Micaela's simulation of 10^5 massless disk stars integrated perturbed by a galactic bar. This ran in 60 seconds. The outer ring has morphology similar to R1'R2 pseudo-ringed galaxies. Alex's simulation of two planets migrating in a disk of 10^4 planetesimals. All force pairs are computed. The x-axis is semi-major axis; the y-axis the eccentricity. The integrator works in democratic heliocentric coordinates so that Pluto sized planetsimal masses can be integrated in single precision floating point. This ran in about 6 hours.

Introduction to Programming in CUDA

One day hands-on programming workshop.

Location:

Bausch and Lomb 407, University of Rochester

Date and Time:

Friday June 20, 2008, 9am-5pm

Organizers:

Richard Edgar, Alice Quillen, Alex Moore

Schedule

9:00-9:30 Lecture
Richard Edgar 		Welcome, Introduction
Why is it fun and exciting to program on GPUs?		 Intro.pdf , ComputerArchitecture.pdf
9:30-10:00 Lecture
Alice Quillen		 Cookbook
Running your first programs with CUDA on the GPU device 		GPUrecipe.ppt , GPUrecipe.pdf
10:00-11:30 Workshop Writing short programs for the GPU device Assignments A
11:30-12:00 Discussion Introductions x
12:00-1:00 Lunch B+L 208 provided
1:00-1:30 Lecture
Alex Moore 		How to build your own computer with a fast video card
Parallel computing on GPUs in context with other settings 		Hardstuff.pdf
1:30-2:45 Lecture
Richard Edgar 		Optimization on the GPU
Benchmarks 		AdvancedCUDA.pdf
2:45-5:00 Workshop Writing fast programs for the GPU device Assignments B

Links:

Supported in part by the departments of Physics and Astronomy, Optics, and Electrical and Computer Engineering at the University of Rochester.