Presenter: Dr. Harisankar Sadasivan, Member of Technical Staff SDE, AMD | Faculty (part-time), UW Seattle | Sr Member, IEEE
Abstract:
In the relentless march of AI, we face an unprecedented challenge: compute demand is skyrocketing, leaving Moore’s Law and even cutting-edge heterogeneous hardware designs in the dust. This isn’t just a gap—it’s a chasm, yawning wider by orders of magnitude with each AI breakthrough. Enter the unsung hero of this technological arms race: software optimization. In this lightning talk, we’ll dive into the heart of this computational battleground, exploring the meteoric rise of AI’s computational hunger and its implications for the tech landscape. We’ll examine why traditional hardware solutions are falling short, how they’re evolving to keep pace, and the critical role of aggressive software optimizations in bridging the ever-widening performance gap. By dissecting emerging patterns in AI workload characteristics that are reshaping our approach to system design, we’ll unveil the symbiotic evolution of hardware and software stacks in response to these unprecedented demands. Join us for a whirlwind tour of the cutting edge, where software ingenuity is rewriting the rules of the game in our quest to satiate AI’s insatiable appetite for compute power.
Presenter Bio:
Dr. Harisankar Sadasivan, Member of Technical Staff SDE, AMD
Dr. Hari Sadasivan is a Member of Technical Staff SDE at AMD, a part-time faculty member at the University of Washington, Seattle, and an IEEE Senior Member. Hari focuses on systems for AI and life-sciences. Hari drives the AMD Center of Excellence in AI at the Allen School of CSE and is the co-founder of the AMD-omics focus group which collaborates with research institutes world over. Hari holds a PhD in CSE from the University of Michigan Ann Arbor and is a peer-reviewer for several IEEE/ACM venues.