Space Filling Curves
Pros
- There is a marginal difference even though index of next position of the curve has to be calculated
 
Cons
- Array has to be padded to a power of 2
 - Parallelism is slightly trickier as the offset of each thread has to calculated
 - It initially looks as if only static scheduling can be done
 
Related work
- Böhm, Christian, Martin Perdacher and Claudia Plant. “A Novel Hilbert Curve for Cache-Locality Preserving Loops.” IEEE Transactions on Big Data 7 (2021): 241-254.
 
Example
TODO
Try optimising hilber_inc_xy function in above gist