arXiv subjects

Weihao Cui

Publications and source records attributed to Weihao Cui.

14 recordsLinked to original sources

arXiv2026

MuxTune: Efficient Multi-Task LLM Fine-Tuning in Multi-Tenant Datacenters via Spatial-Temporal Backbone Multiplexing

Parameter-Efficient Fine-Tuning (PEFT) is widely applied as the backend of fine-tuning APIs for large language model (LLM) customization in datacenters. Service providers deploy separate instances for individual PEFT tasks, giving rise to prominent resource inefficiencies, including (1) GPU underutilization from small-scale, PEFT-native operators and (2) device stalls from communication delays and data dependencies in parallelized execution. To address these issues, this paper presents MuxTune, a fine-tuning system that enables resource-efficient concurrent execution of multiple PEFT tasks. The key idea is to multiplex the backbone across independent tasks in a spatial-temporal manner for improved utilization and reduced stalls. Building on flexible, modularized backbone sharing via unified PEFT representations, MuxTune proposes hierarchical co-scheduling scheme with task, operator, and data-level optimizations. Specifically, it fuses tasks through a hybrid of spatial and temporal multiplexing, and orchestrates multi-task operator execution in two-tiered hybrid parallelism. Additionally, MuxTune employs chunk-based data alignment to mitigate inter-task ineffective tokens. Experimental results demonstrate that MuxTune achieves up to $2.33\times$ higher throughput and $5.29\times$ memory reduction compared to three state-of-the-art baselines.

cs.DC↗

arXiv2025

Harli: SLO-Aware Co-location of LLM Inference and PEFT-based Finetuning on Model-as-a-Service Platforms

Large language models (LLMs) are increasingly deployed under the Model-as-a-Service (MaaS) paradigm. To meet stringent quality-of-service (QoS) requirements, existing LLM serving systems disaggregate the prefill and decode phases of inference. However, decode instances often experience low GPU utilization due to their memory-bound nature and insufficient batching in dynamic workloads, leaving compute resources underutilized. We introduce Harli, a serving system that improves GPU utilization by co-locating parameter-efficient finetuning (PEFT) tasks with LLM decode instances. PEFT tasks are compute-bound and memory-efficient, making them ideal candidates for safe co-location. Specifically, Harli addresses key challenges--limited memory and unpredictable interference--using three components: a unified memory allocator for runtime memory reuse, a two-stage latency predictor for decode latency modeling, and a QoS-guaranteed throughput-maximizing scheduler for throughput maximization. Experimental results show that Harli improves the finetune throughput by 46.2% on average (up to 92.0%) over state-of-the-art serving systems, while maintaining strict QoS guarantees for inference decode.

cs.DC↗

arXiv2025

Boosting Embodied AI Agents through Perception-Generation Disaggregation and Asynchronous Pipeline Execution

Embodied AI systems operate in dynamic environments, requiring seamless integration of perception and generation modules to process high-frequency input and output demands. Traditional sequential computation patterns, while effective in ensuring accuracy, face significant limitations in achieving the necessary "thinking" frequency for real-world applications. In this work, we present Auras, an algorithm-system co-designed inference framework to optimize the inference frequency of embodied AI agents. Auras disaggregates the perception and generation and provides controlled pipeline parallelism for them to achieve high and stable throughput. Faced with the data staleness problem that appears when the parallelism is increased, Auras establishes a public context for perception and generation to share, thereby promising the accuracy of embodied agents. Experimental results show that Auras improves throughput by 2.54x on average while achieving 102.7% of the original accuracy, demonstrating its efficacy in overcoming the constraints of sequential computation and providing high throughput.

cs.AI↗

arXiv2025

Towards High-Goodput LLM Serving with Prefill-decode Multiplexing

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, causing potential resource idleness or compute redundancy. Others split the prefill phase into chunks and fuse it with decode iteration, creating a dilemma between SLO compliance and high utilization. To address these issues, an efficient serving system should dynamically adapt compute allocation, decouple compute from memory management, and execute prefill and decode independently. We present MuxWise, an LLM serving framework that adopts a new paradigm, intra-GPU prefill-decode multiplexing, to meet these requirements. To fully exploit the paradigm, MuxWise integrates a bubble-less multiplex engine, a contention-tolerant estimator, and an SLO-aware dispatcher. Evaluation shows that MuxWise improves peak throughput under SLO guarantees by an average of 2.20x (up to 3.06x) over state-of-the-art baselines.

cs.OS↗

arXiv2025

Efficient Function-as-a-Service for Large Language Models with TIDAL

Large Language Model (LLM) applications have emerged as a prominent use case for Function-as-a-Service (FaaS) due to their high computational demands and sporadic invocation patterns. However, serving LLM functions within FaaS frameworks faces significant GPU-side cold start. A fundamental approach involves leveraging a template with function state saved on GPUs to bypass the cold start for new invocations. Yet, this approach struggles with the high GPU footprint, dynamic initialization behaviors, and lazy GPU kernel loading inherent in LLM functions, primarily due to a lack of insight into the underlying execution details. In this paper, we introduce TIDAL, an optimized FaaS framework for LLM applications that achieves fast startups by tracing fine-grained execution paths. By utilizing the traced execution details, TIDAL generates adaptive function templates, effectively breaking startup barriers for LLM functions. Extensive evaluations demonstrate that TIDAL reduces cold start latency by $1.79\times\text{\textasciitilde}2.11\times$ and improves the $95\%$-ile time-to-first-token by $76.0\%$, surpassing state-of-the-art methods.

