simple fpc vs gcc benchmark

[fcu]

thanks to everybody
the trick is
inlining those functions :

function Vec2(x, y: Single): TVec2;
function Vec3(x, y, z: Single): TVec3;
function Vertex(pos: TVec3; uv: TVec2): TVertex;
function Triangle(v0, v1, v2: TVertex);TTriangle;

and precompute (cc := Vec2(x+0.5,y+0.5) as d4eva mentioned

the fps becomes 46 from 16( original ) , that seems good and its near equal to the c++ version
that shows that fpc is doing great job

[backprop]

Also consider to use sincos function for the same argument, instead to calculate it separately, as these functions are complementary and no extra cycles are wasted to calculate it separately.