You copied the Doc URL to your clipboard.

FRECPS

Floating-point reciprocal step (unpredicated).

Multiply corresponding floating-point elements of the first and second source vectors, subtract the products from 2.0 without intermediate rounding and place the results in the corresponding elements of the destination vector. This instruction is unpredicated.

This instruction can be used to perform a single Newton-Raphson iteration for calculating the reciprocal of a vector of floating-point values.

313029282726252423222120191817161514131211109876543210
01100101size0Zm000110ZnZd

FRECPS <Zd>.<T>, <Zn>.<T>, <Zm>.<T>

if !HaveSVE() then UNDEFINED;
if size == '00' then UNDEFINED;
integer esize = 8 << UInt(size);
integer n = UInt(Zn);
integer m = UInt(Zm);
integer d = UInt(Zd);

Assembler Symbols

<Zd>

Is the name of the destination scalable vector register, encoded in the "Zd" field.

<T> Is the size specifier, encoded in size:
size <T>
00 RESERVED
01 H
10 S
11 D
<Zn>

Is the name of the first source scalable vector register, encoded in the "Zn" field.

<Zm>

Is the name of the second source scalable vector register, encoded in the "Zm" field.

Operation

CheckSVEEnabled();
integer elements = VL DIV esize;
bits(VL) operand1 = Z[n];
bits(VL) operand2 = Z[m];
bits(VL) result;

for e = 0 to elements-1
    bits(esize) element1 = Elem[operand1, e, esize];
    bits(esize) element2 = Elem[operand2, e, esize];
    Elem[result, e, esize] = FPRecipStepFused(element1, element2);

Z[d] = result;