Arm® A64 Instruction Set Architecture: Future Architecture Technologies in the A architecture profile

FCVT

Floating-point convert precision (predicated).

Convert the size and precision of each active floating-point element of the source vector, and place the results in the corresponding elements of the destination vector. Inactive elements in the destination vector register remain unmodified.

Since the input and result types have a different size the smaller type is held unpacked in the least significant bits of elements of the larger size. When the input is the smaller type the upper bits of each source element are ignored. When the result is the smaller type the results are zero-extended to fill each destination element.

It has encodings from 6 classes: Half-precision to single-precision , Half-precision to double-precision , Single-precision to half-precision , Single-precision to double-precision , Double-precision to half-precision and Double-precision to single-precision

Half-precision to single-precision

if !HaveSVE() then UNDEFINED;
integer esize = 32;
integer g = UInt(Pg);
integer n = UInt(Zn);
integer d = UInt(Zd);
integer s_esize = 16;
integer d_esize = 32;

Half-precision to double-precision

if !HaveSVE() then UNDEFINED;
integer esize = 64;
integer g = UInt(Pg);
integer n = UInt(Zn);
integer d = UInt(Zd);
integer s_esize = 16;
integer d_esize = 64;

Single-precision to half-precision

if !HaveSVE() then UNDEFINED;
integer esize = 32;
integer g = UInt(Pg);
integer n = UInt(Zn);
integer d = UInt(Zd);
integer s_esize = 32;
integer d_esize = 16;

Single-precision to double-precision

if !HaveSVE() then UNDEFINED;
integer esize = 64;
integer g = UInt(Pg);
integer n = UInt(Zn);
integer d = UInt(Zd);
integer s_esize = 32;
integer d_esize = 64;

Double-precision to half-precision

if !HaveSVE() then UNDEFINED;
integer esize = 64;
integer g = UInt(Pg);
integer n = UInt(Zn);
integer d = UInt(Zd);
integer s_esize = 64;
integer d_esize = 16;

Double-precision to single-precision

if !HaveSVE() then UNDEFINED;
integer esize = 64;
integer g = UInt(Pg);
integer n = UInt(Zn);
integer d = UInt(Zd);
integer s_esize = 64;
integer d_esize = 32;

Assembler Symbols

CheckSVEEnabled();
integer elements = VL DIV esize;
bits(PL) mask = P[g];
bits(VL) operand  = Z[n];
bits(VL) result = Z[d];

for e = 0 to elements-1
    bits(esize) element = Elem[operand, e, esize];
    if ElemP[mask, e, esize] == '1' then
        bits(d_esize) res = FPConvertSVE(element<s_esize-1:0>, FPCR);
        Elem[result, e, esize] = ZeroExtend(res);

Z[d] = result;

Operational information

This instruction might be immediately preceded in program order by a MOVPRFX instruction. The MOVPRFX instruction must conform to all of the following requirements, otherwise the behavior of the MOVPRFX and this instruction is unpredictable:

• The MOVPRFX instruction must be unpredicated, or be predicated using the same governing predicate register and source element size as this instruction.
• The MOVPRFX instruction must specify the same destination register as this instruction.
• The destination register must not refer to architectural register state referenced by any other source operand register of this instruction.