arXiv subjects

Xu Wen

Publications and source records attributed to Xu Wen.

14 recordsLinked to original sources

arXiv2026

Assessment of turbulent fire dynamics and combustion instabilities using a flamelet model

The Sandia one-meter methane fire plume is an established benchmark for turbulent combustion modeling of large-scale flames. This study investigates the combustion instabilities formed close to the base of the Sandia fire plume. Finite rate chemistry and differential diffusion are considered using a flamelet/progress variable (FPV) approach. The performance of the FPV approach is assessed by comparing with the eddy dissipation model (EDM) and the experimental data for the large-scale fire plume via large eddy simulations (LES). The effects of radiation modeling and mesh resolution on the predictive capability of the model are systematically investigated by comparing the axial and radical velocities against the experimental data at various locations. Although all models successfully capture the primary flow characteristics of fire plumes, the FPV model with differential diffusion yields improved predictions in the near-flame-base region. The formation mechanism of cellular flow structures near the flame base is investigated via a budget analysis of the vorticity equation, and the type of instability governing the formation of the cellular structure is clarified. Finally, the individual effects of finite rate chemistry and differential diffusion on the prediction of the thermo-chemical quantities are quantified. Overall, this study explains the underlying physics governing combustion instabilities at the base of turbulent fire plumes, provides novel insights into the performance of flamelet models for LES of gaseous pool fires, and offers reliable guidance for the high-fidelity numerical simulation of large-scale turbulent buoyancy driven flames.

physics.flu-dyn↗

arXiv2024

Bridging the Gap Between Domain-specific Frameworks and Multiple Hardware Devices

The rapid development of domain-specific frameworks has presented us with a significant challenge: The current approach of implementing solutions on a case-by-case basis incurs a theoretical complexity of O(M*N), thereby increasing the cost of porting applications to different hardware platforms. To address these challenges, we propose a systematic methodology that effectively bridges the gap between domain-specific frameworks and multiple hardware devices, reducing porting complexity to O(M+N). The approach utilizes multi-layer abstractions. Different domain-specific abstractions are employed to represent applications from various domains. These abstractions are then transformed into a unified abstraction, which is subsequently translated into combinations of primitive operators. Finally, these operators are mapped to multiple hardware platforms. The implemented unified framework supports deep learning, classical machine learning, and data analysis across X86, ARM, RISC-V, IoT devices, and GPU. It outperforms existing solutions like scikit-learn, hummingbird, Spark, and pandas, achieving impressive speedups: 1.1x to 3.83x on X86 servers, 1.06x to 4.33x on ARM IoT devices, 1.25x to 3.72x on RISC-V IoT devices, and 1.93x on GPU. The source code is available at https://github.com/BenchCouncil/bridger.git.

cs.SE↗

arXiv2024

Quality at the Tail of Machine Learning Inference

Machine learning inference should be subject to stringent inference time constraints while ensuring high inference quality, especially in safety-critical (e.g., autonomous driving) and mission-critical (e.g., emotion recognition) contexts. Neglecting either aspect can lead to severe consequences, such as loss of life and property damage. Many studies lack a comprehensive consideration of these metrics, leading to incomplete or misleading evaluations. The study unveils a counterintuitive revelation: deep learning inference quality exhibits fluctuations due to inference time. To depict this phenomenon, the authors coin a new term, "tail quality," providing a more comprehensive evaluation, and overcoming conventional metric limitations. Moreover, the research proposes an initial evaluation framework to analyze factors affecting quality fluctuations, facilitating the prediction of the potential distribution of inference quality. The effectiveness of the evaluation framework is validated through experiments conducted on deep learning models for three different tasks across four systems.

cs.LG↗

arXiv2023

CMLCompiler: A Unified Compiler for Classical Machine Learning

Classical machine learning (CML) occupies nearly half of machine learning pipelines in production applications. Unfortunately, it fails to utilize the state-of-the-practice devices fully and performs poorly. Without a unified framework, the hybrid deployments of deep learning (DL) and CML also suffer from severe performance and portability issues. This paper presents the design of a unified compiler, called CMLCompiler, for CML inference. We propose two unified abstractions: operator representations and extended computational graphs. The CMLCompiler framework performs the conversion and graph optimization based on two unified abstractions, then outputs an optimized computational graph to DL compilers or frameworks. We implement CMLCompiler on TVM. The evaluation shows CMLCompiler's portability and superior performance. It achieves up to 4.38$\times$ speedup on CPU, 3.31$\times$ speedup on GPU, and 5.09$\times$ speedup on IoT devices, compared to the state-of-the-art solutions -- scikit-learn, intel sklearn, and hummingbird. Our performance of CML and DL mixed pipelines achieves up to 3.04x speedup compared with cross-framework implementations. The project documents and source code are available at https://www.computercouncil.org/cmlcompiler.

