Scale-CNN : a tool for generating scalable high-throughput CNN inference accelerators on FPGAs

dc.contributor.advisorJohn, Lizy Kurian
dc.creatorRauch, Daniel Levi
dc.creator.orcid0000-0003-2888-4883
dc.date.accessioned2021-09-02T23:11:47Z
dc.date.available2021-09-02T23:11:47Z
dc.date.created2021-05
dc.date.issued2021-04-30
dc.date.submittedMay 2021
dc.date.updated2021-09-02T23:11:47Z
dc.description.abstractIn the past decade, research has shown that CNN inference can be considerably sped up via dedicated hardware accelerators. However, most existing accelerators have limited performance by only working on a single inference at a time and/or relying on slow off-chip memory accesses for hidden layers. These limitations stem from the high memory requirements of CNN inference, which can be 10s of Mb even for small networks with reduction techniques. Despite this, as Moore's law has continued to scale, this level of on-chip memory is now attainable. We propose Scale-CNN, a tool for generating multiple Pareto optimal design points for high-throughput CNN inference accelerators on Xilinx Ultrascale+ FPGAs. The Scale-CNN architecture dedicates separate hardware resources for each layer and stores all feature maps and weights on-chip, enabling a high-throughput network pipeline where each layer works on a different inference simultaneously with no off-chip memory accesses. Using Scale-CNN, we generate several accelerator IPs for Tiny Darknet on the smallest Virtex Ultrascale+ FPGA (XCVU3P) that range from 1.7 to 56.7 inferences per second utilizing 22% to 66% of FPGA resources.
dc.description.departmentElectrical and Computer Engineering
dc.format.mimetypeapplication/pdf
dc.identifier.urihttps://hdl.handle.net/2152/87413
dc.identifier.urihttp://dx.doi.org/10.26153/tsw/14362
dc.language.isoen
dc.subjectCNN
dc.subjectFPGA
dc.subjectAccelerator
dc.subjectHLS
dc.subjectDesign-space exploration
dc.titleScale-CNN : a tool for generating scalable high-throughput CNN inference accelerators on FPGAs
dc.typeThesis
dc.type.materialtext
thesis.degree.departmentElectrical and Computer Engineering
thesis.degree.disciplineElectrical and Computer Engineering
thesis.degree.grantorThe University of Texas at Austin
thesis.degree.levelMasters
thesis.degree.nameMaster of Science in Engineering

Access full-text files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
RAUCH-THESIS-2021.pdf
Size:
587.6 KB
Format:
Adobe Portable Document Format

License bundle

Now showing 1 - 2 of 2
No Thumbnail Available
Name:
PROQUEST_LICENSE.txt
Size:
4.45 KB
Format:
Plain Text
Description:
No Thumbnail Available
Name:
LICENSE.txt
Size:
1.84 KB
Format:
Plain Text
Description: