TBA: Faster Large Language Model Training Using SSD-Based Activation Offloading

Autor: Wu, Kun, Park, Jeongmin Brian, Zhang, Xiaofan, Hidayetoğlu, Mert, Mailthody, Vikram Sharma, Huang, Sitao, Lumetta, Steven Sam, Hwu, Wen-mei

The growth rate of the GPU memory capacity has not been able to keep up with that of the size of large language models (LLMs), hindering the model training process. In particular, activations -- the intermediate tensors produced during forward propag

Externí odkaz: http://arxiv.org/abs/2408.10013

Accelerating Sampling and Aggregation Operations in GNN Frameworks with GPU Initiated Direct Storage Accesses

Autor: Park, Jeongmin Brian, Mailthody, Vikram Sharma, Qureshi, Zaid, Hwu, Wen-mei

Graph Neural Networks (GNNs) are emerging as a powerful tool for learning from graph-structured data and performing sophisticated inference tasks in various application domains. Although GNNs have been shown to be effective on modest-sized graphs, tr

Externí odkaz: http://arxiv.org/abs/2306.16384

Preclinical models of atherosclerosis: An overview

Autor: Priyanka Arya, Vikram Sharma, Surabhi Thapliyal, Rahul Sagar, Priyanka Singh

Publikováno v: Iranian Journal of Basic Medical Sciences, Vol 27, Iss 5, Pp 535-542 (2024)

Atherosclerosis is a primary cause of illness and death globally and its mechanism is still unclear. Different animal models have been created to evaluate the progression of atherosclerosis, allowing researchers to carefully control the circumstances

Externí odkaz: https://doaj.org/article/3dcf60d3d49a46f8b47baf9193e4ca39

A Comprehensive Review of the Use of Plant-Derived Antioxidants in the Management of Non-Alcoholic Liver Toxicity

Autor: Shubham Verma, Akshay Thakur, Kaunava Roy, Vir Vikram Sharma

Publikováno v: Journal of Pharmacy, Vol 4, Iss 2 (2024)

ABSTRACT With its rising incidence, non-alcoholic fatty liver disease (NAFLD) has become a global health problem. Hepatic lipid buildup, inflammation, and oxidative stress are the results of complex interactions between metabolic, genetic, and env

Externí odkaz: https://doaj.org/article/762a9c083e9445c4a4b192dc1a852a22

GPU-Initiated On-Demand High-Throughput Storage Access in the BaM System Architecture

Autor: Qureshi, Zaid, Mailthody, Vikram Sharma, Gelado, Isaac, Min, Seung Won, Masood, Amna, Park, Jeongmin, Xiong, Jinjun, Newburn, CJ, Vainbrand, Dmitri, Chung, I-Hsin, Garland, Michael, Dally, William, Hwu, Wen-mei

Publikováno v: ASPLOS 2023: Proceedings of the 28th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2

Graphics Processing Units (GPUs) have traditionally relied on the host CPU to initiate access to the data storage. This approach is well-suited for GPU applications with known data access patterns that enable partitioning of their dataset to be proce

Externí odkaz: http://arxiv.org/abs/2203.04910