论文检索

General Machine Learning · Hardware and Software

Nina Wiedemann, Quentin Leboutet, Michael Paulitsch, Diana Wofk, Benjamin Ummenhofer

GPU kernel optimization challenges LLMs beyond standard coding tasks, as it requires an understanding of hardware architecture, parallel computing optimization strategies, and profiling outputs. However, most existing approaches leveraging LLMs for kernel generation apply standard prompting and feedback loops, considering hardware only through profiling feedback. We introduce KernelFoundry, an evolutionary framework that efficiently explores the space of GPU kernels through (1) MAP-Elites quality-diversity search with kernel-specific behavioral dimensions to sustain exploration; (2) meta-prompt evolution that co-evolves prompts with kernels to uncover task-specific optimization strategies, and (3) a template-based parameter optimization approach to tune kernels to inputs and hardware. We evaluate this framework on KernelBench, robust-kbench and custom tasks, generating SYCL kernels as a cross-platform GPU programming paradigm, and CUDA kernels for comparison to prior work. Our approach consistently outperforms the baseline methods and achieves an average speedup of 2.3 on KernelBench for SYCL. Moreover, KernelFoundry is implemented as a distributed framework with remote access to diverse hardware, allowing quick benchmarking and featuring a flexible user input layer to support kernel generation for a for a wide range of real use cases beyond benchmarking.

Applications · Everything Else

Shvetank Prakash, Andrew Cheng, Arya Tschand, Mark Mazumder, Varun Gohil, Jeffrey Ma, Jason Yik, Zishen Wan, Jessica A. Quaye, Elisavet Alvanaki 等

The field of computer architecture, which bridges high-level software abstractions and low-level hardware implementations, remains absent from current large language model (LLM) evaluations. To this end, we present QuArch (pronounced 'quark'), the first benchmark designed to facilitate the development and evaluation of LLM knowledge and reasoning capabilities specifically in computer architecture. QuArch provides a comprehensive collection of 2,671 expert-validated question-answer (QA) pairs covering various aspects of computer architecture, including processor design, memory systems, and interconnection networks. Our evaluation reveals that while frontier models possess domain-specific knowledge, they struggle with skills that require higher-order thinking in computer architecture. Frontier model accuracies vary widely (from 34% to 73%) on these advanced questions, highlighting persistent gaps in architectural reasoning across analysis, design, and implementation QAs. Furthermore, via fine-tuning we find that QuArch can translate to improved performance on a realistic memory hierarchy design task, resulting in up to 1.99x more area-efficient solutions and up to 40% more viable solutions overall. By holistically assessing fundamental skills, QuArch provides a foundation for building and measuring LLM capabilities that can accelerate innovation in computing systems.

Deep Learning · Large Language Models

Venmugil Elango, Nidhi Bhatia, Roger Waleffe, Rasoul Shafipour, Tomer Asida, Abhinav Khattar, Nave Assaf, Maximilian Golub, Joseph Guman, Tiyasa Mitra 等

Mixture-of-Experts (MoEs) have become a central component of many state-of-the-art open-source and proprietary large language models. Despite their widespread adoption, it remains unclear how close existing MoE architectures are to optimal with respect to inference cost, as measured by accuracy per floating-point operation and per parameter. In this work, we revisit MoE design from a hardware-software co-design perspective, grounded in empirical and theoretical considerations. We characterize key performance bottlenecks across diverse deployment regimes, spanning offline high-throughput execution and online, latency-critical inference. Guided by these insights, we introduce LatentMoE, a new model architecture resulting from systematic design exploration and optimized for maximized accuracy per unit of compute. Empirical design space exploration at scales of up to 95B parameters and over a 1T-token training horizon, together with supporting theoretical analysis, show that LatentMoE consistently outperforms standard MoE architectures in terms of accuracy per FLOP and per parameter.

