V-Rex: Real-Time Streaming Video LLM Acceleration via Dynamic KV Cache Retrieval

• URL: http://arxiv.org/abs/2512.12284v2
• Date: Fri, 19 Dec 2025 08:02:44 GMT
• Title: V-Rex: Real-Time Streaming Video LLM Acceleration via Dynamic KV Cache Retrieval
• Authors: Donghyuk Kim, Sejeong Yang, Wonjin Shin, Joo-Young Kim,
• Abstract summary: Streaming video large language models (LLMs) are increasingly used for real-time multimodal tasks such as video captioning, question answering, conversational agents, and augmented reality.<n>These models face fundamental memory and computational challenges because their key-value ( KV) caches grow substantially with continuous streaming video input.<n>We propose V-Rex, the first software- hardware co-designed accelerator that addresses both algorithmic and hardware bottlenecks in streaming video LLM inference.
• Score: 1.677021230191566
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Streaming video large language models (LLMs) are increasingly used for real-time multimodal tasks such as video captioning, question answering, conversational agents, and augmented reality. However, these models face fundamental memory and computational challenges because their key-value (KV) caches grow substantially with continuous streaming video input. This process requires an iterative prefill stage, which is a unique feature of streaming video LLMs. Due to its iterative prefill stage, it suffers from significant limitations, including extensive computation, substantial data transfer, and degradation in accuracy. Crucially, this issue is exacerbated for edge deployment, which is the primary target for these models. In this work, we propose V-Rex, the first software-hardware co-designed accelerator that comprehensively addresses both algorithmic and hardware bottlenecks in streaming video LLM inference. At its core, V-Rex introduces ReSV, a training-free dynamic KV cache retrieval algorithm. ReSV exploits temporal and spatial similarity-based token clustering to reduce excessive KV cache memory across video frames. To fully realize these algorithmic benefits, V-Rex offers a compact, low-latency hardware accelerator with a dynamic KV cache retrieval engine (DRE), featuring bit-level and early-exit based computing units. V-Rex achieves unprecedented real-time of 3.9-8.3 FPS and energy-efficient streaming video LLM inference on edge deployment with negligible accuracy loss. While DRE only accounts for 2.2% power and 2.0% area, the system delivers 1.9-19.7x speedup and 3.1-18.5x energy efficiency improvements over AGX Orin GPU. This work is the first to comprehensively tackle KV cache retrieval across algorithms and hardware, enabling real-time streaming video LLM inference on resource-constrained edge devices.

