5 citations · 8 across the 5 of their papers we have counts for
4 papers · 1 filter
POLCA: Power Oversubscription in LLM Cloud Providers
Pratyush Patel, Esha Choukse, Chaojie Zhang +4
Recent innovation in large language models (LLMs), and their myriad use-cases have rapidly driven up the compute capacity demand for datacenter GPUs. Several cloud providers and ot…
Hybrid Computing for Interactive Datacenter Applications
Pratyush Patel, Katie Lim, Kushal Jhunjhunwalla +5
Field-Programmable Gate Arrays (FPGAs) are more energy efficient and cost effective than CPUs for a wide variety of datacenter applications. Yet, for latency-sensitive and bursty w…
Srifty: Swift and Thrifty Distributed Training on the Cloud
Liang Luo, Peter West, Arvind Krishnamurthy +1
Finding the best VM configuration is key to achieve lower cost and higher throughput, two primary concerns in cloud-based distributed neural network (NN) training today. Optimal VM…
A Server-based Approach for Predictable GPU Access with Improved Analysis
Hyoseung Kim, Pratyush Patel, Shige Wang +2
We propose a server-based approach to manage a general-purpose graphics processing unit (GPU) in a predictable and efficient manner. Our proposed approach introduces a GPU server t…