We use cookies

We use cookies to ensure you get the best experience on our website. For more information on how we use cookies, please see our cookie policy.

Back to home

Utilizing Chiplet-Locality For Efficient Memory Mapping In MCM GPUs (ETRI, Sungkyunkwan Univ.)

Source

SemiEngineering

Published

TL;DR

AI Generated

Researchers from ETRI and Sungkyunkwan University published a technical paper on leveraging chiplet-locality for efficient memory mapping in multi-chip module GPUs. The study explores how page size in memory mapping impacts memory system non-uniformity in MCM GPUs. They introduce CLAP, a method to determine the suitable page size for each application based on chiplet-locality patterns. CLAP optimizes data placement within chiplets, improving performance by up to 19.2% compared to previous paging schemes. The paper was presented at the 58th IEEE/ACM International Symposium on Microarchitecture.