王豪杰｜清华大学｜学者主页

2026

Taming Dynamic Diffusion LLM Inference through Virtual Static Execution

Jia-Nian Zhu, Hang Wu, Ying-Hui Li, Haojie Wang, Rui-Xuan Li, Ji-Dong Zhai

International Conference on Supercomputing

2026

UniOrch: A Unified Mixed Framework for High-Efficiency LLM Training on Heterogeneous AI Chips

Jing-Yi Wang, Jia Wang, Jun Gao, Yanlu Cao, Yang Zhai, Haojie Wang, Wanxin Song, Dezhi Kong, Liu Liu, Wei Li, Zhaofeng He

IEEE Transactions on Parallel and Distributed Systems

DOI

2026

Hiergossip: a Hierarchical Gossip Protocol for Heterogeneous Distributed Machine Learning

Haojie Wang, Rui Guo

2026 9th International Conference on Advanced Algorithms and Control Engineering (ICAACE)

DOI

2026

ChituDiffusion: A Data-Characteristic-Aware Serving System for Diffusion Models

Cheng-Zhang Wu, Liyan Zheng, Haojie Wang, Kezhao Huang, Zi-Xuan Ma, Dong Dong, Ji-Dong Zhai

ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming

DOI

2026

Decoupling Readiness from Release for Tail-Aware Scheduling of Agentic LLM Workflows

Bo-Chao Feng, Jian-Jiang Li, Haojie Wang, Lin Qiao, Ying-Hui Li, Yu-Kun Yan, Ji-Dong Zhai

2026

Epoch: Compiling Diffusion Blocks for Sparse MoE Serving

Jia-Nian Zhu, Hang Wu, Ying-Hui Li, Haojie Wang, Rui-Xuan Li, Ji-Dong Zhai

2025

面向国产智能芯片的统一智能计算架构

王豪杰，潘泽众，黄嘉成，李映辉，赵家程，武永卫

《计算》

2025

Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning

Chengze Du, Zhiwei Yu, Heng Xu, Haojie Wang, Bo Liu, Jia-Long Li

IEEE Transactions on Cloud Computing

DOI

2025

SpecRouter: Adaptive Routing for Multi-Level Speculative Decoding in Large Language Models

Hang Wu, Jia-Nian Zhu, Ying-Hui Li, Haojie Wang, Biao Hou, Ji-Dong Zhai

arXiv.org

DOI

2025

Learning Optimal Transport for GPU Allocation: A Deep Learning Approach to Efficient Resource Scheduling

Chengze Du, Zhiwei Yu, Heng Xu, Haojie Wang, Bo Liu, Jia-Long Li

Asia Communications and Photonics conference and Exhibition

DOI

2025

Leveraging Graph Analysis to Pinpoint Root Causes of Scalability Issues for Parallel Applications

Yu-Yang Jin, Haojie Wang, Xiongchao Tang, Zhen-Hua Guo, Yaqian Zhao, T. Hoefler, Tao Liu, Xu Liu, Ji-Dong Zhai

IEEE Transactions on Parallel and Distributed Systems

DOI

2025

TopoFair: A Fairness-Aware Topology Selection for Reconfigurable Data Center Networks

Zhiwei Yu, Chengze Du, Heng Xu, Haojie Wang, Bo Liu, Jia-Long Li

Asia Communications and Photonics conference and Exhibition

DOI

2025

IntelliGen: Instruction-Level Auto-tuning for Tensor Program with Monotonic Memory Optimization

Zi-Xuan Ma, Haojie Wang, Jingze Xing, Shuhong Huang, Liyan Zheng, Chen Zhang, Huan-Qi Cao, Kezhao Huang, Ming-Shu Zhai, Shizhi Tang, Penghan Wang, Ji-Dong Zhai

IEEE/ACM International Symposium on Code Generation and Optimization

DOI

2025

LayerScope: Predictive Cross-Layer Scheduling for Efficient Multi-Batch MoE Inference on Legacy Servers

En-Da Yu, Dezun Dong, Zhao-Ning Zhang, Zhe Bai, Wei-Ling Yang, Haojie Wang, Dongsheng Li, Yong Wu, Xiang-Ke Liao

International Conference on Supercomputing

DOI

2025

FastCache: Optimizing Multimodal LLM Serving through Lightweight KV-Cache Compression Framework

Jia-Nian Zhu, Hang Wu, Haojie Wang, Ying-Hui Li, Biao Hou, Rui-Xuan Li, Ji-Dong Zhai

arXiv.org

DOI

2025

NineToothed: A Triton-Based High-Level Domain-Specific Language for Machine Learning

