VSCSE - Virtual School of Computational Science and Engineering

Prerequisite Course: Introduction to CUDA

July 12–23, 2010

This short online course is available FREE to registered on-site participants to help these students meet prerequisites for Proven Algorithmic Techniques for Many-core Processors.

This course includes eight 45-minute recorded lectures; each week students will be expected to watch four video lectures on their own schedule and must complete two short programming assignments. Teaching assistants will conduct at least two online, synchronous open labs per week, answering questions and assisting students with their assignments. Online discussion forums will enable participants to post questions and get assistance from fellow students and teaching assistants.

Participants will receive access to high-performance computing systems for the duration of the prerequisite course and the corresponding summer course.

Prerequisite: C or C++

Instructor: Josh Anderson, University of Michigan


  • Introduction to Many-Core Processors
  • CUDA Programming Model
  • CUDA Threading Model
  • CUDA Memory Model
  • Basic Performance Considerations
  • Practical Performance Tuning
  • Floating-Point Precision
  • Installing and Using the CUDA Toolkit

Open labs:

  • Tuesdays @ 11 a.m. CT
  • Thursdays @ 1 p.m. CT

Additional open labs will be scheduled if necessary.