4 ms·
AFAIK this kind of problem is easily parallelized, and it should be possible to accelerate it using GPGPU... So, they may not be able to beat a well-tuned imple
by tachyonbeam 11y ago
AFAIK this kind of problem is easily parallelized, and it should be possible to accelerate it using GPGPU... So, they may not be able to beat a well-tuned implementation on a desktop workstation with a few GPUs costing less than 10K.