[행사/세미나] 메모리 시뮬레이터(Ramulator) 세미나 (ETH Haocong Luo) (4/29(수) 11:10)
- ice
- 조회수672
- 2026-04-23
일자: 2026년 4월 29일 (수) 11:10
장소: 화학관 330102호
연사: Haocong (Richard) Luo (ETH Zurich)
제목: A Sneak Peek of Ramulator 2.1 & Identifying the Gaps and Inconsistencies Between Device-Level Study and Experimental Characterization of DRAM Read Disturbance
요약: In this talk, we cover two crucial topics on DRAM research: 1) DRAM and memory controller simulation infrastructure, and 2) fundamentally understanding DRAM read disturbance.
First, we introduce Ramulator 2.1, a modern, modular, and extensible cycle-level DRAM simulator. It is the successor of Ramulator 1.0 and a major overhaul of Ramulator 2.0. The goal of Ramulator 2.1 is to enable rapid and agile implementation and evaluation of design changes in the memory controller and DRAM to meet the increasing research effort in improving the performance, security, and reliability of memory systems. In this talk, we will demonstrate Ramulator 2.1’s 1) key features and functionality, 2) key modeling and simulation methodology, and 3) how to extend and modify Ramulator 2.1.
Second, we present our research on fundamentally understanding DRAM read disturbance phenomena. We first examine prior device-level studies on DRAM read disturbance and extract key characteristics of DRAM read disturbance bitflips from these device-level works. Then, we perform real-chip characterization with the goal of reproducing the device-level characteristics of read disturbance bitflips.
Our results show that even the state-of-the-art device-level studies of DRAM read disturbance bitflips do not explain all the bitflip characteristics observed from experimental characterization studies.
연사:
Haocong Luo is a senior PhD student of the SAFARI Research Group at ETH Zurich, advised by Prof. Onur Mutlu. His research interests include 1) DRAM robustness (retention failure, read disturbance, etc.), 2) DRAM architecture and memory system performance, 3) Processing-in-Memory (PIM), and 4) understanding and improving the memory system performance for emerging workloads and applications.


