Hardware architecture and software stack for PIM based on commercial DRAM technology: Industrial product

S Lee, S Kang, J Lee, H Kim, E Lee… - 2021 ACM/IEEE 48th …, 2021 - ieeexplore.ieee.org
Emerging applications such as deep neural network demand high off-chip memory
bandwidth. However, under stringent physical constraints of chip packages and system …

A modern primer on processing in memory

O Mutlu, S Ghose, J Gómez-Luna… - … computing: from devices …, 2022 - Springer
Modern computing systems are overwhelmingly designed to move data to computation. This
design choice goes directly against at least three key trends in computing that cause …

Benchmarking a new paradigm: Experimental analysis and characterization of a real processing-in-memory system

J Gómez-Luna, I El Hajj, I Fernandez… - IEEE …, 2022 - ieeexplore.ieee.org
Many modern workloads, such as neural networks, databases, and graph processing, are
fundamentally memory-bound. For such workloads, the data movement between main …

Drisa: A dram-based reconfigurable in-situ accelerator

S Li, D Niu, KT Malladi, H Zheng, B Brennan… - Proceedings of the 50th …, 2017 - dl.acm.org
Data movement between the processing units and the memory in traditional von Neumann
architecture is creating the" memory wall" problem. To bridge the gap, two approaches, the …

Processing data where it makes sense: Enabling in-memory computation

O Mutlu, S Ghose, J Gómez-Luna… - Microprocessors and …, 2019 - Elsevier
Today's systems are overwhelmingly designed to move data to computation. This design
choice goes directly against at least three key trends in systems that cause performance …

Rowhammer: A retrospective

O Mutlu, JS Kim - … Transactions on Computer-Aided Design of …, 2019 - ieeexplore.ieee.org
This retrospective paper describes the RowHammer problem in dynamic random access
memory (DRAM), which was initially introduced by Kim et al. at the ISCA 2014 Conference …

Recnmp: Accelerating personalized recommendation with near-memory processing

L Ke, U Gupta, BY Cho, D Brooks… - 2020 ACM/IEEE 47th …, 2020 - ieeexplore.ieee.org
Personalized recommendation systems leverage deep learning models and account for the
majority of data center AI cycles. Their performance is dominated by memory-bound sparse …

Tensordimm: A practical near-memory processing architecture for embeddings and tensor operations in deep learning

Y Kwon, Y Lee, M Rhu - Proceedings of the 52nd Annual IEEE/ACM …, 2019 - dl.acm.org
Recent studies from several hyperscalars pinpoint to embedding layers as the most memory-
intensive deep learning (DL) algorithm being deployed in today's datacenters. This paper …

Processing-in-memory: A workload-driven perspective

S Ghose, A Boroumand, JS Kim… - IBM Journal of …, 2019 - ieeexplore.ieee.org
Many modern and emerging applications must process increasingly large volumes of data.
Unfortunately, prevalent computing paradigms are not designed to efficiently handle such …

DAMOV: A new methodology and benchmark suite for evaluating data movement bottlenecks

GF Oliveira, J Gómez-Luna, L Orosa, S Ghose… - IEEE …, 2021 - ieeexplore.ieee.org
Data movement between the CPU and main memory is a first-order obstacle against improv
ing performance, scalability, and energy efficiency in modern systems. Computer systems …