cs.LG↗

arXiv2023

Flamelet modeling of thermo-diffusively unstable hydrogen-air flames

In order to reduce CO2 emissions, hydrogen combustion has become increasingly relevant for technical applications. In this context, lean H2-air flames show promising features but, among other characteristics, they tend to exhibit thermo-diffusive instabilities. The formation of cellular structures associated with these instabilities leads to an increased flame surface area which further promotes the flame propagation speed, an important reference quantity for design, control, and safe operation of technical combustors. While many studies have addressed the physical phenomena of intrinsic flame instabilities in the past, there is also a demand to predict such flame characteristics with reduced-order models to allow computationally efficient simulations. In this work, a H2-air spherical expanding flame, which exhibits thermo-diffusive instabilities, is studied with flamelet-based modeling approaches both in a-priori and a-posteriori manner. A recently proposed Flamelet/Progress Variable (FPV) model, with a manifold based on unstretched planar flames, and a novel FPV approach, which takes into account a large curvature variation in the tabulated manifold, are compared to detailed chemistry (DC) calculations. First, both FPV approaches are assessed in terms of an a-priori test with the DC reference dataset. Thereafter, the a-posteriori assessment contains two parts: a linear stability analysis of perturbed planar flames and the simulation of the spherical expanding flame. Both FPV models are systematically analyzed considering global and local flame properties in comparison to the DC reference data. It is shown that the new FPV model, incorporating large curvature variations in the manifold, leads to improved predictions for the microstructure of the corrugated flame front and the formation of cellular structures, while global flame properties are reasonably well reproduced by both models.

physics.flu-dyn↗

arXiv2022

Numerical modeling of pulverized iron flames in a multidimensional hot counterflow burner

Pulverized iron flames stabilized in a multidimensional hot counterflow burner are simulated using a numerical model, which is extended from the state-of-the-art model developed by Hazenberg and van Oijen (PCI, 2021) considering unsteady effects. The results are compared to available experimental data (McRae et al., PCI, 2019), including particle image velocimetry measurements, a direct flame photo, the flow field velocity and the flame speed for different iron and oxygen concentrations. The comparison shows that the particle dynamics and flame shape can be reasonably well predicted. The flow field velocity and flame speed also show quantitative agreement between the simulation and the experiment. Based on the validated simulation results, the iron combustion characteristics, including the thermal structures and the multidimensional effects, are analyzed for different oxidizer environments. The analysis shows that the iron particles undergo a transition from kinetic-controlled regime (up to ignition) to a diffusion-controlled regime (burning) at the central axis for both environments with the particle temperature being higher than the gas temperature at the flame front, which is indicated by the Damköhler number. For the hot counterflow burner, there exist multidimensional effects, i.e., the temperature and Damköhler number change along the radial direction.

physics.flu-dyn↗

arXiv2021

AIBench Scenario: Scenario-distilling AI Benchmarking

Modern real-world application scenarios like Internet services consist of a diversity of AI and non-AI modules with huge code sizes and long and complicated execution paths, which raises serious benchmarking or evaluating challenges. Using AI components or micro benchmarks alone can lead to error-prone conclusions. This paper presents a methodology to attack the above challenge. We formalize a real-world application scenario as a Directed Acyclic Graph-based model and propose the rules to distill it into a permutation of essential AI and non-AI tasks, which we call a scenario benchmark. Together with seventeen industry partners, we extract nine typical scenario benchmarks. We design and implement an extensible, configurable, and flexible benchmark framework. We implement two Internet service AI scenario benchmarks based on the framework as proxies to two real-world application scenarios. We consider scenario, component, and micro benchmarks as three indispensable parts for evaluating. Our evaluation shows the advantage of our methodology against using component or micro AI benchmarks alone. The specifications, source code, testbed, and results are publicly available from \url{https://www.benchcouncil.org/aibench/scenario/}.

