEGO-xtal) that overcomes these limitations. Our method generates initial structures using AI models trained on an augmented small dataset, and then optimizes them using machine learning structure descriptors rather than traditional energy-based optimization. We demonstrate the effectiveness of LEGO-xtal by expanding from 25 known low-energy sp2 carbon allotropes to over 1,700, all within 0.5 eV/atom of the ground-state energy of graphite. This framework offers a generalizable strategy for the targeted design of materials with modular building blocks, such as metal-organic frameworks and next-generation battery materials.

cond-mat.mtrl-sci↗

arXiv2025

TSUE: A Two-Stage Data Update Method for an Erasure Coded Cluster File System

Compared to replication-based storage systems, erasure-coded storage incurs significantly higher overhead during data updates. To address this issue, various parity logging methods have been pro- posed. Nevertheless, due to the long update path and substantial amount of random I/O involved in erasure code update processes, the resulting long latency and low throughput often fail to meet the requirements of high performance applications. To this end, we propose a two-stage data update method called TSUE. TSUE divides the update process into a synchronous stage that records updates in a data log, and an asynchronous stage that recycles the log in real-time. TSUE effectively reduces update latency by transforming random I/O into sequential I/O, and it significantly reduces recycle overhead by utilizing a three-layer log and the spatio-temporal locality of access patterns. In SSDs cluster, TSUE significantly im- proves update performance, achieving improvements of 7.6X under Ali-Cloud trace, 5X under Ten-Cloud trace, while it also extends the SSD's lifespan by up to 13X through reducing the frequencies of reads/writes and of erase operations.

cs.DC↗

arXiv2025

Cross-Modal Interactive Perception Network with Mamba for Lung Tumor Segmentation in PET-CT Images

Lung cancer is a leading cause of cancer-related deaths globally. PET-CT is crucial for imaging lung tumors, providing essential metabolic and anatomical information, while it faces challenges such as poor image quality, motion artifacts, and complex tumor morphology. Deep learning-based models are expected to address these problems, however, existing small-scale and private datasets limit significant performance improvements for these methods. Hence, we introduce a large-scale PET-CT lung tumor segmentation dataset, termed PCLT20K, which comprises 21,930 pairs of PET-CT images from 605 patients. Furthermore, we propose a cross-modal interactive perception network with Mamba (CIPA) for lung tumor segmentation in PET-CT images. Specifically, we design a channel-wise rectification module (CRM) that implements a channel state space block across multi-modal features to learn correlated representations and helps filter out modality-specific noise. A dynamic cross-modality interaction module (DCIM) is designed to effectively integrate position and context information, which employs PET images to learn regional position information and serves as a bridge to assist in modeling the relationships between local features of CT images. Extensive experiments on a comprehensive benchmark demonstrate the effectiveness of our CIPA compared to the current state-of-the-art segmentation methods. We hope our research can provide more exploration opportunities for medical image segmentation. The dataset and code are available at https://github.com/mj129/CIPA.

eess.IV↗

arXiv2024

Cross-modal Medical Image Generation Based on Pyramid Convolutional Attention Network

The integration of multimodal medical imaging can provide complementary and comprehensive information for the diagnosis of Alzheimer's disease (AD). However, in clinical practice, since positron emission tomography (PET) is often missing, multimodal images might be incomplete. To address this problem, we propose a method that can efficiently utilize structural magnetic resonance imaging (sMRI) image information to generate high-quality PET images. Our generation model efficiently utilizes pyramid convolution combined with channel attention mechanism to extract multi-scale local features in sMRI, and injects global correlation information into these features using self-attention mechanism to ensure the restoration of the generated PET image on local texture and global structure. Additionally, we introduce additional loss functions to guide the generation model in producing higher-quality PET images. Through experiments conducted on publicly available ADNI databases, the generated images outperform previous research methods in various performance indicators (average absolute error: 0.0194, peak signal-to-noise ratio: 29.65, structural similarity: 0.9486) and are close to real images. In promoting AD diagnosis, the generated images combined with their corresponding sMRI also showed excellent performance in AD diagnosis tasks (classification accuracy: 94.21 %), and outperformed previous research methods of the same type. The experimental results demonstrate that our method outperforms other competing methods in quantitative metrics, qualitative visualization, and evaluation criteria.

cs.CE↗

arXiv2024

A Reinforcement Learning Based Backfilling Strategy for HPC Batch Jobs

High Performance Computing (HPC) systems are used across a wide range of disciplines for both large and complex computations. HPC systems often receive many thousands of computational tasks at a time, colloquially referred to as jobs. These jobs must then be scheduled as optimally as possible so they can be completed within a reasonable timeframe. HPC scheduling systems often employ a technique called backfilling, wherein low-priority jobs are scheduled earlier to use the available resources that are waiting for the pending high-priority jobs. To make it work, backfilling largely relies on job runtime to calculate the start time of the ready-to-schedule jobs and avoid delaying them. It is a common belief that better estimations of job runtime will lead to better backfilling and more effective scheduling. However, our experiments show a different conclusion: there is a missing trade-off between prediction accuracy and backfilling opportunities. To learn how to achieve the best trade-off, we believe reinforcement learning (RL) can be effectively leveraged. Reinforcement Learning relies on an agent which makes decisions from observing the environment, and gains rewards or punishments based on the quality of its decision-making. Based on this idea, we designed RLBackfilling, a reinforcement learning-based backfilling algorithm. We show how RLBackfilling can learn effective backfilling strategies via trial-and-error on existing job traces. Our evaluation results show up to 59% better scheduling performance (based on average bounded job slowdown) compared to EASY backfilling using user-provided job runtime and 30% better performance compared with EASY using the ideal predicted job runtime (the actual job runtime).

