16 October 2010


  • City of Wonder
  • Performance

And I reckon I can beat your best case too.

Ugh… getting the bit representation of a Number (aka float) can’t be inlined.

I was impressed that removing the duplicated p1.x-p2.x and p1.y and p2.y didn’t give a speed improvement, the optimisation must be doing some common expression removal.

Good job.

More sqrt implementations than you can shake a stick at: http://www.codeproject.com/KB/cpp/Sqrt_Prec_VS_Speed.aspx