Deep Learning · Large Language Models

Xiaoxue Zhu, Jilin Hu, Fuyuan Zhang, Jianyu Zhang, Yongwang Zhao

Recent advances in large language models have accelerated neural theorem proving (NTP). Isabelle is a mature and important formal theorem prover that has been widely used in software and hardware verification. However, progress in the Isabelle setting remains limited. Existing approaches either optimize search strategies or train on highly imbalanced raw proof corpora. At the same time, the specialized structure of Isabelle proofs limits the effectiveness of general-purpose data selection methods. To address these challenges, we adopt a data-centric framework for neural theorem proving in Isabelle. We characterize high-quality formal proof data along three complementary dimensions—proof complexity, semantic coverage, and reasoning diversity (PSR)—and propose a PSR-guided data selection pipeline to construct a compact, high-quality training subset. In addition, we leverage verifier feedback as a dynamic data signal during inference, introducing a dynamic feedback-based prompt optimization that iteratively incorporates Isabelle verifier feedback to guide proof generation. We construct and release a 4k high-quality Isabelle dataset based on the PSR criterion. On the miniF2F-test, fine-tuning solely on PSR-selected data achieves 84.8% Pass@64. When further combined with dynamic feedback–based prompt optimization, the full framework improves performance to 90.6% Pass@64, establishing a new state of the art for neural theorem proving in Isabelle.

General Machine Learning · Hardware and Software

Zongze Li, Jingyu Liu, Zach Xu, Yineng Zhang, Tahseen Rabbani, Ce Zhang

Prefill-Decode (PD) disaggregation has become the standard architecture for modern LLM inference engines, which alleviates the interference of two distinctive workloads. With the growing demand for multi-turn interactions in chatbots and agentic systems, we re-examined PD in this case and found two fundamental inefficiencies: (1) every turn requires prefilling the new prompt and response from the last turn, and (2) repeated KV transfers between prefill and decode nodes saturate the bandwidth, leading to high latency and even service degradation. Our key insight is that not all prefill operations are equally disruptive: append-prefill---processing only the new input tokens while reusing cached KV states---incurs substantially less decoding slowdown than full prefill. This motivates routing append-prefill to decode nodes locally. However, through comprehensive analysis, we show that no single fixed routing strategy satisfies all Service Level Objectives (SLOs) simultaneously. Based on this insight, we propose Prefill Prefill-capable Decode (PPD) disaggregation, a dynamic routing system that decides when to process Turn 2+ requests locally on decode nodes using cached KV states. PPD adapts to varying SLOs via configurable weights and seamlessly integrates with traditional PD deployments. With extensive evaluations, we show that PPD reduces Turn 2+ time-to-first-token (TTFT) by 68% while maintaining competitive time-per-output-token (TPOT), effectively alleviating KV transfer congestion under high load. We believe PPD represents a flexible and efficient paradigm for multi-turn LLM serving.

General Machine Learning · Hardware and Software

Duc Hoang, Aarush Gupta, Philip Harris

Ultrafast online learning is essential for high-frequency systems, such as controls for quantum computing and nuclear fusion, where adaptation must occur on sub-microsecond timescales. Meeting these requirements demands low-latency, fixed-precision computation under strict memory constraints, a regime in which conventional Multi-Layer Perceptrons (MLPs) are both inefficient and numerically unstable. We identify key properties of Kolmogorov-Arnold Networks (KANs) that align with these constraints. Specifically, we show that: (i) KAN updates exploiting B-spline locality are sparse, enabling superior on-chip resource scaling, and (ii) KANs are inherently robust to fixed-point quantization. By implementing fixed-point online training on Field-Programmable Gate Arrays (FPGAs), a representative platform for on-chip computation, we demonstrate that KAN-based online learners are significantly more efficient and expressive than MLPs across a range of low-latency and resource-constrained tasks. To our knowledge, this work is the first to demonstrate model-free online learning at sub-microsecond latencies.

General Machine Learning · Hardware and Software

Caleb Winston, Ron Yifeng Wang, Azalia Mirhoseini, Christoforos Kozyrakis

Computer-use agents (CUA) automate tasks specified with natural language such as "order the cheapest item from Taco Bell", by generating sequences of calls to tools such as click, type, and scroll on a browser. Current CUA implementations follow a sequential fetch-screenshot-execute loop where each iteration requires an LLM call, resulting in high latency and frequent errors from incorrect tool use. We present agent JIT compilation, an alternative that compiles task descriptions directly into executable code that is free to include LLM calls, tool calls, and parallelization. Our approach comprises three components: (1) JIT-Planner, which generates multiple code plans from a task, validates each against tool specifications, and selects the minimum-cost candidate; (2) JIT-Scheduler, which explores parallelization strategies via Monte Carlo cost estimation from learned latency distributions; and (3) an invariant-enforcing tool protocol specifying pre/postcondition state requirements that reduces the rate of generating plans with incorrect tool use. Evaluation across 5 applications demonstrates that JIT-Planner achieves $10.4\times$ speedup and $+28\%$ accuracy improvement over Browser-Use, while JIT-Scheduler achieves $2.6\times$ speedup and $+9\%$ accuracy improvement over OpenAI CUA.

