Project Info

Spin and thermal transport in complex nanostructures

Recurrent neural networks is widely used in domains, such as document understanding, machine translation, and sequence prediction. Although GPUs have shown their tremendous power in model training, inference of recurrent neural networks cannot efficiently utilize GPUs due to the limited parallelism. This project aims at designing a scheduling framework to accelerate inference on GPUs by more than 10 times.

More Information

1. https://iopscience.iop.org/article/10.1088/2043-6262/5/3/033001
2. https://iopscience.iop.org/article/10.1088/1367-2630/16/9/093029/pdf
3. https://nanohub.org/resources/18350/download/NikonovBeyondCMOS_3_NEGF.pdf

Grand Engineering Challenge: Engineer the tools of scientific discovery

Student Preparation

Qualifications

1. Proficient in some programming language
2. Had an introduction to differential equations and matrices

Time Commitment

40 hours/month

Skills/Techniques Gained

1. Understanding of quantum effects in transport of charge, spin, and heat
2. Introduction to basic quantum mechanics
3. Introduction to Green’s functions
3. Translating between physical problems, mathematical representations, and code to solve them
4. Presentation skills

Mentoring Plan

1. The student will work in a sub-group working on thermal and spin transport in nanostructures. There will be a graduate student and a senior design student on the project to help the student on a day to day basis.
2. I will meet with the student once a week in a sub-group meeting to help figure out ways around roadblocks, set goals for the coming week, and assess progress.
3. There will be a group meeting once every two weeks. The student will present at this meeting once every 2-3 months.