Jia-Cheng Huang, Zimin Li, Ying-Hui Li, Haojie Wang

arXiv.org

DOI

2025

FlexPie: Accelerate Distributed Inference on Edge Devices with Flexible Combinatorial Optimization[Technical Report]

Run-Hua Zhang, Hong-Xu Jiang, Jinkun Geng, Yuhang Ma, Chenhui Zhu, Haojie Wang

International Conference on Database Systems for Advanced Applications

DOI

2025

RailS: Load Balancing for All-to-All Communication in Distributed Mixture-of-Experts Training

Heng Xu, Zhiwei Yu, Chengze Du, Ying Zhou, Letian Li, Haojie Wang, Wei-Qiang Cheng, Jia-Long Li

IEEE Transactions on Networking

DOI

2024

A comparative study of network processor chips for G-SRv6

Weiqiang Cheng, Jinming Li, Shu-Kun Zhang, Haojie Wang, Yu Yan, Wenying Jiang

Other Conferences

DOI

2024

WiseGraph: Optimizing GNN with Joint Workload Partition of Graph and Operations

Kezhao Huang, Ji-Dong Zhai, Liyan Zheng, Haojie Wang, Yu-Yang Jin, Qihao Zhang, Run-Qing Zhang, Zhen Zheng, Youngmin Yi, Xi-Peng Shen

European Conference on Computer Systems

DOI

2024

MAGPY: Compiling Eager Mode DNN Programs by Monitoring Execution States

Chen Zhang, Rong-Chao Dong, Haojie Wang, Runxin Zhong, Jike Chen, Ji-Dong Zhai

USENIX Annual Technical Conference

2024

AdaPipe: Optimizing Pipeline Parallelism with Adaptive Recomputation and Partitioning

Zhenbo Sun, Huanqi Cao, Yuanwei Wang, Guanyu Feng, Sheng-Qi Chen, Haojie Wang, Wen-Guang Chen

International Conference on Architectural Support for Programming Languages and Operating Systems

DOI

2024

Graph-Centric Performance Analysis for Large-Scale Parallel Applications

Yu-Yang Jin, Haojie Wang, Runxin Zhong, Chen Zhang, Xia Liao, Feng Zhang, Ji-Dong Zhai

IEEE Transactions on Parallel and Distributed Systems

DOI

2024

Optimal Kernel Orchestration for Tensor Programs with Korch

Mu-Yan Hu, Ashwin Venkatram, Shreyashri Biswas, Balamurugan Marimuthu, Bohan Hou, Gabriele Oliaro, Haojie Wang, Liyan Zheng, Xu-Peng Miao, Ji-Dong Zhai, Zhi-Hao Jia

International Conference on Architectural Support for Programming Languages and Operating Systems

DOI

2024

A high-performance dataflow-centric optimization framework for deep learning inference on the edge

Run-Hua Zhang, Hong-Xu Jiang, Jinkun Geng, Fangzheng Tian, Yuhang Ma, Haojie Wang

Journal of systems architecture

DOI

2024

An efficient CNN accelerator for pattern-compressed sparse neural networks on FPGA

Yonghua Zhang, Haojie Wang, Zhenhua Pan

Neurocomputing

DOI

2023

GraphSet: High Performance Graph Mining through Equivalent Set Transformations

Tianhui Shi, Ji-Dong Zhai, Haojie Wang, Qiqian Chen, Ming-Shu Zhai, Zixu Hao, Haoyu Yang, Wen-Guang Chen

International Conference for High Performance Computing, Networking, Storage and Analysis

DOI

2023

PowerFusion: A Tensor Compiler with Explicit Data Movement Description and Instruction-level Graph IR

Zi-Xuan Ma, Haojie Wang, Jingze Xing, Liyan Zheng, Chen Zhang, Huanqi Cao, Kezhao Huang, Shizhi Tang, Penghan Wang, Ji-Dong Zhai

arXiv.org

DOI

2023

Optimizing DNNs With Partially Equivalent Transformations and Automated Corrections

Haojie Wang, Ji-Dong Zhai, Mingyu Gao, Feng Zhang, Tuowei Wang, Zi-Xuan Ma, Shizhi Tang, Liyan Zheng, Wen Wang, Kaiyuan Rong, Yuanyong Chen, Zhi-Hao Jia

IEEE transactions on computers

DOI

2023

Xenos: Dataflow-Centric Optimization to Accelerate Model Inference on Edge Devices

