Hello all, I''m trying to improve my exponential function and need some help. The current version is posted below:

float pow(float base,float exp) { float tmpVal; _asm { fld [exp]; fld [base]; fyl2x; // exp * log2(base) fld1; fld st(1); fprem; // remainder ((exp * log2(base)) / 1) f2xm1; // (2 ^ remainder) - 1 faddp st(1),st(0); // (exp * log2(base)) + ((2 ^ remainder) - 1) fscale; // (2 ^ int(st1)) * st0 fstp [tmpVal]; // Phew done ;) } return tmpVal; }

Now, what I''d like to do is split up on of these operations into an integer/float part so that it can work in parallel. Unfortunately this has proven much harder to do then I thought. For example, all the fprem does is find the fractional part of st(0), but coming up with a method to do that quickly in integer math is proving very difficult. Any suggestions? Gary

Maybe a look-up table?