General Machine Learning · Hardware and Software

Xiao Liu, Lijun Zhang, Deepak Ganesan, Hui Guan

Multi-device inference can reduce Transformer latency by parallelizing computation. However, existing methods require high inter-device bandwidth, making them impractical for bandwidth-constrained environments. We present ASTRA, a communication-efficient framework that integrates sequence parallelism with mixed-precision attention, where non-local token embeddings are transmitted as low-bit vector-quantized codes while local attention remains full precision. To preserve accuracy under aggressive compression, ASTRA introduces Noise-Augmented Quantization and Distributed Class Tokens. Across vision and language models (e.g., ViT and GPT2), ASTRA achieves up to 2.64$\times$ speedup over single-device inference and up to 15.25$\times$ over prior multi-device baselines while operating at bandwidths as low as 10 Mbps. ASTRA remains robust on large models (e.g., Llama-3-8B) even under non-ideal network conditions such as packet loss and dynamic networks.

Deep Learning · Robustness

Aviv Adler, Jennifer Tang

It is well-known in the field of lossless data compression that probabilistic next-symbol prediction can be used to compress sequences of symbols. Deep neural networks are able to capture rich dependencies in data, offering a powerful means of estimating these probabilities and hence an avenue towards more effective compression algorithms. However, both compressor and decompressor must have exactly matching predictions; even small differences from non-determinism (which often happen with learned models due to hardware, software, or computation order) can lead to cascading decoding failures. In this paper, we formalize the problem of prediction mismatch in model-driven compression, and introduce Probability Matching Interval Coding (PMATIC), a model-agnostic algorithm that tolerates bounded prediction mismatch with low overhead. PMATIC works with the predicted probabilities, making it compatible as a drop-in replacement for the arithmetic encoder in model-driven compression tools. We show theoretical correctness and performance bounds for PMATIC, and validate these results on text data. These results confirm that, when paired an advanced prediction model, PMATIC is robust to prediction mismatch while achieving compression rates that out-perform standard modern compression tools.

Applications · Neuroscience, Cognitive Science

Zhanglu Yan, Jiayi Mao, Qianhui Liu, Fanfan Li, Tao Luo, Gang Pan, Bowen Zhu, Weng-Fai Wong

Spiking neural networks (SNNs) promise high energy efficiency, particularly with time-to-first-spike (TTFS) encoding, which maximizes sparsity by emitting at most one spike per neuron. However, this energy advantage is often unrealized because inference requires evaluating a temporal decay function and then multiplying the result by the synaptic weights. This paper challenges this costly approach by repurposing a physical hardware `bug', namely, the natural signal decay in optoelectronic devices, as the core computation of TTFS. We fabricated a custom indium oxide optoelectronic synapse that demonstrates how its intrinsic physical decay directly implements the required temporal function. By treating the device's analog output as the fused product of the synaptic weight and temporal decay, optoelectronic synaptic TTFS (named Otters) eliminates these expensive digital operations. To use the Otters' paradigm in complex architectures such as the transformer, which are challenging to train directly due to sparsity, we introduce a novel quantized neural network-to-SNN conversion algorithm. This complete hardware-software co-design enables our model to achieve state-of-the-art accuracy across seven GLUE benchmark datasets and demonstrates a 1.77$\times$ improvement in energy efficiency over previous leading SNNs, based on a comprehensive analysis of compute, data movement, and memory access costs using energy measurements from a commercial 22nm process. Our work thus establishes a new paradigm for energy-efficient SNNs that translates fundamental device physics directly into powerful computational primitives. All codes and data are open source.