Run-Hua Zhang, Hongxu Jiang, Fangzheng Tian, Jin-Kun Geng, Xiaobin Li, Yuhang Ma, Chenhui Zhu, Dong Dong, Xin Li, Haojie Wang

International Conference on Database Systems for Advanced Applications

DOI

2023

Unified Programming Models for Heterogeneous High-Performance Computers

Zi-Xuan Ma, Yu-Yang Jin, Shizhi Tang, Haojie Wang, Wei Xue, Ji-Dong Zhai, Wei-Min Zheng

Journal of Computational Science and Technology

DOI

2023

EINNET: Optimizing Tensor Programs with Derivation-Based Transformations

Liyan Zheng, Haojie Wang, Ji-Dong Zhai, Mu-Yan Hu, Zi-Xuan Ma, Tuowei Wang, Shuhong Huang, Xu-Peng Miao, Shizhi Tang, Kezhao Huang, Zhi-Hao Jia

USENIX Symposium on Operating Systems Design and Implementation

2022

FasterMoE: modeling and optimizing training of large-scale dynamic pre-trained models

Jiaao He, Ji-Dong Zhai, Tiago Antunes, Haojie Wang, Fuwen Luo, Shangfeng Shi, Qin Li

ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming

DOI

2022

An Efficient Sparse CNNs Accelerator on FPGA

Yonghua Zhang, Hongxu Jiang, Xiaobin Li, Haojie Wang, Dong Dong, Yong-Xiang Cao

IEEE International Conference on Cluster Computing

DOI

2022

Efficiently emulating high-bitwidth computation with low-bitwidth hardware

Zi-Xuan Ma, Haojie Wang, Guanyu Feng, Chen Zhang, Lei Xie, Jiaao He, Sheng-Qi Chen, Ji-Dong Zhai

International Conference on Supercomputing

DOI

2022

BaGuaLu: targeting brain scale pretrained models with over 37 million cores

Zi-Xuan Ma, Jiaao He, Jie-Zhong Qiu, Huan Cao, Yuanwei Wang, Zhenbo Sun, Liyan Zheng, Haojie Wang, Shizhi Tang, Tian-Yu Zheng, Jun-Yang Lin, Guanyu Feng, Zeqiang Huang, Jie Gao, Ao-Han Zeng, Jian-Wei Zhang, Runxin Zhong, Tianhui Shi, Sha Liu, Wei-Min Zheng, Jie Tang, Hong-Xia Yang, Xin Liu, Ji-Dong Zhai, Wen-Guang Chen

ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming

DOI

2022

Detecting Performance Variance for Parallel Applications without Source Code

Ji-Dong Zhai, Liyan Zheng, Feng Zhang, Xiongchao Tang, Haojie Wang, Teng Yu, Yu-Yang Jin, S. Song, Wen-Guang Chen

IEEE Transactions on Parallel and Distributed Systems

DOI

2022

Scaling graph traversal to 281 trillion edges with 40 million cores

Huanqi Cao, Yuanwei Wang, Haojie Wang, Heng Lin, Zi-Xuan Ma, Wan-Wang Yin, Wen-Guang Chen

ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming

DOI

2022

UniQ: A Unified Programming Model for Efficient Quantum Circuit Simulation

Chen Zhang, Haojie Wang, Zi-Xuan Ma, Lei Xie, Ze-Yu Song, Ji-Dong Zhai

International Conference for High Performance Computing, Networking, Storage and Analysis

DOI

2022

FreeTensor: a free-form DSL with holistic optimizations for irregular tensor programs

Shizhi Tang, Ji-Dong Zhai, Haojie Wang, Lin Jiang, Liyan Zheng, Zhenhao Yuan, Chen Zhang

ACM-SIGPLAN Symposium on Programming Language Design and Implementation

DOI

2022

Vapro: performance variance detection and diagnosis for production-run parallel applications

Liyan Zheng, Ji-Dong Zhai, Xiongchao Tang, Haojie Wang, Teng Yu, Yu-Yang Jin, S. Song, Wen-Guang Chen

ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming

DOI

2022

OLLIE: Derivation-based Tensor Program Optimizer

Liyan Zheng, Haojie Wang, Ji-Dong Zhai, Mu-Yan Hu, Zi-Xuan Ma, Tuowei Wang, Shizhi Tang, Lei Xie, Kezhao Huang, Zhi-Hao Jia

arXiv.org

DOI

2022

PerFlow: a domain specific framework for automatic performance analysis of parallel applications

