Closed rageworx closed 5 years ago
Current simple OpenMP is not so powerfully optimized for AARCH64 like nVidia Jetson TX2. Performance gone for down when it adjusted runnig for single loop for OpenMP.
Maybe yasm required ?
Tested for many different aarch64 targets, but it depends on each kernel performance.
Current simple OpenMP is not so powerfully optimized for AARCH64 like nVidia Jetson TX2. Performance gone for down when it adjusted runnig for single loop for OpenMP.