Optimization · Large Scale, Parallel and Distributed

Jingyi Zhao, Linxin Yang, Haohua Zhang, Qile He, Tian Ding

Dynamic programming (DP) is central to combinatorial optimization, optimal control, and reinforcement learning, yet its perceived sequentiality has long hindered scalability. We introduce a general-purpose GPU framework that reformulates broad classes of forward DP recursions as batched min--plus matrix--vector products over layered DAGs, collapsing actions into masked state-to-state transitions that map directly to GPU kernels. This approach removes a major bottleneck in scenario-based stochastic programming (SP), where the use of DP has traditionally restricted the number of scenarios due to excessive computational cost. Our framework exposes massive parallelism across scenarios, transition layers, and, when applicable, route or action options, via self-designed GPU kernels that implement Bellman updates with warp-/block-level reductions and numerically safe masking. In a single GPU pass, these kernels can process over $10^6$ uncertainty realizations, far beyond the capacity of prior scenario-based methods. We demonstrate the approach in two canonical SP applications: (i) a vectorized split operator for the capacitated vehicle routing problem with stochastic demand, exploiting **2D** parallelism (scenarios $\times$ transitions); and (ii) a forward inventory reinsertion DP under an order-up-to policy, exploiting **3D** parallelism (scenarios $\times$ inventory transitions $\times$ route options). Across benchmarks, the implementation scales nearly linearly in the number of scenarios and achieves one to three orders of magnitude speedups over multithreaded CPU baselines, yielding tighter SAA estimates and consistently stronger first-stage decisions under identical wall-clock budgets. Viewed as hardware-aware software primitives, our min--plus DP kernels offer a drop-in path to scalable, GPU-accelerated stochastic discrete optimization.

Applications · Everything Else

Yilong Li, Yijing Zeng, Shuai Zhang, Hao Zhang, Xinmiao Xiong, Jingyu Liu, Pan Hu, Suman Banerjee

Large Multimodal Models (LMMs) are inherently modular, consisting of vision and audio encoders, projectors, and large language models. Yet, they are almost always executed monolithically, which underutilizes the heterogeneous accelera- tors (NPUs, GPUs, DSPs) in modern SoCs and leads to high end-to-end latency. In this paper, we present NANOMIND, a hardware–software co-design inference framework for Large Multimodal Models (LMMs) that breaks large models into modular “bricks” (vision, language, audio, etc.) and maps each to its ideal accelera- tor. The key insight is that large models can be broken into modular components and scheduled to run on the most appropriate compute units. It performs module-level dynamic offloading across accelerators on unified-memory SoCs. By combining customized hardware design, system-level scheduling, and optimized low-bit com- putation kernels, we demonstrate our framework with a compact, battery-powered device capable of running LMMs entirely on-device. This prototype functions as a self-contained intelligent assistant that requires no network connectivity, while achieving higher throughput and superior power efficiency under strict resource constraints. The design further bypasses CPU bottlenecks and reduces redundant memory usage through token-aware buffer management and module-level coordi- nation. Our system outperforms existing implementations in resource efficiency, cutting energy consumption by 42.3% and GPU memory usage by 11.2%. This enables a battery-powered device to run LlaVA-OneVision-qwen2-05B with a camera for nearly 20.8 hours.

Applications · Robotics

Shiming Wang, Holger Caesar, Liangliang Nan, Julian Kooij

In autonomous driving, multi-modal perception tasks like 3D object detection typically rely on well-synchronized sensors, both at training and inference. However, despite the use of hardware- or software-based synchronization algorithms, perfect synchrony is rarely guaranteed: Sensors may operate at different frequencies, and real-world factors such as network latency, hardware failures, or processing bottlenecks often introduce time offsets between sensors. Such asynchrony degrades perception performance, especially for dynamic objects. To address this challenge, we propose AsyncBEV, a trainable, lightweight, and generic module to improve the robustness of 3D Bird's Eye View (BEV) object detection models against sensor asynchrony. Inspired by scene flow estimation, AsyncBEV first estimates the 2D flow from the BEV features of two different sensor modalities, taking into account the known time offset between these sensor measurements. The predicted feature flow is then used to warp and spatially align the feature maps, which we show can easily be integrated into different current BEV detector architectures (e.g., BEV grid-based and token-based). Extensive experiments demonstrate AsyncBEV improves robustness against both small and large asynchrony between LiDAR or camera sensors in both the token-based CMT and grid-based UniBEV, especially for dynamic objects. We significantly outperform the ego motion compensated CMT and UniBEV baselines, notably by $16.6$ % and $11.9$ % NDS on dynamic objects in the worst-case scenario of a $0.5 s$ time offset. Code is available at \url{https://github.com/tudelft-iv/AsyncBEV}.