Related papers

• Flow caching for autoregressive video generation [72.10021661412364]
  We present FlowCache, the first caching framework specifically designed for autoregressive video generation.<n>Our method achieves remarkable speedups of 2.38 times on MAGI-1 and 6.7 times on SkyReels-V2, with negligible quality degradation.
  arXiv  Detail & Related papers  (2026-02-11T13:11:04Z)
• Quant VideoGen: Auto-Regressive Long Video Generation via 2-Bit KV-Cache Quantization [83.406036390582]
  Quant VideoGen (QVG) is a training free KV cache quantization framework for autoregressive video diffusion models.<n>It reduces KV memory by up to 7.0 times with less than 4% end to end latency overhead.<n>It consistently outperforms existing baselines in generation quality.
  arXiv  Detail & Related papers  (2026-02-03T00:54:32Z)
• VidLaDA: Bidirectional Diffusion Large Language Models for Efficient Video Understanding [52.69880888587866]
  Current Video Large Language Models (Video LLMs) typically encode frames via a encoder vision and employ an autoregressive (AR) LLM for understanding and generation.<n>We propose VidLaDA, a Diffusion Video LLM based on Language Models (DLMs) that leverages bidirectional attention to unlock comprehensive modeling and decode tokens in parallel.<n>Experiments show VidLaDA rivals state-of-the-art AR baselines and outperforms DLM baselines, with MARS-Cache delivering over 12x speedup without compromising accuracy.
  arXiv  Detail & Related papers  (2026-01-25T15:02:01Z)
• HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding [92.59317281526239]
  HERMES is a training-free architecture for real-time and accurate understanding of video streams.<n>Hermes reuses a compact KV cache, enabling efficient streaming understanding under resource constraints.<n>Hermes achieves superior or comparable accuracy across all benchmarks, with up to 11.4% gains on streaming datasets.
  arXiv  Detail & Related papers  (2026-01-21T07:26:15Z)
• StreamKV: Streaming Video Question-Answering with Segment-based KV Cache Retrieval and Compression [95.59657871147846]
  We propose textbfStreamKV, a framework that seamlessly equips Video-LLMs with advanced KV cache retrieval and compression.<n>Experiments on public StreamingVQA benchmarks demonstrate that StreamKV significantly outperforms existing Online Video-LLMs.
  arXiv  Detail & Related papers  (2025-11-10T16:25:03Z)
• Video-XL-2: Towards Very Long-Video Understanding Through Task-Aware KV Sparsification [9.615466029246694]
  Video-XL-2 is a novel MLLM that delivers superior cost-effectiveness for long-video understanding based on task-aware KV sparsification.<n>It is capable of processing over 10,000 frames on a single NVIDIA A100 (80GB) GPU and thousands of frames in just a few seconds.
  arXiv  Detail & Related papers  (2025-06-24T01:19:56Z)
• FCA2: Frame Compression-Aware Autoencoder for Modular and Fast Compressed Video Super-Resolution [68.77813885751308]
  State-of-the-art (SOTA) compressed video super-resolution (CVSR) models face persistent challenges, including prolonged inference time, complex training pipelines, and reliance on auxiliary information.<n>We propose an efficient and scalable solution inspired by the structural and statistical similarities between hyperspectral images (HSI) and video data.<n>Our approach introduces a compression-driven dimensionality reduction strategy that reduces computational complexity, accelerates inference, and enhances the extraction of temporal information across frames.
  arXiv  Detail & Related papers  (2025-06-13T07:59:52Z)
• dKV-Cache: The Cache for Diffusion Language Models [53.85291644298835]
  Diffusion Language Models (DLMs) have been seen as a promising competitor for autoregressive language models.<n>We propose a KV-cache-like mechanism, delayed KV-Cache, for the denoising process of DLMs.<n>Our approach is motivated by the observation that different tokens have distinct representation dynamics throughout the diffusion process.
  arXiv  Detail & Related papers  (2025-05-21T17:32:10Z)
• LeanVAE: An Ultra-Efficient Reconstruction VAE for Video Diffusion Models [17.29580459404157]
  We propose LeanVAE, a novel and ultra-efficient Video VAE framework.<n>Our model offers up to 50x fewer FLOPs and 44x faster inference speed.<n>Our experiments validate LeanVAE's superiority in video reconstruction and generation.
  arXiv  Detail & Related papers  (2025-03-18T14:58:59Z)
• STORM: Token-Efficient Long Video Understanding for Multimodal LLMs [116.4479155699528]
  STORM is a novel architecture incorporating a dedicated temporal encoder between the image encoder and the Video-LLMs.<n>We show that STORM achieves state-of-the-art results across various long video understanding benchmarks.
  arXiv  Detail & Related papers  (2025-03-06T06:17:38Z)
• Streaming Video Question-Answering with In-context Video KV-Cache Retrieval [10.990431921021585]
  We propose ReKV, a training-free approach that enables efficient streaming video question-answering (StreamingVQA)<n>Our approach analyzes long videos in a streaming manner, allowing for prompt responses as soon as user queries are received.
  arXiv  Detail & Related papers  (2025-03-01T15:53:33Z)
• Ca2-VDM: Efficient Autoregressive Video Diffusion Model with Causal Generation and Cache Sharing [66.66090399385304]
  Ca2-VDM is an efficient autoregressive VDM with Causal generation and Cache sharing.<n>For causal generation, it introduces unidirectional feature computation, which ensures that the cache of conditional frames can be precomputed in previous autoregression steps.<n>For cache sharing, it shares the cache across all denoising steps to avoid the huge cache storage cost.
  arXiv  Detail & Related papers  (2024-11-25T13:33:41Z)
• SparseTem: Boosting the Efficiency of CNN-Based Video Encoders by Exploiting Temporal Continuity [19.900719882624028]
  We propose a memory-efficient scheduling method to eliminate memory overhead and an online adjustment mechanism to minimize accuracy degradation.<n>SparseTem achieves speedup of 1.79x for EfficientDet and 4.72x for CRNN, with minimal accuracy drop and no additional memory overhead.
  arXiv  Detail & Related papers  (2024-10-28T07:13:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.