<div dir="auto">Hi all, this one is happening at 1pm. See you soon. <br></div><div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Mon, Nov 11, 2024 at 16:58 Columbia EE Events <<a href="mailto:ee-events@ee.columbia.edu">ee-events@ee.columbia.edu</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><blockquote type="cite"><div><div dir="ltr"><div><div dir="ltr" class="gmail_signature"><div dir="ltr"><div><div><div></div></div></div></div></div></div></div></div></blockquote><font face="arial, sans-serif">Hi all,<br><br>This Thursday (November 14, 2024), Prashant Nair (<a href="https://urldefense.proofpoint.com/v2/url?u=https-3A__prashantnair.bitbucket.io&d=DwMFaQ&c=009klHSCxuh5AI1vNQzSO0KGjl4nbi2Q0M1QLJX9BeE&r=SD-9ztpHDQVkXLFc0zdc9RZWqGhdQ2vAJZ0O73eMlt0&m=XyQOwwB1AOGpyMIB1-oOgHWQMJ-nTno2aHWOWSU144rgeM7ezax7jnj3IvZEevKn&s=c-D5R1vShqpvp2suhlRY-_B8RhSeKbhn0LOcTF-yK8M&e=" target="_blank">https://prashantnair.bitbucket.io</a>) from the University of British Columbia (UBC) will be visiting and giving a talk at 1pm at EE conference room. <br><br>Professor Tanvir Ahmed Khan is hosting this seminar (reach out to him if you have any questions.)<br><br>-------<br>Title: Ender’s Game: Architectural Techniques to Develop Scalable and Secure Memory Systems <br><br>Abstract: The emergence of new applications, especially in machine learning, has increased the demand for scalable memory systems. DRAM cells, the foundation of modern memory systems, face challenges to meet the capacity needs of next-generation applications. This scalability issue has led designers to rethink memory architectures by examining data semantics and communication costs. Additionally, this shift has prompted memory security to become a first-order design criteria.<br><br>In this talk, I will discuss two recent initiatives from my research group to address these challenges. The first part of the presentation will introduce Hotline (ISCA 2024), an acceleration framework that reduces memory and communication overheads in the training of Deep Learning Recommendation Models by leveraging data access patterns. While Hotline effectively mitigates performance overheads, it does not tackle security vulnerabilities -- a growing concern as the memory footprint of applications increase. The second part of my talk will focus on memory security challenges, such as RowHammer, in high-capacity systems. I will present our work on Scalable and Secure Row Swap (Best Paper at HPCA 2023) that introduces the Juggernaut attack and proposes cost-effective defense mechanisms. The talk will conclude with a discussion of the broader research efforts in my group aimed at improving memory scalability for next-generation applications.<br><br>Biography: Prashant Nair is an Assistant Professor at the University of British Columbia (UBC), where he leads the Systems and Architectures (STAR) Lab. Additionally, he serves as an Affiliate Fellow at the Quantum Algorithms Institute. His research primarily focuses on memory systems. In recognition of his contributions, he received the 2024 TCCA Young Architect Award and was inducted into the MICRO and HPCA Halls of Fame. He has also received the 2023 HPCA Best Paper Award, two IEEE MICRO Top Picks honorable mentions, and the ECE Graduate Research Assistant Excellence Award from Georgia Tech.</font></div>
</blockquote></div></div>