Predavanje: "Overcoming the von Neumann Bottleneck with In-Memory Computing"

Odjel za računarstvo i Odjel za obrazovanje Hrvatske sekcije IEEE pozivaju vas na predavanje pod naslovom:

Overcoming the von Neumann Bottleneck with In-Memory Computing

koje će održati dr. sc. William Andrew Simon u utorak, 7. veljače 2023. godine s početkom u 17:30 sati, u dvorani D160-2 Fakulteta elektrotehnike i računarstva Sveučilišta u Zagrebu, Unska 3.
 
Predavanje je otvoreno za sve zainteresirane, a posebno se pozivaju studenti. Jezik predavanja je engleski.

Predavanje će se moći pratiti uživo i preko platforme YouTube na sljedećoj poveznici: FER ZESOI D160-2

Životopis predavača i sažetak predavanja nalaze se u nastavku obavijesti.

In-memory computing as a field of research proposes the abolition of the division between compute and memory components of a compute device, commonly referred to as the von Neumann bottleneck. A variety of innovative in-memory implementations have been proposed recently, and come in basic flavors, digital- and analog-based. This talk will discuss one example from each domain, that of the digital-based BLADE, and that of IBM’s analog-based HERMES core for performing in-memory matrix multiplications.

O predavaču:

William Andrew Simon received his Ph.D. in Electrical Engineering from Embedded Systems Laboratory (ESL) at the Swiss Federal Institute of Technology Lausanne (EPFL), Switzerland, in 2022. He is currently a post-doctoral researcher at IBM Zurich in the Cloud Computing and AI Design Department. His current work focuses on in-memory computing centric architectures for accelerating deep neural networks.