FPGAs have found their way into data centers as accelerator cards, making
reconfigurable computing more accessible for high-performance applications. At
the same time, new high-level synthesis compilers like Xilinx Vitis and runtime
libraries such as XRT attract software programmers into the reconfigurable
domain. While software programmers are familiar with task-level and
data-parallel programming, FPGAs often require different types of parallelism.
For example, data-driven parallelism is mandatory to obtain satisfactory
hardware designs for pipelined dataflow architectures. However, software
programmers are often not acquainted with dataflow architectures - resulting in
poor hardware designs.
In this work we present FLOWER, a comprehensive compiler infrastructure that
provides automatic canonical transformations for high-level synthesis from a
domain-specific library. This allows programmers to focus on algorithm
implementations rather than low-level optimizations for dataflow architectures.
We show that FLOWER allows to synthesize efficient implementations for
high-performance streaming applications targeting System-on-Chip and FPGA
accelerator cards, in the context of image processing and computer vision.
Description
FLOWER: A comprehensive dataflow compiler for high-level synthesis
%0 Generic
%1 amiri2021flower
%A Amiri, Puya
%A Pérard-Gayot, Arsène
%A Membarth, Richard
%A Slusallek, Philipp
%A Leißa, Roland
%A Hack, Sebastian
%D 2021
%K Compiler Dataflow FPGA HLS
%R 10.1109/ICFPT52863.2021.9609930
%T FLOWER: A comprehensive dataflow compiler for high-level synthesis
%U http://arxiv.org/abs/2112.07789
%X FPGAs have found their way into data centers as accelerator cards, making
reconfigurable computing more accessible for high-performance applications. At
the same time, new high-level synthesis compilers like Xilinx Vitis and runtime
libraries such as XRT attract software programmers into the reconfigurable
domain. While software programmers are familiar with task-level and
data-parallel programming, FPGAs often require different types of parallelism.
For example, data-driven parallelism is mandatory to obtain satisfactory
hardware designs for pipelined dataflow architectures. However, software
programmers are often not acquainted with dataflow architectures - resulting in
poor hardware designs.
In this work we present FLOWER, a comprehensive compiler infrastructure that
provides automatic canonical transformations for high-level synthesis from a
domain-specific library. This allows programmers to focus on algorithm
implementations rather than low-level optimizations for dataflow architectures.
We show that FLOWER allows to synthesize efficient implementations for
high-performance streaming applications targeting System-on-Chip and FPGA
accelerator cards, in the context of image processing and computer vision.
@conference{amiri2021flower,
abstract = {FPGAs have found their way into data centers as accelerator cards, making
reconfigurable computing more accessible for high-performance applications. At
the same time, new high-level synthesis compilers like Xilinx Vitis and runtime
libraries such as XRT attract software programmers into the reconfigurable
domain. While software programmers are familiar with task-level and
data-parallel programming, FPGAs often require different types of parallelism.
For example, data-driven parallelism is mandatory to obtain satisfactory
hardware designs for pipelined dataflow architectures. However, software
programmers are often not acquainted with dataflow architectures - resulting in
poor hardware designs.
In this work we present FLOWER, a comprehensive compiler infrastructure that
provides automatic canonical transformations for high-level synthesis from a
domain-specific library. This allows programmers to focus on algorithm
implementations rather than low-level optimizations for dataflow architectures.
We show that FLOWER allows to synthesize efficient implementations for
high-performance streaming applications targeting System-on-Chip and FPGA
accelerator cards, in the context of image processing and computer vision.},
added-at = {2021-12-16T16:46:08.000+0100},
author = {Amiri, Puya and Pérard-Gayot, Arsène and Membarth, Richard and Slusallek, Philipp and Leißa, Roland and Hack, Sebastian},
biburl = {https://www.bibsonomy.org/bibtex/2a47172a1974839b29735df9f2aa24205/pooya_ww},
description = {FLOWER: A comprehensive dataflow compiler for high-level synthesis},
doi = {10.1109/ICFPT52863.2021.9609930},
interhash = {4fcad487f6981c3022ab5e0c2db8d9c1},
intrahash = {a47172a1974839b29735df9f2aa24205},
keywords = {Compiler Dataflow FPGA HLS},
note = {cite arxiv:2112.07789},
timestamp = {2021-12-16T16:46:08.000+0100},
title = {FLOWER: A comprehensive dataflow compiler for high-level synthesis},
url = {http://arxiv.org/abs/2112.07789},
year = 2021
}