cs.DC↗

arXiv2024

DGAP: Efficient Dynamic Graph Analysis on Persistent Memory

Dynamic graphs, featuring continuously updated vertices and edges, have grown in importance for numerous real-world applications. To accommodate this, graph frameworks, particularly their internal data structures, must support both persistent graph updates and rapid graph analysis simultaneously, leading to complex designs to orchestrate `fast but volatile' and `persistent but slow' storage devices. Emerging persistent memory technologies, such as Optane DCPMM, offer a promising alternative to simplify the designs by providing data persistence, low latency, and high IOPS together. In light of this, we propose DGAP, a framework for efficient dynamic graph analysis on persistent memory. Unlike traditional dynamic graph frameworks, which combine multiple graph data structures (e.g., edge list or adjacency list) to achieve the required performance, DGAP utilizes a single mutable Compressed Sparse Row (CSR) graph structure with new designs for persistent memory to construct the framework. Specifically, DGAP introduces a \textit{per-section edge log} to reduce write amplification on persistent memory; a \textit{per-thread undo log} to enable high-performance, crash-consistent rebalancing operations; and a data placement schema to minimize in-place updates on persistent memory. Our extensive evaluation results demonstrate that DGAP can achieve up to $3.2\times$ better graph update performance and up to $3.77\times$ better graph analysis performance compared to state-of-the-art dynamic graph frameworks for persistent memory, such as XPGraph, LLAMA, and GraphOne.

cs.DS↗

arXiv2023

PROV-IO+: A Cross-Platform Provenance Framework for Scientific Data on HPC Systems

Data provenance, or data lineage, describes the life cycle of data. In scientific workflows on HPC systems, scientists often seek diverse provenance (e.g., origins of data products, usage patterns of datasets). Unfortunately, existing provenance solutions cannot address the challenges due to their incompatible provenance models and/or system implementations. In this paper, we analyze four representative scientific workflows in collaboration with the domain scientists to identify concrete provenance needs. Based on the first-hand analysis, we propose a provenance framework called PROV-IO+, which includes an I/O-centric provenance model for describing scientific data and the associated I/O operations and environments precisely. Moreover, we build a prototype of PROV-IO+ to enable end-to-end provenance support on real HPC systems with little manual effort. The PROV-IO+ framework can support both containerized and non-containerized workflows on different HPC platforms with flexibility in selecting various classes of provenance. Our experiments with realistic workflows show that PROV-IO+ can address the provenance needs of the domain scientists effectively with reasonable performance (e.g., less than 3.5% tracking overhead for most experiments). Moreover, PROV-IO+ outperforms a state-of-the-art system (i.e., ProvLake) in our experiments.

cs.DC↗

arXiv2023

ClusterLog: Clustering Logs for Effective Log-based Anomaly Detection

With the increasing prevalence of scalable file systems in the context of High Performance Computing (HPC), the importance of accurate anomaly detection on runtime logs is increasing. But as it currently stands, many state-of-the-art methods for log-based anomaly detection, such as DeepLog, have encountered numerous challenges when applied to logs from many parallel file systems (PFSes), often due to their irregularity and ambiguity in time-based log sequences. To circumvent these problems, this study proposes ClusterLog, a log pre-processing method that clusters the temporal sequence of log keys based on their semantic similarity. By grouping semantically and sentimentally similar logs, this approach aims to represent log sequences with the smallest amount of unique log keys, intending to improve the ability of a downstream sequence-based model to effectively learn the log patterns. The preliminary results of ClusterLog indicate not only its effectiveness in reducing the granularity of log sequences without the loss of important sequence information but also its generalizability to different file systems' logs.

cs.DC↗

arXiv2023

IOPathTune: Adaptive Online Parameter Tuning for Parallel File System I/O Path

Parallel file systems contain complicated I/O paths from clients to storage servers. An efficient I/O path requires proper settings of multiple parameters, as the default settings often fail to deliver optimal performance, especially for diverse workloads in the HPC environment. Existing tuning strategies have shortcomings in being adaptive, timely, and flexible. We propose IOPathTune, which adaptively tunes PFS I/O Path online from the client side without characterizing the workloads, doing expensive profiling, and communicating with other machines. We implemented IOPathTune on Lustre and leveraged CloudLab to conduct the evaluations on 20 different Filebench workloads in three different scenarios. We observed either on-par or better performance than the default configuration, as high as 231% on standalone executions. IOPathTune also delivers 89.57% better overall performance than CAPES in multiple client executions.

cs.DC↗