SIMD instructions[2] lowered the amount of cpu cycles needed per operation.
Novel algorithms[3] lowered the amount of operations needed per iteration.
[1] https://people.csail.mit.edu/rivest/lcs35-puzzle-description... (short description at the bottom).
[2] https://en.wikipedia.org/wiki/SIMD
[3] https://gmplib.org/manual/Algorithms.html
The problem is designed so that iterations have to be sequential, but a single iteration can use parallelism.
SIMD instructions[2] lowered the amount of cpu cycles needed per operation.
Novel algorithms[3] lowered the amount of operations needed per iteration.
[1] https://people.csail.mit.edu/rivest/lcs35-puzzle-description... (short description at the bottom).
[2] https://en.wikipedia.org/wiki/SIMD
[3] https://gmplib.org/manual/Algorithms.html