cs.OS↗

arXiv2025

VQ-LLM: High-performance Code Generation for Vector Quantization Augmented LLM Inference

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 cache to optimize codebook access efficiency and support the integration of VQ with various computations. The codebook cache adaptively stores different entries across the GPU's memory hierarchy, including off-chip global memory, on-chip shared memory, and registers. Centered around the codebook cache, we design an efficient computation engine that optimizes memory traffic during computations involving codebooks. This compute engine adopts the codebook-centric dataflow and fusion optimizations. Additionally, we provide adaptive heuristics to tailor parameter selection in our optimizations to diverse VQ configurations. Our optimizations achieve an average latency reduction of 46.13% compared to unoptimized versions. Compared to existing open-source implementations, our methods decrease latency by 64.36% to 99.1%. A final comparison with state-of-the-art element-wise quantization methods like AWQ and KVQuant shows that our VQ-LLM is practically viable, achieving latencies close or even better latencies to those at equivalent bit-widths, potentially offering greater accuracy.

cs.DC↗

arXiv2025

Comet: Fine-grained Computation-communication Overlapping for Mixture-of-Experts

Mixture-of-experts (MoE) has been extensively employed to scale large language models to trillion-plus parameters while maintaining a fixed computational cost. The development of large MoE models in the distributed scenario encounters the problem of large communication overhead. The inter-device communication of a MoE layer can occupy 47% time of the entire model execution with popular models and frameworks. Therefore, existing methods suggest the communication in a MoE layer to be pipelined with the computation for overlapping. However, these coarse grained overlapping schemes introduce a notable impairment of computational efficiency and the latency concealing is sub-optimal. To this end, we present COMET, an optimized MoE system with fine-grained communication-computation overlapping. Leveraging data dependency analysis and task rescheduling, COMET achieves precise fine-grained overlapping of communication and computation. Through adaptive workload assignment, COMET effectively eliminates fine-grained communication bottlenecks and enhances its adaptability across various scenarios. Our evaluation shows that COMET accelerates the execution of a single MoE layer by $1.96\times$ and for end-to-end execution, COMET delivers a $1.71\times$ speedup on average. COMET has been adopted in the production environment of clusters with ten-thousand-scale of GPUs, achieving savings of millions of GPU hours.

cs.DC↗

arXiv2025

Flare: Anomaly Diagnostics for Divergent LLM Training in GPU Clusters of Thousand-Plus Scale

The rapid proliferation of large language models has driven the need for efficient GPU training clusters. However, it is challenging due to the frequent occurrence of training anomalies. Since existing diagnostic tools are narrowly tailored to specific issues, there are gaps in their ability to address anomalies spanning the entire training stack. In response, we introduce Flare, a diagnostic framework designed for distributed LLM training at scale. Flare first integrates a lightweight tracing daemon for full-stack and backend-extensible tracing. Additionally, it features a diagnostic engine that automatically diagnoses anomalies, with a focus on performance regressions. The deployment of Flare across 6,000 GPUs has demonstrated significant improvements in pinpointing deficiencies in real-world scenarios, with continuous operation for over eight months.

cs.OS↗

arXiv2024

Vortex: Efficient Sample-Free Dynamic Tensor Program Optimization via Hardware-aware Strategy Space Hierarchization

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 compilation optimization methods for such networks often rely heavily on predefined samples to guide the compilation process, which restricts their adaptability and efficiency. These sample-driven methods struggle to efficiently manage the diverse and unpredictable shapes encountered in real-world scenarios, often resulting in suboptimal performance. To tackle these issues, we introduce Vortex, a hardware-driven and sample-free compiler tailored for dynamic-shape tensor programs. Vortex capitalizes on detailed hardware information and hierarchizes the strategy space to facilitate high-performance code generation without relying on runtime shape samples. It features a unique bidirectional compilation workflow, combining top-down abstraction for aligning tensor program execution with hardware hierarchies and bottom-up kernel construction to narrow the search space, enabling Vortex to achieve remarkable efficiency. Comprehensive evaluations confirm that Vortex reduces compilation time by $176\times$ compared to the existing dynamic-shape compiler. Additionally, it substantially outperforms existing vendor-provided libraries and dynamic-shape compilers on both CPU and GPU platforms, delivering speedups of $2.53\times$ and $3.01\times$, respectively.

cs.DC↗

arXiv2024

The CAP Principle for LLM Serving: A Survey of Long-Context Large Language Model Serving

