FTMAD

Floating-point trigonometric multiply-add coefficient

The FTMAD instruction calculates the series terms for either sin(x) or cos(x), where the argument x has been adjusted to be in the range -π/4 < x ≤ π/4.

To calculate the series terms of sin(x) and cos(x) the initial source operands of FTMAD should be zero in the first source vector and x² in the second source vector. The FTMAD instruction is then executed eight times to calculate the sum of eight series terms, which gives a result of sufficient precision.

The FTMAD instruction multiplies each element of the first source vector by the absolute value of the corresponding element of the second source vector and performs a fused addition of each product with a value obtained from a table of hard-wired coefficients, and places the results destructively in the first source vector.

The coefficients are different for sin(x) and cos(x), and are selected by a combination of the sign bit in the second source element and an immediate index in the range 0 to 7.

This instruction is unpredicated.

if !HaveSVE() then UNDEFINED; if size == '00' then UNDEFINED; integer esize = 8 << UInt(size); integer dn = UInt(Zdn); integer m = UInt(Zm); integer imm = UInt(imm3);