vTensor: Flexible Virtual Tensor Management for Efficient LLM Serving

• URL: http://arxiv.org/abs/2407.15309v1
• Date: Mon, 22 Jul 2024 14:37:58 GMT
• Title: vTensor: Flexible Virtual Tensor Management for Efficient LLM Serving
• Authors: Jiale Xu, Rui Zhang, Cong Guo, Weiming Hu, Zihan Liu, Feiyang Wu, Yu Feng, Shixuan Sun, Changxu Shao, Yuhong Guo, Junping Zhao, Ke Zhang, Minyi Guo, Jingwen Leng,
• Abstract summary: Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests.
• Score: 53.972175896814505
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large Language Models (LLMs) are widely used across various domains, processing millions of daily requests. This surge in demand poses significant challenges in optimizing throughput and latency while keeping costs manageable. The Key-Value (KV) cache, a standard method for retaining previous computations, makes LLM inference highly bounded by memory. While batching strategies can enhance performance, they frequently lead to significant memory fragmentation. Even though cutting-edge systems like vLLM mitigate KV cache fragmentation using paged Attention mechanisms, they still suffer from inefficient memory and computational operations due to the tightly coupled page management and computation kernels. This study introduces the vTensor, an innovative tensor structure for LLM inference based on GPU virtual memory management (VMM). vTensor addresses existing limitations by decoupling computation from memory defragmentation and offering dynamic extensibility. Our framework employs a CPU-GPU heterogeneous approach, ensuring efficient, fragmentation-free memory management while accommodating various computation kernels across different LLM architectures. Experimental results indicate that vTensor achieves an average speedup of 1.86x across different models, with up to 2.42x in multi-turn chat scenarios. Additionally, vTensor provides average speedups of 2.12x and 3.15x in kernel evaluation, reaching up to 3.92x and 3.27x compared to SGLang Triton prefix-prefilling kernels and vLLM paged Attention kernel, respectively. Furthermore, it frees approximately 71.25% (57GB) of memory on the NVIDIA A100 GPU compared to vLLM, enabling more memory-intensive workloads.

Related papers

• Efficient LLM Inference with I/O-Aware Partial KV Cache Recomputation [7.204881999658682]
  Inference for Large Language Models (LLMs) is computationally demanding. To reduce the cost of auto-regressive decoding, Key-Value ( KV) caching is used to store intermediate activations. The memory required for KV caching grows rapidly, often exceeding the capacity of GPU memory. A cost-effective alternative is to offload KV cache to CPU memory, which alleviates GPU memory pressure but shifts the bottleneck to the limited bandwidth of the PCIe connection between the CPU and GPU.
  arXiv  Detail & Related papers  (2024-11-26T04:03:14Z)
• LiVOS: Light Video Object Segmentation with Gated Linear Matching [116.58237547253935]
  LiVOS is a lightweight memory network that employs linear matching via linear attention. For longer and higher-resolution videos, it matched STM-based methods with 53% less GPU memory and supports 4096p inference on a 32G consumer-grade GPU.
  arXiv  Detail & Related papers  (2024-11-05T05:36:17Z)
• Efficiently Training 7B LLM with 1 Million Sequence Length on 8 GPUs [24.066283519769968]
  Large Language Models (LLMs) have been trained using extended context lengths to foster more creative applications. We propose MEMO, a novel framework for fine-grained activation memory management. We show that MEMO achieves an average of 2.42x and 2.26x MFU compared to Megatron-LM and DeepSpeed.
  arXiv  Detail & Related papers  (2024-07-16T18:59:49Z)
• vAttention: Dynamic Memory Management for Serving LLMs without PagedAttention [8.20523619534105]
  VLLM proposed PagedAttention to enable dynamic memory allocation for KV-cache. PagedAttention changes the layout of KV-cache from contiguous virtual memory to non-contiguous virtual memory. We propose vAttention, a new approach for dynamic KV-cache memory management.
  arXiv  Detail & Related papers  (2024-05-07T16:00:32Z)
• Efficient Video Object Segmentation via Modulated Cross-Attention Memory [123.12273176475863]
  We propose a transformer-based approach, named MAVOS, to model temporal smoothness without requiring frequent memory expansion. Our MAVOS achieves a J&F score of 63.3% while operating at 37 frames per second (FPS) on a single V100 GPU.
  arXiv  Detail & Related papers  (2024-03-26T17:59:58Z)
• AI and Memory Wall [81.06494558184049]
  We show how memory bandwidth can become the dominant bottleneck for decoder models. We argue for a redesign in model architecture, training, and deployment strategies to overcome this memory limitation.
  arXiv  Detail & Related papers  (2024-03-21T04:31:59Z)
• Efficient Memory Management for Large Language Model Serving with PagedAttention [44.70922552274376]
  High throughput serving of large language models (LLMs) requires sufficiently many requests at a time. Existing systems struggle because the key-value cache ( KV cache) memory for each request is huge and grows and shrinks dynamically. We propose PagedAttention, an attention algorithm inspired by the classical virtual memory and paging techniques in operating systems.
  arXiv  Detail & Related papers  (2023-09-12T12:50:04Z)
• PowerFusion: A Tensor Compiler with Explicit Data Movement Description and Instruction-level Graph IR [10.059491353103526]
  We propose IntelliGen, a tensor compiler that can generate high-performance code for memory-intensive operators. IntelliGen considers both computation and data movement optimizations. We evaluate IntelliGen on NVIDIA GPU, AMD GPU, and Cambricon MLU, showing speedup up to 1.97x, 2.93x, and 16.91x (1.28x, 1.23x, and 2.31x on average)
  arXiv  Detail & Related papers  (2023-07-11T03:17:40Z)
• NumS: Scalable Array Programming for the Cloud [82.827921577004]
  We present NumS, an array programming library which optimize NumPy-like expressions on task-based distributed systems. This is achieved through a novel scheduler called Load Simulated Hierarchical Scheduling (LSHS) We show that LSHS enhances performance on Ray by decreasing network load by a factor of 2x, requiring 4x less memory, and reducing execution time by 10x on the logistic regression problem.
  arXiv  Detail & Related papers  (2022-06-28T20:13:40Z)

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.