Age | Commit message (Collapse) | Author |
|
This reverts commit f962e9656116cfe611500809344fb5bc1959d935.
It didn't actually work in many common cases, and making it work is
outside the scope of what we want to do when finding capsets. Thus, we
will try to solve this with a post-processing script instead, which
really fits with our needs better anyway (to prevent having to find
solutions twice).
Example of case that failed to be detected:
[000 001 010 011 100 101 112 122 212]
[000 001 010 011 100 101 112 212]
|
|
|
|
|
|
|
|
|
|
It was intended to be a Flush(), but even Flush() is unnecessary because
we aren't doing buffered I/O. This speeds up fully solving dimension 3
from 20 seconds to .1 seconds.
|
|
|
|
|
|
Using bitwise operations yields over a 2x speedup.
The iteration order of hasSet was reversed which causes it to be within
60% of the performance of hasSetWithLast, instead of 6x slower. It was
also rewritten to use hasSetWithLast for better code reuse.
|
|
|
|
|
|
|
|
|
|
|
|
It is around 60 times faster
|