Course project
This course project focuses on using the K20 GPU at coit-grid08.uncc.edu for teams of 2 graduates or 2-3 undergraduates (individual work allowed). Each team must propose a unique, pre-approved project by the deadlines. Potential topics include parallel algorithms (sorting, searching, AI), languages/tools (OpenCL, jCUDA, CUDA), hardware evaluations, and graphics applications (CUDA, OpenGL). Key deadlines include topic submission, progress report, and final project report with presentation. All projects are first come, first served.
Course project
E N D
Presentation Transcript
Course project • To use the new K20 GPU coit-grid08.uncc.edu • Teams of 2 graduates or 2-3 undergraduates (Can work individually) Each team to propose a unique project. Every project must be different and pre-approved, first come first served ITCS 4/5010 GPU Programming, B. Wilkinson, CourseProject.ppt Feb 11, 2013
Possible topics • Parallel Algorithms or Applications – such as sorting, searching, AI, numeric methods, image processing, … • Languages and Tools – such as OpenCL, OpenACC, jCUDA, hiCUDA, PyCUDA, hybrid programming CUDA/OpenMP and/or MPI, CUDA/OpenCL, NVIDIA tools, comparative study of competing tools, … • Hardware – Evaluating new features of K20, multiple GPUs, … • Graphics – using GPUs for graphics with CUDA, OpenGL, …, • … Ideas!
Deadlines • Monday February 25th, 2013 Finalize a topic and submit a 1/2- 1 page description of proposed work • Tuesday February 26th, 2013 Approval for work • March 4th-9th, 2013 Spring break • Monday March 18th, 2013 Submit two-page progress report. • Tuesday March 19th, 2013 Report to class on progress • Tuesday April 16th, 2013 Submit project report (6 pages minimum) plus 6-10 presentation slides • April 16th/18th/23th, 2013 10-15-minute presentation to be scheduled during this time