Chi-Jui Ho, Yash Belhe, Steve Rotenberg, Ravi Ramamoorthi, Tzu-Mao Li, Nicholas Antipa

End-to-end optimization, which integrates differentiable optics simulators with computational algorithms, enables the joint design of hardware and software for data-driven imaging systems. However, existing methods usually compromise physical accuracy by neglecting wave optics or off-axis effects due to the high computational cost of modeling both aberration and diffraction. This limitation raises concerns about the robustness of optimized designs. In this paper, we propose a differentiable optics simulator that accurately and efficiently models aberration and diffraction in compound optics and allows us to analyze the role and impact of diffraction in end-to-end optimization. Experimental results demonstrate that compared with ray-optics-based optimization, diffraction-aware optimization improves system robustness to diffraction blur. Through accurate wave optics modeling, we also apply the simulator to optimize the Fizeau interferometer and freeform optics elements. These findings underscore the importance of accurate wave optics modeling in robust end-to-end optimization. Our code is publicly available at: https://github.com/JerryHoTaiwan/DeepWaveOptics

Linshen Liu, Boyan Su, Junyue Jiang, Guanlin Wu, Cong Guo, Ceyu Xu, Hao Frank Yang

This paper presents Edge-based Mixture of Experts (MoE) Collaborative Computing (EMC2), an optimal computing system designed for autonomous vehicles (AVs) that simultaneously achieves low-latency and high-accuracy 3D object detection. Unlike existing works, the EMC2 introduces a novel scenario-aware MoE architecture optimized for fusing complementary sparse 3D point clouds and dense 2D images to achieve robust multimodal representations for detection. Furthermore, EMC2 integrates an adaptive multimodal data bridge with multi-scale region proposing and scenario-aware routing, dynamically dispatching features to complementary experts based on object visibility and distance. In addition, EMC2 integrates joint hardware-software optimizations, including hardware resource utilization optimization and computational graph simplification, to ensure efficient and real-time inference on resource-constrained edge devices. Experiments on open-source benchmarks clearly show the EMC2 advancements as an end-to-end system. On the KITTI dataset, it achieves an average accuracy improvement of 3.58% and a 159.06% inference speedup compared to 15 baseline methods on Jetson platforms, with similar performance gains on the nuScenes dataset, highlighting its capability to advance reliable, real-time 3D object detection tasks for AVs.

Minsoo Park, Youngkwon Lim, Yangwoo Kim, Sam Richards, Min Woo Park, Kwang Pyo Choi

Over the past few decades, significant resources have been invested in developing video codecs for the storage and delivery of video data. International standards development organizations, such as MPEG (Moving Picture Experts Group) and AOM (Alliance for Open Media), have fostered large-scale competitive collaboration among industry players. While open-source software developed alongside these video codec standards has accelerated verification of the standard and enabled rapid adoption, as it is used as a starting point for implementation and future research, this practice has not been fully applied to the ecosystem of mezzanine video codecs used in high-quality video capture and post-production. To address this gap, the industry's first open source project for collaborative innovation in professional video codecs, OpenAPV, has been established. The project aims to facilitate collaborative research and development of a royalty free, open source, and open standard video codec for professional use. This paper outlines the technical aspects of the video codec, together with the open source software implementation. The codec has been successfully tested across various platforms and shows excellent R-D (Rate-Distortion) performance and coding speed, even without hardware acceleration. Furthermore, it demonstrates robust quality resilience against multiple rounds of encoding and decoding cycles, which are common in video post-production. The open source project is hosted at https://github.com/AcademySoftwareFoundation/openapv.

Xiaoyi Shi, Lin He 0004, Jiasheng Zhou, Yifan Yang 0009, Ying Liu 0024

As online cloud services expand rapidly, layer-7 load balancing has become indispensable for maintaining service availability and performance. The emergence of programmable switches with both high performance and a certain degree of flexibility has made it possible to apply programmable switches to load balancing. Nevertheless, the limited memory capacity and the relatively sluggish speed of table entry insertion and deletion of programmable switches have severely constrained their performance. To this end, we introduce Miresga, a hybrid and high-performance layer-7 load balancing system by co-designing hardware and software. The core idea of Miresga is to maximize the utilization of hardware and software resources by rationally partitioning the layer-7 load balancing task, thereby improving performance. To achieve this, Miresga offloads the elephant flows, which account for the majority of traffic, to programmable switches that excel at packet processing, and Miresga utilizes general-purpose servers with stronger computational capabilities to parse application layer protocols and apply load balancing rules. To alleviate memory pressure on the programmable switch, Miresga employs a back-end agent to handle memory-intensive tasks, working in conjunction with the programmable switch to complete the offloaded tasks. This design leverages the performance advantages of the programmable switch while avoiding bottlenecks caused by its limited memory and table insertion speed. We implement the Miresga prototype with a 3.2 Tbps Intel Tofino switch and general-purpose servers. The evaluation results show that Miresga achieves 3.9× throughput and 0.4× latency compared to software load balancing solutions. Compared to the state-of-the-art design employing programmable switches, Miresga achieves almost the same throughput and latency for delivering large objects and 5.0× throughput and 0.2× latency when transmitting small objects.

