PIM Course: Lecture 3: Real-world PIM: UPMEM PIM - Fall 2022

Projects & Seminars, ETH Zürich, Fall 2022
Data-Centric Architectures: Fundamentally Improving Performance and Energy

Lecture 3: Real-world Processing-in-Memory Architectures: UPMEM PIM Architecture
Lecturer: Dr. Juan Gómez Luna
Date: October 25, 2022

Recommended Readings:
====================
Intelligent Architectures for Intelligent Computing Systems

A Modern Primer on Processing in Memory

RowHammer: A Retrospective

RECOMMENDED LECTURE VIDEOS & PLAYLISTS:
========================================
Heterogeneous Systems Spring 2022 Lectures Playlist:

Processing-in-Memory Spring 2022 Lectures Playlist:

Computer Architecture Fall 2022 Lectures Playlist:

Digital Design and Computer Architecture Spring 2022 Livestream Lectures Playlist:

Featured Lectures:

The Story of RowHammer Lecture:

Accelerating Genome Analysis Lecture:

Memory-Centric Computing Systems Tutorial at IEDM 2021:

Intelligent Architectures for Intelligent Machines Lecture:

Public Lectures by Onur Mutlu, Playlist:

[YintaoHe]

Thanks for the great lecture. I have two questions: 1. Why we still use DRAM as main memory in the whole system, instead of using all the PIM-enabled DRAM as main memory? If so, we can directly execute the computing without data transfer between host CPU's main memory and DPU's MRAM banks.

[YintaoHe]

2. The UPMEM's task offloading process is very similar to GPU's. Will the differences from GPU be discussed in follow-up lectures?