Lines Matching refs:reciprocal
79 // [1, 2.0) and get a Q32 approximate reciprocal using a small minimax
80 // polynomial approximation: reciprocal = 3/4 + 1/sqrt(2) - b/2. This
85 // Now refine the reciprocal estimate using a Newton-Raphson iteration:
109 uint64_t correction, reciprocal;
113 reciprocal = (uint64_t)recip32*cHi + ((uint64_t)recip32*cLo >> 32);
116 // 64-bit reciprocal estimate downward to ensure that it is strictly smaller
117 // than the infinitely precise exact reciprocal. Because the computation
120 reciprocal -= 2;
122 // The numerical reciprocal is accurate to within 2^-56, lies in the
123 // interval [0.5, 1.0), and is strictly smaller than the true reciprocal
124 // of b. Multiplying a by this reciprocal thus gives a numerical q = a/b
135 wideMultiply(aSignificand << 2, reciprocal, "ient, "ientLo);