Lines Matching full:precision
333 * "Normalize" the terms in the infinite-precision expression a + b for
336 * the same precision as 'a' and the resulting b is the rounding error.)
343 * extra precision in a + b. This is required by C standards but broken
352 * any extra precision into the type of 'a' -- 'a' should have type float_t,
355 * reduce their own extra-precision and efficiciency problems. In
389 * Set x += c, where x is represented in extra precision as a + b.
399 * 2**20 times smaller than 'a' to give about 20 extra bits of precision.
741 /* double precision kernel functions */
753 /* float precision kernel functions */
771 /* long double precision kernel functions */
