English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Today's computing systems require moving data back-and-forth betweencomputing resources (e.g., CPUs, GPUs, accelerators) and off-chip main memoryso that computation can take place on the data. Unfortunately, this datamovement is a major bottleneck for system performance and energy consumption.One promising execution paradigm that alleviates the data movement bottleneckin modern and emerging applications is processing-in-memory (PIM), where thecost of data movement to/from main memory is reduced by placing computationcapabilities close to memory.  Naively employing PIM to accelerate data-intensive workloads can lead tosub-optimal performance due to the many design constraints PIM substratesimpose. Therefore, many recent works co-design specialized PIM accelerators andalgorithms to improve performance and reduce the energy consumption of (i)applications from various application domains; and (ii) various computingenvironments, including cloud systems, mobile systems, and edge devices.  We showcase the benefits of co-designing algorithms and hardware in a waythat efficiently takes advantage of the PIM paradigm for two moderndata-intensive applications: (1) machine learning inference models for edgedevices and (2) hybrid transactional/analytical processing databases for cloudsystems. We follow a two-step approach in our system design. In the first step,we extensively analyze the computation and memory access patterns of eachapplication to gain insights into its hardware/software requirements and majorsources of performance and energy bottlenecks in processor-centric systems. Inthe second step, we leverage the insights from the first step to co-designalgorithms and hardware accelerators to enable high-performance andenergy-efficient data-centric architectures for each application.

Heterogeneous Data-Centric Architectures for Modern Data-Intensive Applications: Case Studies in Machine Learning and Databases