Changes so far:
-
Faster search (about 1.6X; more for asymetric searches)
-
Added support for widths > 10
-
Reduced memory consumption (only 19GB needed for width 11 searches)
-
Table generation much faster (more than 5x for large tables)
-
Table generation dynamic (as needed) rather than up-front
-
Rule parsing integrated; no need for separate python script
-
Added randomized search order; are you feeling lucky?
-
Asymmetric searches now twice as fast because mirror images not searched
-
Various bugs fixed from original sources
I compile with g++ -std=c++11 -O3 -march=native -o ntzfind ntzfind.cpp; make sure to enable C++11 support in your compiler.