Project Info

LARGE-SCALE GRAPH ANALYTICS USING MULTIPLE GPUS

1. Understanding the bottleneck of multi-GPU graph analytics.
2. Building a runtime to minimize inter-GPU communication.
3. Demonstrating >10X speedups over CPU-based graph processing systems.

More Information

https://gunrock.github.io/docs

Grand Engineering Challenge: Not applicable

Student Preparation

Qualifications

The student should be familiar with CUDA programming.

Time Commitment

80 hours/month

Skills/Techniques Gained

Performance optimization
GPU programming
Graph analytics

Mentoring Plan

The student can attend our weekly group meetings. I’ll also have bi-weekly individual meetings with the student.