cs.PF↗

arXiv2021

AIBench Training: Balanced Industry-Standard AI Training Benchmarking

Earlier-stage evaluations of a new AI architecture/system need affordable benchmarks. Only using a few AI component benchmarks like MLPerfalone in the other stages may lead to misleading conclusions. Moreover, the learning dynamics are not well understood, and the benchmarks' shelf-life is short. This paper proposes a balanced benchmarking methodology. We use real-world benchmarks to cover the factors space that impacts the learning dynamics to the most considerable extent. After performing an exhaustive survey on Internet service AI domains, we identify and implement nineteen representative AI tasks with state-of-the-art models. For repeatable performance ranking (RPR subset) and workload characterization (WC subset), we keep two subsets to a minimum for affordability. We contribute by far the most comprehensive AI training benchmark suite. The evaluations show: (1) AIBench Training (v1.1) outperforms MLPerfTraining (v0.7) in terms of diversity and representativeness of model complexity, computational cost, convergent rate, computation, and memory access patterns, and hotspot functions; (2) Against the AIBench full benchmarks, its RPR subset shortens the benchmarking cost by 64%, while maintaining the primary workload characteristics; (3) The performance ranking shows the single-purpose AI accelerator like TPU with the optimized TensorFlowframework performs better than that of GPUs while losing the latter's general support for various AI models. The specification, source code, and performance numbers are available from the AIBench homepage https://www.benchcouncil.org/aibench-training/index.html.

cs.AI↗

arXiv2020

AIBench: An Agile Domain-specific Benchmarking Methodology and an AI Benchmark Suite

Domain-specific software and hardware co-design is encouraging as it is much easier to achieve efficiency for fewer tasks. Agile domain-specific benchmarking speeds up the process as it provides not only relevant design inputs but also relevant metrics, and tools. Unfortunately, modern workloads like Big data, AI, and Internet services dwarf the traditional one in terms of code size, deployment scale, and execution path, and hence raise serious benchmarking challenges. This paper proposes an agile domain-specific benchmarking methodology. Together with seventeen industry partners, we identify ten important end-to-end application scenarios, among which sixteen representative AI tasks are distilled as the AI component benchmarks. We propose the permutations of essential AI and non-AI component benchmarks as end-to-end benchmarks. An end-to-end benchmark is a distillation of the essential attributes of an industry-scale application. We design and implement a highly extensible, configurable, and flexible benchmark framework, on the basis of which, we propose the guideline for building end-to-end benchmarks, and present the first end-to-end Internet service AI benchmark. The preliminary evaluation shows the value of our benchmark suite---AIBench against MLPerf and TailBench for hardware and software designers, micro-architectural researchers, and code developers. The specifications, source code, testbed, and results are publicly available from the web site \url{http://www.benchcouncil.org/AIBench/index.html}.

cs.PF↗

arXiv2020

AI-oriented Medical Workload Allocation for Hierarchical Cloud/Edge/Device Computing

In a hierarchically-structured cloud/edge/device computing environment, workload allocation can greatly affect the overall system performance. This paper deals with AI-oriented medical workload generated in emergency rooms (ER) or intensive care units (ICU) in metropolitan areas. The goal is to optimize AI-workload allocation to cloud clusters, edge servers, and end devices so that minimum response time can be achieved in life-saving emergency applications. In particular, we developed a new workload allocation method for the AI workload in distributed cloud/edge/device computing systems. An efficient scheduling and allocation strategy is developed in order to reduce the overall response time to satisfy multi-patient demands. We apply several ICU AI workloads from a comprehensive edge computing benchmark Edge AIBench. The healthcare AI applications involved are short-of-breath alerts, patient phenotype classification, and life-death threats. Our experimental results demonstrate the high efficiency and effectiveness in real-life health-care and emergency applications.

