Open william-silversmith opened 3 years ago
The compiler can't easily make inferences about how code will be accessed within the core neighborhood loop so if we instead write it explicitly, will we see a performance improvement from e.g. autovectorization?
The compiler can't easily make inferences about how code will be accessed within the core neighborhood loop so if we instead write it explicitly, will we see a performance improvement from e.g. autovectorization?