Yu-Yang Jin, Haojie Wang, Runxin Zhong, Chen Zhang, Ji-Dong Zhai

ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming

DOI

2022

<inline-formula><tex-math notation="LaTeX">$TC-Stream$</tex-math><alternatives><mml:math><mml:mrow><mml:mi>T</mml:mi><mml:mi>C</mml:mi><mml:mo>-</mml:mo><mml:mi>S</mml:mi><mml:mi>t</mml:mi><mml:mi>r</mml:mi><mml:mi>e</mml:mi><mml:mi>a</mml:mi><mml:mi>m</mml:mi></mml:mrow></mml:math><inline-graphic x

Jian-Qiang Huang, Haojie Wang, Xiang Fei, Xiaoying Wang, Wen-Guang Chen

IEEE Transactions on Parallel and Distributed Systems

DOI

2021

Sparker: Efficient Reduction for More Scalable Machine Learning with Spark

Bowen Yu, Huanqi Cao, Tianyi Shan, Haojie Wang, Xiongchao Tang, Wen-Guang Chen

International Conference on Parallel Processing

DOI

2021

HyQuas: hybrid partitioner based quantum circuit simulation system on GPU

Chen Zhang, Ze-Yu Song, Haojie Wang, Kaiyuan Rong, Ji-Dong Zhai

International Conference on Supercomputing

DOI

2021

Chukonu: A Fully-Featured Big Data Processing System by Efficiently Integrating a Native Compute Engine into Spark

Bowen Yu, Guanyu Feng, Huanqi Cao, Xiaohan Li, Zhenbo Sun, Haojie Wang, Xiao-Wei Zhu, Wei-Min Zheng, Wen-Guang Chen

Proceedings of the VLDB Endowment

2021

LotusSQL: SQL engine for high-performance big data systems

Xiaohan Li, Bowen Yu, Guanyu Feng, Haojie Wang, Wenguang Chen

Big Data Mining and Analytics

DOI

2021

Chukonu

Bowen Yu, Guanyu Feng, Huanqi Cao, Xiaohan Li, Zhenbo Sun, Haojie Wang, Xiao-Wei Zhu, Wei-Min Zheng, Wen-Guang Chen

Proceedings of the VLDB Endowment

DOI

2021

PET: Optimizing Tensor Programs with Partially Equivalent Transformations and Automated Corrections

Haojie Wang, Ji-Dong Zhai, Mingyu Gao, Zi-Xuan Ma, Shizhi Tang, Liyan Zheng, Yuanzhi Li, Kaiyuan Rong, Yuanyong Chen, Zhi-Hao Jia

USENIX Symposium on Operating Systems Design and Implementation

2020

Identifying scalability bottlenecks for large-scale parallel programs with graph analysis

Yu-Yang Jin, Haojie Wang, Xiongchao Tang, T. Hoefler, Xu Liu, Ji-Dong Zhai

ACM SIGPLAN Symposium on Principles & Practice of Parallel Programming

DOI

2020

SCALANA: Automating Scaling Loss Detection with Graph Analysis

Yu-Yang Jin, Haojie Wang, Teng Yu, Xiongchao Tang, T. Hoefler, Xu Liu, Ji-Dong Zhai

International Conference for High Performance Computing, Networking, Storage and Analysis

DOI

2019

Spread-n-Share: Improving Application Performance and Cluster Throughput with Resource-aware Job Placement

Xiongchao Tang, Haojie Wang, Xiao-Song Ma, Nosayba El-Sayed, Ji-Dong Zhai, Wen-Guang Chen, Ashraf Aboulnaga

International Conference for High Performance Computing, Networking, Storage and Analysis

DOI

2018

Spindle: Informed Memory Access Monitoring

Haojie Wang, Ji-Dong Zhai, Xiongchao Tang, Bowen Yu, Xiao-Song Ma, Wen-Guang Chen

USENIX Annual Technical Conference

Chukonu: A Fully-Featured High-Performance Big Data Framework that Integrates a Native Compute Engine into Spark

Bowen Yu, Guanyu Feng, Huanqi Cao, Xiaohan Li, Zhenbo Sun, Haojie Wang, Xiao-Wei Zhu, Wei-Min Zheng, Wen-Guang Chen

This paper is included in the Proceedings of the 17th USENIX Symposium on Operating Systems Design and Implementation.

