News

• [Aug 2026] The paper “Laplacian Frequency Hierarchies for Efficient 3D Gaussian Splatting Training” has been accepted by Pacific Graphics (PG) 2026. Congratulations to Dr. Yang Yixiong and Sisheng Zhang.
• [July 2026] The paper “SpatialGrammar: A Domain-Specific Language for LLM-Based 3D Indoor Scene Generation” has been accepted by ACMMM 2026.
• [July 2026] The paper “Same Cache, Different Latency: Understanding and Exploiting L2 Domain Locality in GPUs” has been accepted by MICRO 2026. Congratulations to Weile Luo and Yibin Ma.
• [July 2026] The paper “Exploiting Low-Level Sparsity for Efficient Large Language Model Inference on GPU with SpInfer” has been accepted by TOCS 2026.
• [July 2026] The paper “WAQ-LLM: Optimizing Multi-Instance LLM Deployment via Workload-Aware Queueing Model” has been accepted by ICPP 2026. Congratulations to Jiaxin Lai and Yizhou Luo.
• [April 2026] Our paper “DOSA: Optimizing DVFS Policy for Stochastic Workloads with Analytical Solutions” has been accepted by IWQoS 2026. Congratulations to Xuanhao Feng.
• [Dec 2025] The paper “FastTT: Accelerating Shift-XOR Erasure Coding for Data Storage” has been accepted by IPDPS 2026. Congratulations to Duo Sun.
• [Nov 2025] The paper “ZipServ: Fast and Memory-Efficient LLM Inference with Hardware-Aware Lossless Compression” has been accepted by ASPLOS 2026.
• [Nov 2025] The paper “ROME: Maximizing GPU Efficiency for All-Pairs Shortest Path via Taming Fine-Grained Irregularities” has been accepted by PPoPP 2026.
• [Nov 2025] The papers “SALR: Sparsity-Aware Low-Rank Representation for Efficient Fine-Tuning of Large Language Models” and “PipeDiT: Accelerating Diffusion Transformers in Video Generation with Task Pipelining and Model Decoupling” have been accepted by AAAI 2026.
• [Oct 2025] The paper “Castor: Optimizing Deep Learning Job Scheduling in Multi-Tenant GPU Clusters via Intelligent Colocation” has been accepted by TCC. Congratulations to Yizhou Luo.
• [Oct 2025] The paper “AUE: A Normalized Energy Efficiency Metric for AI Servers under LLM Workloads” has been accepted by ICPADS 2025. This paper proposes a new metric to evaluate the energy efficiency of AI servers, and has received the Best Presentation Award.
• [July 2025] The paper “City-VLM: Towards Multidomain Perception Scene Understanding via Multimodal Incomplete Learning” has been accepted by ACMMM 2025.
• [June 2025] The papers “SAFormer: Spatially Adaptive Transformer for Efficient and Multi-Resolution Occupancy Prediction” and “RA-NeRF: Robust Neural Radiance Field Reconstruction with Accurate Camera Pose Estimation under Complex Trajectories” have been accepted by IROS 2025.
• [May 2025] The paper “UnrealLLM: Towards Highly Controllable and Interactable 3D Scene Generation by LLM-powered Procedural Content Generation” has been accepted by ACL 2025 (Findings).
• [May 2025] The paper “BurstGPT: A Real-World Workload Dataset to Optimize LLM Serving Systems” has been accepted by KDD 2025 (Datasets and Benchmarks Track).
• [May 2025] The paper “SCFusion: Enhance Infrared and Visible Modality Fusion by Preserving Salient Object Consistency” has been accepted by IoTJ.
• [April 2025] The paper “SpInfer: Leveraging Low-Level Sparsity for Efficient Large Language Model Inference on GPUs” has received the Best Paper Award of EuroSys 2025.
• [March 2025] The paper “CMRFusion: Efficient Feature Decomposition for RGB-T Fusion via Cross Modality Mask Reconstruction” has been accepted by ICME 2025.
• [Jan 2025] The paper “SpInfer: Leveraging Low-Level Sparsity for Efficient Large Language Model Inference on GPUs” has been accepted by EuroSys 2025.
• [Jan 2025] The paper “STBLLM: Breaking the 1-Bit Barrier with Structured Binary LLMs” has been accepted by ICLR 2025.
