In this project I wanted to prove the efficiency of GPUs over CPUs for scientific calculations.
Problem: calculate the two-point angular correlation function for two sets of galaxies.
- Implementations:
- C implementation, sequential (no parallel programming)
- CUDA implementation, parallel programming
I've obtained a code speed-up of 70x faster (on average) compared to computing the same calculation on the CPU.
More details on the info file
- There are some auxiliar files attached to this project:
- hello world, CUDA programming
- simple sum, CUDA programming