A team from DIG won the KAUST-NVIDIA Hackathon 2022. For this 7th hackathon edition, twelve teams of students and researchers enjoyed the opportunity to accelerate their codes on GPUs under the guidance of expert mentors from NVIDIA, Codee and KAUST Supercomputing Lab computational scientists.

The team composed by Juan Daniel Romero, Nick Luiken and Miguel Angel Corrales Guerrero won the Best Performance achievement. They were able to leverage GPU capabilities to optimize the computation of thousands of bisection problems used within a primal-dual-based segmentation algorithm. In simpler terms, they manage to get an algorithm that takes 80 seconds to compute an element-by-element operation of a large matrix to only take 1 second! 

The winning code can be found at https://github.com/DIG-Kaust/HPC_Hackathon_DIG.