IEEE Circuits and Systems Magazine - Q2 2021 - 3

From the
Guest Editors
Zhiru Zhang
Cornell University
Hongbo Rong
Intel Labs
Yu Wang
Tsinghua University
Introduction of Special Issue on FPGA-Based Computing
O
ver the past three decades, field-programmable
gate arrays (FPGAs) have evolved from small
chips with a few thousand logic blocks to billion-transistor
system-on-chips. Modern FPGAs provide
massive fine-grained parallelism, high energy efficiency,
and more predictable performance than CPUs
and GPUs. More importantly, they can be reconfigured
to implement a special-purpose architecture that is specifically
customized based on the key characteristics of
a target application (or domain). For these reasons, recent
years have seen a rapidly increasing use of FPGAs
for computing, where a plethora of new architectures,
compilers, and FPGA-accelerated applications are being
demonstrated and actively developed. This special issue
includes five survey papers from renowned experts in
the field of FPGAs. The goal is to summarize some of the
recent advances on FPGA-based computing, to elucidate
the architecture- and application-level trends, and to promote
new tools to enable productive FPGA programming.
The first survey walks us through the evolution of
the FPGA architectures accompanying ever-changing
application requirements and recent trends in technology
scaling. Design and analysis of an FPGA architecture
is broken down into design decisions at various
levels, from transistors to embedded blocks. Major
challenges with the existing solutions are analyzed,
Digital Object Identifier 10.1109/MCAS.2021.3071606
Date of current version: 24 May 2021
raising questions to address for future progress of
FPGA architectures.
The second paper presents new views on FPGA accelerators
for near-future big-data analytics, their challenges,
solutions, and opportunities. The paper also
describes a prototype system for seamless integration
of FPGAs in cloud infrastructure for big data analytics.
The third paper presents a detailed survey of the
state-of-the-art FPGA work in different robotic applications,
including sensing, perception, localization, and
planning. It also introduces two example systems: one
for commercial autonomous vehicles and the other for
space robots.
The fourth survey covers the recent trends in domainspecific
computing for deep learning, from modern software
optimization, prevailing hardware backends, and
intermediate representations. It offers commentary on
the general tradeoffs and challenges during the design
process and application deployment, which leads to further
discussion on the applicability of FPGA platforms in
this domain and an in-depth introduction to the Xilinx
Versal architecture.
The last paper focuses on the compilation for FPGA
accelerators. It details a drastically different form of
high-level synthesis, which generates dynamically
scheduled dataflow circuits out of software programs.
Such dataflow circuits achieve out-of-order memory accesses
and speculation, similar to some of the features
in high-performance microprocessors.
SECOND QUARTER 2021
IEEE CIRCUITS AND SYSTEMS MAGAZINE
3
IEEE Circuits and Systems Magazine - Q2 2021

Table of Contents for the Digital Edition of IEEE Circuits and Systems Magazine - Q2 2021
