Key is of 'int' type in all benchmarks. Values of keys are 0...N in all tests (I know it is too primitively).

bench_size) This test is very artificial
  1. Hash function is Identity function, i.e. hash_func(k) = k
  2. Key compare function is as fast as it is possible, i.e. bool operator < (int, int)
  3. X-axis is a number of elements in array

http://judyhash.sourceforge.net/benchmarks/bench_size.html

bench_size65599) Much more real test with real hash function and compare function close to the real one
  1. Hash function is "polynom 65599" (Look at the source code ;-) )
  2. Key compare function is slower than in 'bench_size' but is not very slow (slowness_level=SLOW_LEVEL_DEF)
  3. X-axis is a number of elements in array

http://judyhash.sourceforge.net/benchmarks/bench_size65599.html

bench_slowness) This benchmark is about how performance is changed in case of sloooow key comparing function.
  1. Hash function is "polynom 65599" as in 'bench_size65599'
  2. ITEMS_DEF elements in array
  3. X-axis is a slowness of the key comparing function

http://judyhash.sourceforge.net/benchmarks/bench_slowness.html