Systems Design, Liyan Zheng, Haojie Wang, Ji-Dong Zhai, Mu-Yan Hu, Zi-Xuan Ma, Tuowei Wang, Shuhong Huang, Xupeng Miao, Shizhi Tang, Kezhao Huang, ZhihaoJia

精选作品

期刊论文

Taming Dynamic Diffusion LLM Inference through Virtual Static Execution

UniOrch: A Unified Mixed Framework for High-Efficiency LLM Training on Heterogeneous AI Chips

Hiergossip: a Hierarchical Gossip Protocol for Heterogeneous Distributed Machine Learning

ChituDiffusion: A Data-Characteristic-Aware Serving System for Diffusion Models

Decoupling Readiness from Release for Tail-Aware Scheduling of Agentic LLM Workflows

Epoch: Compiling Diffusion Blocks for Sparse MoE Serving

面向国产智能芯片的统一智能计算架构

Temporal-Aware GPU Resource Allocation for Distributed LLM Inference via Reinforcement Learning

SpecRouter: Adaptive Routing for Multi-Level Speculative Decoding in Large Language Models

Learning Optimal Transport for GPU Allocation: A Deep Learning Approach to Efficient Resource Scheduling

Leveraging Graph Analysis to Pinpoint Root Causes of Scalability Issues for Parallel Applications

TopoFair: A Fairness-Aware Topology Selection for Reconfigurable Data Center Networks

IntelliGen: Instruction-Level Auto-tuning for Tensor Program with Monotonic Memory Optimization

LayerScope: Predictive Cross-Layer Scheduling for Efficient Multi-Batch MoE Inference on Legacy Servers

FastCache: Optimizing Multimodal LLM Serving through Lightweight KV-Cache Compression Framework

NineToothed: A Triton-Based High-Level Domain-Specific Language for Machine Learning

FlexPie: Accelerate Distributed Inference on Edge Devices with Flexible Combinatorial Optimization[Technical Report]

RailS: Load Balancing for All-to-All Communication in Distributed Mixture-of-Experts Training

A comparative study of network processor chips for G-SRv6

WiseGraph: Optimizing GNN with Joint Workload Partition of Graph and Operations

MAGPY: Compiling Eager Mode DNN Programs by Monitoring Execution States

AdaPipe: Optimizing Pipeline Parallelism with Adaptive Recomputation and Partitioning

Graph-Centric Performance Analysis for Large-Scale Parallel Applications

Optimal Kernel Orchestration for Tensor Programs with Korch

A high-performance dataflow-centric optimization framework for deep learning inference on the edge

An efficient CNN accelerator for pattern-compressed sparse neural networks on FPGA

GraphSet: High Performance Graph Mining through Equivalent Set Transformations

PowerFusion: A Tensor Compiler with Explicit Data Movement Description and Instruction-level Graph IR

Optimizing DNNs With Partially Equivalent Transformations and Automated Corrections

Xenos: Dataflow-Centric Optimization to Accelerate Model Inference on Edge Devices

Unified Programming Models for Heterogeneous High-Performance Computers

EINNET: Optimizing Tensor Programs with Derivation-Based Transformations

FasterMoE: modeling and optimizing training of large-scale dynamic pre-trained models

An Efficient Sparse CNNs Accelerator on FPGA

Efficiently emulating high-bitwidth computation with low-bitwidth hardware

BaGuaLu: targeting brain scale pretrained models with over 37 million cores

Detecting Performance Variance for Parallel Applications without Source Code

Scaling graph traversal to 281 trillion edges with 40 million cores

UniQ: A Unified Programming Model for Efficient Quantum Circuit Simulation

FreeTensor: a free-form DSL with holistic optimizations for irregular tensor programs

Vapro: performance variance detection and diagnosis for production-run parallel applications

OLLIE: Derivation-based Tensor Program Optimizer

PerFlow: a domain specific framework for automatic performance analysis of parallel applications

Sparker: Efficient Reduction for More Scalable Machine Learning with Spark

HyQuas: hybrid partitioner based quantum circuit simulation system on GPU

Chukonu: A Fully-Featured Big Data Processing System by Efficiently Integrating a Native Compute Engine into Spark

LotusSQL: SQL engine for high-performance big data systems

Chukonu

PET: Optimizing Tensor Programs with Partially Equivalent Transformations and Automated Corrections

Identifying scalability bottlenecks for large-scale parallel programs with graph analysis

SCALANA: Automating Scaling Loss Detection with Graph Analysis

Spread-n-Share: Improving Application Performance and Cluster Throughput with Resource-aware Job Placement

Spindle: Informed Memory Access Monitoring

Chukonu: A Fully-Featured High-Performance Big Data Framework that Integrates a Native Compute Engine into Spark

This paper is included in the Proceedings of the 17th USENIX Symposium on Operating Systems Design and Implementation.

预印本

Xenos: Accelerating Edge-Based Model Inference with Dataflow-Centric Optimization

研究方向