11 citations · 20 across the 19 of their papers we have counts for
19 papers
Weave: Fine-Grained Dynamic SM Scheduling in an MoE Megakernel for Compute-Communication Overlap
Ziyu Huang, Yangjie Zhou, Chenhao Zhu +15
Mixture-of-Experts (MoE) inference under expert parallelism (EP) turns each MoE layer into a distributed computation with costly dispatch and combine communication. State-of-the-ar…
Dissecting How Die Scaling Breaks GPU Fine-grained Scheduling
Xiaoze Fan, Jianhao Wang, Weihao Cui +8
Modern GPUs are no longer physically symmetric. Die scaling leads to both manufacturing-driven floorsweeping and cache and memory partitioning. The former creates chip-specific com…
A Thread-Register Decoupled GPU Execution Model for Efficient Tensor Computation
Zihan Liu, Jingwen Leng, Yangjie Zhou +12
Modern GPUs increasingly integrate Tensor Cores into the execution pipeline. Although aggregate tensor throughput continues to grow, aided by an operand supply that has evolved fro…
GPU-Tile-Sim: A Tile-Centric GPU Simulation Framework for LLM Hardware-Software Co-Design
Yitong Ding, Jiawei Huang, Renyang Guan +7
Modern LLM (large language model) workloads increasingly rely on optimized GPU kernels through hardware-software co-design. These kernels achieve high-performance through fine-grai…
Optimus: Elastic Decoding for Efficient Diffusion LLM Serving
Chiyue Wei, Cong Guo, Bowen Duan +7
Large language model (LLM) serving is fundamentally limited by inefficient hardware utilization. Autoregressive (AR) decoding underutilizes GPUs due to its strictly sequential exec…
CuBridge: An LLM-Based Framework for Understanding and Reconstructing High-Performance Attention Kernels
Xing Ma, Yangjie Zhou, Wu Sun +6
Efficient CUDA implementations of attention mechanisms are critical to modern deep learning systems, yet supporting diverse and evolving attention variants remains challenging. Exi…
Yggdrasil: Bridging Dynamic Speculation and Static Runtime for Latency-Optimal Tree-Based LLM Decoding
Yue Guan, Changming Yu, Shihan Fang +8
Speculative decoding improves LLM inference by generating and verifying multiple tokens in parallel, but existing systems suffer from suboptimal performance due to a mismatch betwe…
FlashFuser: Expanding the Scale of Kernel Fusion for Compute-Intensive Operators via Inter-Core Connection
Ziyu Huang, Yangjie Zhou, Zihan Liu +8
The scaling of computation throughput continues to outpace improvements in memory bandwidth, making many deep learning workloads memory-bound. Kernel fusion is a key technique to a…
ClusterFusion: Expanding Operator Fusion Scope for LLM Inference via Cluster-Level Collective Primitive
Xinhao Luo, Zihan Liu, Yangjie Zhou +8
Large language model (LLM) decoding suffers from high latency due to fragmented execution across operators and heavy reliance on off-chip memory for data exchange and reduction. Th…
eLLM: Elastic Memory Management Framework for Efficient LLM Serving
Jiale Xu, Rui Zhang, Yi Xiong +11
Large Language Models are increasingly being deployed in datacenters. Serving these models requires careful memory management, as their memory usage includes static weights, dynami…
Towards High-Goodput LLM Serving with Prefill-decode Multiplexing
Yukang Chen, Weihao Cui, Han Zhao +7
Large Language Model (LLM) serving must meet stringent Service Level Objectives (SLOs) for both the prefill and decode phases. Some existing solutions disaggregate the two phases,…
VQ-LLM: High-performance Code Generation for Vector Quantization Augmented LLM Inference
Zihan Liu, Xinhao Luo, Junxian Guo +11
In this work, we design and implement VQ-LLM, an efficient fused Vector Quantization (VQ) kernel generation framework. We first introduce a software abstraction called codebook cac…
Vortex: Efficient Sample-Free Dynamic Tensor Program Optimization via Hardware-aware Strategy Space Hierarchization
Yangjie Zhou, Honglin Zhu, Qian Qiu +9
Dynamic-shape deep neural networks (DNNs) are rapidly evolving, attracting attention for their ability to handle variable input sizes in real-time applications. However, existing c…
Accelerating Generic Graph Neural Networks via Architecture, Compiler, Partition Method Co-Design
Shuwen Lu, Zhihui Zhang, Cong Guo +3
Graph neural networks (GNNs) have shown significant accuracy improvements in a variety of graph learning domains, sparking considerable research interest. To translate these accura…
DistSim: A performance model of large-scale hybrid distributed DNN training
Guandong Lu, Runzhe Chen, Yakai Wang +8
With the ever-increasing computational demand of DNN training workloads, distributed training has been widely adopted. A combination of data, model and pipeline parallelism strateg…
AdaptGear: Accelerating GNN Training via Adaptive Subgraph-Level Kernels on GPUs
Yangjie Zhou, Yaoxu Song, Jingwen Leng +7
Graph neural networks (GNNs) are powerful tools for exploring and learning from graph structures and features. As such, achieving high-performance execution for GNNs becomes crucia…
Efficient Adaptive Activation Rounding for Post-Training Quantization
Zhengyi Li, Cong Guo, Zhanda Zhu +5
Post-training quantization attracts increasing attention due to its convenience in deploying quantized neural networks. Although rounding-to-nearest remains the prevailing method f…
Characterizing and Demystifying the Implicit Convolution Algorithm on Commercial Matrix-Multiplication Accelerators
Yangjie Zhou, Mengtian Yang, Cong Guo +5
Many of today's deep neural network accelerators, e.g., Google's TPU and NVIDIA's tensor core, are built around accelerating the general matrix multiplication (i.e., GEMM). However…
Balancing Efficiency and Flexibility for DNN Acceleration via Temporal GPU-Systolic Array Integration
Cong Guo, Yangjie Zhou, Jingwen Leng +6
The research interest in specialized hardware accelerators for deep neural networks (DNN) spikes recently owing to their superior performance and efficiency. However, today's DNN a…