all AI news
vAttention: Dynamic Memory Management for Serving LLMs without PagedAttention
May 8, 2024, 4:42 a.m. | Ramya Prabhu, Ajay Nayak, Jayashree Mohan, Ramachandran Ramjee, Ashish Panwar
cs.LG updates on arXiv.org arxiv.org
Abstract: Efficient use of GPU memory is essential for high throughput LLM inference. Prior systems reserved memory for the KV-cache ahead-of-time, resulting in wasted capacity due to internal fragmentation. Inspired by OS-based virtual memory systems, vLLM proposed PagedAttention to enable dynamic memory allocation for KV-cache. This approach eliminates fragmentation, enabling high-throughput LLM serving with larger batch sizes. However, to be able to allocate physical memory dynamically, PagedAttention changes the layout of KV-cache from contiguous virtual memory …
abstract arxiv cache capacity cs.lg cs.os dynamic fragmentation gpu inference llm llms management memory prior systems type virtual
More from arxiv.org / cs.LG updates on arXiv.org
Jobs in AI, ML, Big Data
Senior Machine Learning Engineer
@ GPTZero | Toronto, Canada
ML/AI Engineer / NLP Expert - Custom LLM Development (x/f/m)
@ HelloBetter | Remote
Doctoral Researcher (m/f/div) in Automated Processing of Bioimages
@ Leibniz Institute for Natural Product Research and Infection Biology (Leibniz-HKI) | Jena
Seeking Developers and Engineers for AI T-Shirt Generator Project
@ Chevon Hicks | Remote
Principal Data Architect - Azure & Big Data
@ MGM Resorts International | Home Office - US, NV
GN SONG MT Market Research Data Analyst 11
@ Accenture | Bengaluru, BDC7A