[PDF][PDF] At the Locus of Performance: Quantifying the Effects of Copious 3D-Stacked Cache on HPC Workloads
Over the last three decades, innovations in the memory subsystem were primarily targeted at
overcoming the data movement bottleneck. In this paper, we focus on a specific market trend …
overcoming the data movement bottleneck. In this paper, we focus on a specific market trend …
TransPimLib: A Library for Efficient Transcendental Functions on Processing-in-Memory Systems
Processing-in-memory (PIM) promises to alleviate the data movement bottleneck in modern
computing systems. However, current real-world PIM systems have the inherent …
computing systems. However, current real-world PIM systems have the inherent …
Heterogeneous Data-Centric Architectures for Modern Data-Intensive Applications: Case Studies in Machine Learning and Databases
Today's computing systems require moving data back-and-forth between computing
resources (eg, CPUs, GPUs, accelerators) and off-chip main memory so that computation …
resources (eg, CPUs, GPUs, accelerators) and off-chip main memory so that computation …
Ndp-rank: Prediction and ranking of ndp systems performance using machine learning
The near-data processing (NDP) paradigm has recently gained popularity as a promising
method for mitigating the memory wall challenges of future computing systems. Modern 3D …
method for mitigating the memory wall challenges of future computing systems. Modern 3D …
Analysis of Conventional, Near-Memory, and In-Memory DNN Accelerators
Various DNN accelerators based on Conventional compute Hardware Accelerator (CHA),
Near-Data-Processing (NDP) and Processing-in-Memory (PIM) paradigms have been …
Near-Data-Processing (NDP) and Processing-in-Memory (PIM) paradigms have been …
Accelerating Irregular Applications via Efficient Synchronization and Data Access Techniques
C Giannoula - arXiv preprint arXiv:2211.05908, 2022 - arxiv.org
Irregular applications comprise an increasingly important workload domain for many fields,
including bioinformatics, chemistry, physics, social sciences and machine learning …
including bioinformatics, chemistry, physics, social sciences and machine learning …
Methodologies, Workloads, and Tools for Processing-in-Memory: Enabling the Adoption of Data-Centric Architectures
The increasing prevalence and growing size of data in modern applications have led to high
costs for computation in tra-ditional processor-centric computing systems. Moving large …
costs for computation in tra-ditional processor-centric computing systems. Moving large …
PUMA: Efficient and Low-Cost Memory Allocation and Alignment Support for Processing-Using-Memory Architectures
GF Oliveira, EG Esposito, J Gómez-Luna… - arXiv preprint arXiv …, 2024 - arxiv.org
Processing-using-DRAM (PUD) architectures impose a restrictive data layout and alignment
for their operands, where source and destination operands (i) must reside in the same …
for their operands, where source and destination operands (i) must reside in the same …
Improving DRAM Performance, Reliability, and Security by Rigorously Understanding Intrinsic DRAM Operation
H Hassan - arXiv preprint arXiv:2303.07445, 2023 - arxiv.org
DRAM is the primary technology used for main memory in modern systems. Unfortunately,
as DRAM scales down to smaller technology nodes, it faces key challenges in both data …
as DRAM scales down to smaller technology nodes, it faces key challenges in both data …
TCAM-SSD: A Framework for Search-Based Computing in Solid-State Drives
As the amount of data produced in society continues to grow at an exponential rate, modern
applications are incurring significant performance and energy penalties due to high data …
applications are incurring significant performance and energy penalties due to high data …