Implementation and benchmarking of processor architectures for application-specific instruction set processors for implantable medical devices

Master Thesis (2021)
Author(s)

J.R.H.R. Smit (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Contributor(s)

Z. Al-Ars – Mentor (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Faculty
Electrical Engineering, Mathematics and Computer Science
More Info
expand_more
Publication Year
2021
Language
English
Graduation Date
26-04-2021
Awarding Institution
Delft University of Technology
Programme
Computer Engineering
Faculty
Electrical Engineering, Mathematics and Computer Science
Downloads counter
194
Collections
thesis
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

An application-specific instruction set processor (ASIP) can provide for higher power and computationalefficiency compared to general-purpose processors. These attributes are essential for implantable medicaldevices which often run computationally intensive tasks on a strict power budget. This thesis compilesa collection of benchmarks by porting the existing benchmark suites ImpBench and CoreMark, and byimplementing a novel benchmark for artificial neural networks. Four architectures are selected in thecomparison; RISC, DSP, VLIW, and TTA. Implementations of these architectures are produced by theASIP Designer and OpenASIP toolsets. The benchmarks are simulated on these implementations andthe power consumption is measured on an FPGA. The thesis concludes that the implementations of theDSP and VLIW architectures do not deliver enough performance for their heavier use of resources, andrecommends a follow-up research by extending the TTA PeLoTTA and RISC-V Tzscale processors withapplication-specific instructions and running simulations for ASIC power and area numbers.

Files

Thesis.pdf
(pdf | 1.81 Mb)
License info not available