Federico Berto, Chuanbo Hua, Junyoung Park 0002, Laurin Luttmann, Yining Ma 0001, Fanchen Bu, Jiarui Wang 0002, Haoran Ye, Minsu Kim 0004, Sanghyeok Choi 等

Combinatorial optimization (CO) is fundamental to several real-world applications, from logistics and scheduling to hardware design and resource allocation. Deep reinforcement learning (RL) has recently shown significant benefits in solving CO problems, reducing reliance on domain expertise and improving computational efficiency. However, the absence of a unified benchmarking framework leads to inconsistent evaluations, limits reproducibility, and increases engineering overhead, raising barriers to adoption for new researchers. To address these challenges, we introduce RL4CO, a unified and extensive benchmark with in-depth library coverage of 27 CO problem environments and 23 state-of-the-art baselines. Built on efficient software libraries and best practices in implementation, RL4CO features modularized implementation and flexible configurations of diverse environments, policy architectures, RL algorithms, and utilities with extensive documentation. RL4CO helps researchers build on existing successes while exploring and developing their own designs, facilitating the entire research process by decoupling science from heavy engineering. We finally provide extensive benchmark studies to inspire new insights and future work. RL4CO has already attracted numerous researchers in the community and is open-sourced at https://github.com/ai4co/rl4co.

Ashutosh Gupta 0001, Thomas A. Henzinger, Konstantin Kueffner, Kaushik Mallik, David Pape

In automated decision-making, it is desirable that outputs of decision-makers be robust to slight perturbations in their inputs, a property that may be called input-output robustness. Input-output robustness appears in various different forms in the literature, such as robustness of AI models to adversarial or semantic perturbations and individual fairness of AI models that make decisions about humans. We propose runtime monitoring of input-output robustness of deployed, black-box AI models, where the goal is to design monitors that would observe one long execution sequence of the model, and would raise an alarm whenever it is detected that two similar inputs from the past led to dissimilar outputs. This way, monitoring will complement existing offline ''robustification'' approaches to increase the trustworthiness of AI decision-makers. We show that the monitoring problem can be cast as the fixed-radius nearest neighbor (FRNN) search problem, which, despite being well-studied, lacks suitable online solutions. We present our tool Clemont, which offers a number of lightweight monitors, some of which use upgraded online variants of existing FRNN algorithms, and one uses a novel algorithm based on binary decision diagrams--a data-structure commonly used in software and hardware verification. We have also developed an efficient parallelization technique that can substantially cut down the computation time of monitors for which the distance between input-output pairs is measured using the L∞norm. Using standard benchmarks from the literature of adversarial and semantic robustness and individual fairness, we perform a comparative study of different monitors in Clemont, and demonstrate their effectiveness in correctly detecting robustness violations at runtime.

Kaiyuan Hu, Yili Jin, Hao Zhou, Linfeng Du, Jiangchuan Liu, Xue Liu

Immersive video serves as a key component of eXtended Reality (XR) that aims to create and interact with simulated virtual or hybrid environments. Such a technology allows users to experience immersive sensations that transcend time and space, and meanwhile continuously providing training data for emerging technologies like Embodied AI. Thanks to the advancements in sensing, computing, and display, recent years have witnessed many excellent works for XR and related hardware or software systems. However, challenges like high creation cost, lack of immersion, and limited scalability hinder the practical application of immersive video services. Whilst recently emerged generative artificial intelligence (GenAI) provides us with new insights in tackling existing challenges. In this paper, we conduct a comprehensive survey into the recent advances and future opportunities on how GenAI can benefit immersive video services. By introducing a systematic taxonomy, we meticulously classify the pertinent techniques and applications into three well-defined categories aligned with the pipeline of immersive video service: content creation, network delivery, and client-side display. This categorization enables a structured exploration of the diverse roles on how GenAI can benefit immersive video service, providing a framework for a more comprehensive understanding and evaluation of these technologies. To the best of our knowledge, this work is the first systematic survey of GenAI in XR settings, laying a foundation for future research in this interdisciplinary domain.