Join us for our upcoming SAFARI Live Seminar:
Speaker: Abdullah Giray Yağlıkçı, SAFARI Research Group, ETH Zürich
Date: Friday, February 2, 17:00 Zurich time (CET).
Where: Livestream on YouTube (Link) & Location (to be confirmed)
Title: Efficiently and Scalably Mitigating RowHammer in Modern and Future DRAM-Based Memory Systems
This is part two of a two-part talk in which Giray will present his PhD work.
Details on part one can be found here.
Abstract:
RowHammer is a circuit-level DRAM vulnerability where repeatedly accessing (i.e., hammering) a DRAM row can cause bit flips in physically nearby rows. As the RowHammer vulnerability worsens with shrinking DRAM cell size and cell-to-cell spacing, ensuring RowHammer-safe operation becomes more expensive in terms of performance overhead, energy consumption, and hardware complexity. To address these challenges, we present three new mechanisms: BlockHammer, HiRA, and Svärd.
BlockHammer selectively throttles memory accesses that may otherwise cause RowHammer bitflips. By doing so, BlockHammer prevents RowHammer bitflips efficiently and scalably without knowledge of or modifications to DRAM internals. BlockHammer provides 1) competitive performance and energy when the system is not under a RowHammer attack and 2) significantly better performance and energy when the system is under a RowHammer attack.
HiRA hides a DRAM refresh operation’s latency by refreshing a DRAM row concurrently with accessing or refreshing another row within the same DRAM bank. Unlike prior works, HiRA achieves this parallelism without any modifications to off-the-shelf DRAM chips. To do so, it leverages the new observation that two rows in the same bank can be activated without data loss if the rows are connected to different charge restoration circuitry. HiRA significantly reduces the time spent on refresh operations and reduces the performance degradation due to periodic refreshes and refreshes for RowHammer protection (preventive refreshes).
Svärd leverages the heterogeneity in read disturbance vulnerability across DRAM rows and dynamically tunes the aggressiveness of RowHammer mitigation mechanisms based on the vulnerability level of potential victim rows. By doing so, Svärd avoids unnecessarily aggressive protection on already strong rows and reduces the performance overhead of existing RowHammer mitigation mechanisms.
Speaker Bio:
Abdullah Giray Yaglikci is a Ph.D. candidate in the Safari Research Group at ETH Zürich, working with Prof. Onur Mutlu. His current broader research interests are computer architecture, systems, and hardware security, focusing on DRAM robustness and performance. In particular, his Ph.D. research focuses on understanding and solving the RowHammer vulnerability, on which he has several publications. Giray received his B.S. degree in Electrical Engineering and M.Sc. degree in Computer Engineering from TOBB University of Economics and Technology, Ankara, Turkiye and he holds a second M.Sc. in Computer Science from University of Notre Dame, IN, USA. Giray holds an honorable mention from the Intel Hardware Security Academic Award, 2021, and is the First-Place winner in the graduate category of Student Research Competition in PACT, 2023. Giray’s research is partly supported by Google and the Microsoft Swiss Joint Research Center.
Relevant Papers
- A. Giray Yağlıkçı, Minesh Patel, Jeremie S. Kim, Roknoddin Azizi, Ataberk Olgun, Lois Orosa, Hasan Hassan, Jisung Park, Konstantinos Kanellopoulos, Taha Shahroodi, Saugata Ghose, and Onur Mutlu, “BlockHammer: Preventing RowHammer at Low Cost by Blacklisting Rapidly-Accessed DRAM Rows,” in HPCA, 2021.
- A. Giray Yağlıkçı, Ataberk Olgun, Minesh Patel, Haocong Luo, Hasan Hassan, Lois Orosa, Oguz Ergin, and Onur Mutlu, “HiRA: Hidden Row Activation for Reducing Refresh Latency of Off-the-Shelf DRAM Chips,” in MICRO, 2022.
- Abdullah Giray Yağlıkçı, Geraldo Francisco de Oliveira, Yahya Can Tugrul, Ismail Yuksel, Ataberk Olgun, Haocong Luo, and Onur Mutlu, “Spatial Variation-Aware Read Disturbance Defenses: Experimental Analysis of Real DRAM Chips and Implications on Future Solutions,” to appear in HPCA, 2024. (A short poster preprint is available here)