seung-lab / dijkstra3d

Dijkstra's Shortest Path for 6, 18, and 26-Connected 3D (Volumetric) Image Volumes
GNU General Public License v3.0
71 stars 13 forks source link

perf(dijkstra): prefetch instructions ~10-20% improvement #14

Closed william-silversmith closed 3 years ago

william-silversmith commented 3 years ago

For unidirectional dijkstra (compass=False) prefetch expected memory locations for a performance bump using _mm_prefetch.

https://software.intel.com/sites/landingpage/IntrinsicsGuide/#text=_mm_prefetch