• [Dec 2024] The paper “ParZC: Parametric Zero-Cost Proxies for Efficient NAS” has been accepted by AAAI 2025.
• [Nov 2024] The paper “SphereFusion: Efficient Panorama Depth Estimation via Gated Fusion” has been accepted by 3DV 2025.
• [Sept 2024] The paper “Discovering Sparsity Allocation for Layer-wise Pruning of Large Language Models” has been accepted by NeurIPS 2024.
• [Sept 2024] The paper “LPZero: Language Model Zero-cost Proxy Search from Zero” has been accepted by EMNLP Findings 2024.
• [July 2024] The papers “3D Question Answering for City Scene Understanding” and “Data Generation Scheme for Thermal Modality with Edge-Guided Adversarial Conditional Diffusion Model” have been accepted by ACMMM 2024.
• [July 2024] The paper “Multi-Task Domain Adaptation for Language Grounding with 3D Objects” has been accepted by ECCV 2024.
• [May 2024] The paper “Pruner-Zero: Evolving Symbolic Pruning Metric From Scratch for Large Language Models” has been accepted by ICML 2024.
• [April 2024] The papers “Scheduling Deep Learning Jobs in Multi-Tenant GPU Clusters via Wise Resource Sharing” and “DSO: A GPU Energy Efficiency Optimizer by Fusing Dynamic and Static Information” have been accepted by IWQoS 2024. Congratulations to Yizhou Luo and Laiyi Li.
• [Jan 2024] The paper “ScheMoE: An Extensible Mixture-of-Experts Distributed Training System with Tasks Scheduling” has been accepted by EuroSys 2024.
• [Dec 2023] The paper “Benchmarking and Dissecting the Nvidia Hopper GPU Architecture” has been accepted by IPDPS 2024.
• [Dec 2023] The paper “CF-NeRF: Camera Parameter Free Neural Radiance Fields with Incremental Learning” has been accepted by AAAI 2024.
• [Sept 2023] The paper “Improving GPU Energy Efficiency through an Application-transparent Frequency Scaling Policy with Performance Assurance” has been accepted by EuroSys 2024.
• [July 2023] The paper “Explicify Neural Implicit Fields for Efficient Dynamic Human Avatar Modeling via a Neural Explicit Surface” has been accepted by ACMMM 2023.
• [November 2022] The paper “Rethinking Disparity: A depth range free Multi-View Stereo based on Disparity” has been accepted by AAAI 2023.
• [August 2022] The paper “SphereDepth: Panorama Depth Estimation from Spherical Domain” has been accepted by 3DV 2022.
• [August 2022] The paper “Scale-Consistent Fusion: from Heterogeneous Local Sampling to Global Immersive Rendering” has been accepted by TIP.
• [July 2022] The paper “EASNet:Searching Elastic and Accurate Network Architecture for Stereo Matching” has been accepted by ECCV 2022.
• [May 2022] The paper “Energy-aware Non-preemptive Task Scheduling with Deadline Constraint in DVFS-enabled Heterogeneous Clusters” has been accepted by IEEE TPDS.
• [April 2022] The paper “Energy-efficient Online Scheduling of Transformer Inference Services on GPU Servers” has been accepted by IEEE Transactions on Green Communications and Networking (TGCN).
• [March 2021] The paper “IRS: A Large Naturalistic Indoor Robotics Stereo Dataset to Train Deep Models for Disparity and Surface Normal Estimation” has been accepted by ICME 2021.
• [March 2021] The paper “EDNet: Efficient Disparity Estimation with Cost Volume Combination and Attention-based Spatial Residual” has been accepted by CVPR 2021.
• [December 2020] The paper “Energy-Efficient Inference Service of Transformer-Based Deep Learning Models on GPUs” has received the Best Paper Award of IEEE GreenCom 2020.
• [September 2020] I started my academic position as a Research Assistant Professor at Department of Computer Science, Hong Kong Baptist University.
• [June 2020] The paper “GPGPU Performance Estimation with Core and Memory Frequency Scaling” has been accepted by IEEE TPDS.
• [January 2020] The paper “FADNet: A Fast and Accurate Network for Disparity Estimation” has been accepted by ICRA 2020.