cs.DC↗

arXiv2019

HPC AI500: A Benchmark Suite for HPC AI Systems

In recent years, with the trend of applying deep learning (DL) in high performance scientific computing, the unique characteristics of emerging DL workloads in HPC raise great challenges in designing, implementing HPC AI systems. The community needs a new yard stick for evaluating the future HPC systems. In this paper, we propose HPC AI500 --- a benchmark suite for evaluating HPC systems that running scientific DL workloads. Covering the most representative scientific fields, each workload from HPC AI500 is based on real-world scientific DL applications. Currently, we choose 14 scientific DL benchmarks from perspectives of application scenarios, data sets, and software stack. We propose a set of metrics for comprehensively evaluating the HPC AI systems, considering both accuracy, performance as well as power and cost. We provide a scalable reference implementation of HPC AI500. HPC AI500 is a part of the open-source AIBench project, the specification and source code are publicly available from \url{http://www.benchcouncil.org/AIBench/index.html}.

cs.PF↗

arXiv2019

Edge AIBench: Towards Comprehensive End-to-end Edge Computing Benchmarking

In edge computing scenarios, the distribution of data and collaboration of workloads on different layers are serious concerns for performance, privacy, and security issues. So for edge computing benchmarking, we must take an end-to-end view, considering all three layers: client-side devices, edge computing layer, and cloud servers. Unfortunately, the previous work ignores this most important point. This paper presents the BenchCouncil's coordinated e ort on edge AI benchmarks, named Edge AIBench. In total, Edge AIBench models four typical application scenarios: ICU Patient Monitor, Surveillance Camera, Smart Home, and Autonomous Vehicle with the focus on data distribution and workload collaboration on three layers. Edge AIBench is a part of the open-source AIBench project, publicly available from http://www.benchcouncil.org/AIBench/index.html. We also build an edge computing testbed with a federated learning framework to resolve performance, privacy, and security issues.

cs.PF↗

arXiv2018

BigDataBench: A Scalable and Unified Big Data and AI Benchmark Suite

Several fundamental changes in technology indicate domain-specific hardware and software co-design is the only path left. In this context, architecture, system, data management, and machine learning communities pay greater attention to innovative big data and AI algorithms, architecture, and systems. Unfortunately, complexity, diversity, frequently-changed workloads, and rapid evolution of big data and AI systems raise great challenges. First, the traditional benchmarking methodology that creates a new benchmark or proxy for every possible workload is not scalable, or even impossible for Big Data and AI benchmarking. Second, it is prohibitively expensive to tailor the architecture to characteristics of one or more application or even a domain of applications. We consider each big data and AI workload as a pipeline of one or more classes of units of computation performed on different initial or intermediate data inputs, each class of which we call a data motif. On the basis of our previous work that identifies eight data motifs taking up most of the run time of a wide variety of big data and AI workloads, we propose a scalable benchmarking methodology that uses the combination of one or more data motifs---to represent diversity of big data and AI workloads. Following this methodology, we present a unified big data and AI benchmark suite---BigDataBench 4.0, publicly available from~\url{http://prof.ict.ac.cn/BigDataBench}. This unified benchmark suite sheds new light on domain-specific hardware and software co-design: tailoring the system and architecture to characteristics of the unified eight data motifs other than one or more application case by case. Also, for the first time, we comprehensively characterize the CPU pipeline efficiency using the benchmarks of seven workload types in BigDataBench 4.0.

cs.DC↗

arXiv2018

Data Motifs: A Lens Towards Fully Understanding Big Data and AI Workloads

The complexity and diversity of big data and AI workloads make understanding them difficult and challenging. This paper proposes a new approach to modelling and characterizing big data and AI workloads. We consider each big data and AI workload as a pipeline of one or more classes of units of computation performed on different initial or intermediate data inputs. Each class of unit of computation captures the common requirements while being reasonably divorced from individual implementations, and hence we call it a data motif. For the first time, among a wide variety of big data and AI workloads, we identify eight data motifs that take up most of the run time of those workloads, including Matrix, Sampling, Logic, Transform, Set, Graph, Sort and Statistic. We implement the eight data motifs on different software stacks as the micro benchmarks of an open-source big data and AI benchmark suite ---BigDataBench 4.0 (publicly available from http://prof.ict.ac.cn/BigDataBench), and perform comprehensive characterization of those data motifs from perspective of data sizes, types, sources, and patterns as a lens towards fully understanding big data and AI workloads. We believe the eight data motifs are promising abstractions and tools for not only big data and AI benchmarking, but also domain-specific hardware and software co-design.

cs.DC↗