We survey the large language model (LLM) serving area to understand the intricate dynamics between cost-efficiency and accuracy, which is magnified by the growing need for longer contextual understanding when deploying models at a massive scale. Our findings reveal that works in this space optimize along three distinct but conflicting goals: improving serving context length (C), improving serving accuracy (A), and improving serving performance (P). Drawing inspiration from the CAP theorem in databases, we propose a CAP principle for LLM serving, which suggests that any optimization can improve at most two of these three goals simultaneously. Our survey categorizes existing works within this framework. We find the definition and continuity of user-perceived measurement metrics are crucial in determining whether a goal has been met, akin to prior CAP databases in the wild. We recognize the CAP principle for LLM serving as a guiding principle, rather than a formal theorem, to inform designers of the inherent and dynamic trade-offs in serving models. As serving accuracy and performance have been extensively studied, this survey focuses on works that extend serving context length and address the resulting challenges.

cs.DB↗

arXiv2024

Towards Fast Setup and High Throughput of GPU Serverless Computing

Integrating GPUs into serverless computing platforms is crucial for improving efficiency. However, existing solutions for GPU-enabled serverless computing platforms face two significant problems due to coarse-grained GPU management: long setup time and low function throughput. To address these issues, we propose SAGE, a GPU serverless framework with fast setup and high throughput. First, based on the data knowability of GPU function ahead of actual execution, SAGE first devises the parallelized function setup mechanism, which parallelizes the data preparation and context creation. In this way, SAGE achieves fast setup of GPU function invocations.Second, SAGE further proposes the sharing-based memory management mechanism, which shares the read-only memory and context memory across multiple invocations of the same function. The memory sharing mechanism avoids repeated data preparation and then unnecessary data-loading contention. As a consequence, the function throughput could be improved. Our experimental results show that SAGE reduces function duration by 11.3X and improves function density by 1.22X compared to the state-of-the-art serverless platform.

cs.DC↗

arXiv2024

Arena: Efficiently Training Large Models via Dynamic Scheduling and Adaptive Parallelism Co-Design

Efficiently training large-scale models (LMs) in GPU clusters involves two separate avenues: inter-job dynamic scheduling and intra-job adaptive parallelism (AP). However, existing dynamic schedulers struggle with large-model scheduling due to the mismatch between static parallelism (SP)-aware scheduling and AP-based execution, leading to cluster inefficiencies such as degraded throughput and prolonged job queuing. This paper presents Arena, a large-model training system that co-designs dynamic scheduling and adaptive parallelism to achieve high cluster efficiency. To reduce scheduling costs while improving decision quality, Arena designs low-cost, disaggregated profiling and AP-tailored, load-aware performance estimation, while unifying them by sharding the joint scheduling-parallelism optimization space via a grid abstraction. Building on this, Arena dynamically schedules profiled jobs in elasticity and heterogeneity dimensions, and executes them using efficient AP with pruned search space. Evaluated on heterogeneous testbeds and production workloads, Arena reduces job completion time by up to $49.3\%$ and improves cluster throughput by up to $1.60\times$.

cs.DC↗

arXiv2024

Accelerating Sparse DNNs Based on Tiled GEMM

Network pruning can reduce the computation cost of deep neural network (DNN) models. However, sparse models often produce randomly-distributed weights to maintain accuracy, leading to irregular computations. Consequently, unstructured sparse models cannot achieve meaningful speedup on commodity hardware built for dense matrix computations. Accelerators are usually modified or designed with structured sparsity-optimized architectures for exploiting sparsity. For example, the Ampere architecture introduces a sparse tensor core, which adopts the 2:4 sparsity pattern. We propose a pruning method that builds upon the insight that matrix multiplication generally breaks the large matrix into multiple smaller tiles for parallel execution. We present the tile-wise sparsity pattern, which maintains a structured sparsity pattern at the tile level for efficient execution but allows for irregular pruning at the global scale to maintain high accuracy. In addition, the tile-wise sparsity is implemented at the global memory level, and the 2:4 sparsity executes at the register level inside the sparse tensor core. We can combine these two patterns into a tile-vector-wise (TVW) sparsity pattern to explore more fine-grained sparsity and further accelerate the sparse DNN models. We evaluate the TVW on the GPU, achieving averages of $1.85\times$, $2.75\times$, and $22.18\times$ speedups over the dense model, block sparsity, and unstructured sparsity.

cs.DC↗

arXiv2023

AdaptGear: Accelerating GNN Training via Adaptive Subgraph-Level Kernels on GPUs

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 crucially important. Prior works have proposed to explore the sparsity (i.e., low density) in the input graph to accelerate GNNs, which uses the full-graph-level or block-level sparsity format. We show that they fail to balance the sparsity benefit and kernel execution efficiency. In this paper, we propose a novel system, referred to as AdaptGear, that addresses the challenge of optimizing GNNs performance by leveraging kernels tailored to the density characteristics at the subgraph level. Meanwhile, we also propose a method that dynamically chooses the optimal set of kernels for a given input graph. Our evaluation shows that AdaptGear can achieve a significant performance improvement, up to $6.49 \times$ ($1.87 \times$ on average), over the state-of-the-art works on two mainstream NVIDIA GPUs across various datasets.

cs.DC↗