Thanks, Lucas. Will test it on my other PC...
At this time my code (non-optimized, compiled with old compiler Delphi 7 ) for example compresses in 1 thread enwik8 up to 43.5% for 1.5 sec. on CPU with L1 cashe = 64Kb & L2 cashe = 1Mb, CPU 3.1GHz, RAM 667 MHz (sorry, old PC: Athlon II X2 255). It seems, my code beats many famous compressors... I'm going to port it in FASM to achieve and see highest speed.
Now I want to clarify whether I measure the speed. Maybe I'm doing something wrong...
1. My code does not calculate check sum.
2. I measure only the running time of the subroutine, that compresses the data. Like this:
Is it correct?
Anyway I will 10 times run my code and other compressors on RAM disk on 1 core enabled and compare the speed of work by eye.