University of Pennsylvania, CIS 565: GPU Programming and Architecture, Project 1 - Flocking
- Weiyu Du
- Tested on: CETS virtual lab MOR100B-09, Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz
- Plot of number of boids versus frame rate without visualization (x axis: 5000, 10000, 20000, 50000)
- For each implementation, how does changing the number of boids affect performance? Why do you think this is?
Performance decreases as the number of boids increases. This is because more boids require more threads for kernels that require significant operations, for example kernUpdateVelNeighborSearch, kernUpdatePos, kernComputeIndices.
- For each implementation, how does changing the block count and block size affect performance? Why do you think this is?
Performance does not change as block size increase. This is because larger block size would only require more threads for kernels that have very simple operation, for example kernResetIntBuffer.
- For the coherent uniform grid: did you experience any performance improvements with the more coherent uniform grid? Was this the outcome you expected? Why or why not?
Yes, I experienced performance improvements with coherent uniform grid from scattered grid, but not as significant as from naive to scattered grid. This is the outcome I expect because we cut out the time to access dev_particleArrayIndices memory for each thread, which improves performance.
- Did changing cell width and checking 27 vs 8 neighboring cells affect performance? Why or why not?
It does not affect performance much. We still have the same number of threads.