fmaxnm

Floating-Point Max Number (Scalar)

FMAXNM <Hd|Sd|Dd>, <Hn|Sn|Dn>, <Hm|Sm|Dm>

Returns larger value, handling NaNs according to IEEE 754-2008 'maxNum'.

Details

Floating-point maximum number: returns the larger of two floating-point numbers Vn and Vm according to IEEE 754-2008 'maxNum' semantics, where if one operand is NaN and the other is a number, the number is returned. Supports half-precision (H), single-precision (S), and double-precision (D) floating-point formats. No condition flags are affected; exceptions may be generated for invalid operations or input denormals depending on FPCR settings. AArch64 only.

Pseudocode Operation

if HaveFPExt() then
  Vd ← FPMaxNum(Vn, Vm)
else
  UNDEFINED

Example

FMAXNM Dd, Dn, Dm

Encoding

Binary Layout
0
0
0
11110
00
1
Rm
01
10
10
Rn
Rd
 
Format FP Data Processing
Opcode 0x1E206800
Extension Floating Point

Operands

  • Vd
    Destination SIMD/FP vector register
  • Vn
    First source SIMD/FP vector register
  • Vm
    Second source SIMD/FP vector register

Reference (Arm A64 ISA)

Instruction Forms

Encoding Instruction ISA Bit pattern
0x0E400400 FMAXNM <Vd>.<T>, <Vn>.<T>, <Vm>.<T> A64 0 | Q | 0 | 01110 | 0 | 10 | Rm | 00 | 000 | 1 | Rn | Rd
0x0E20C400 FMAXNM <Vd>.<T>, <Vn>.<T>, <Vm>.<T> A64 0 | Q | 0 | 01110 | 0 | sz | 1 | Rm | 11000 | 1 | Rn | Rd
0x1EE06800 FMAXNM <Hd>, <Hn>, <Hm> A64 0 | 0 | 0 | 11110 | 11 | 1 | Rm | 01 | 10 | 10 | Rn | Rd
0x1E206800 FMAXNM <Sd>, <Sn>, <Sm> A64 0 | 0 | 0 | 11110 | 00 | 1 | Rm | 01 | 10 | 10 | Rn | Rd
0x1E606800 FMAXNM <Dd>, <Dn>, <Dm> A64 0 | 0 | 0 | 11110 | 01 | 1 | Rm | 01 | 10 | 10 | Rn | Rd
0xC120A120 FMAXNM { <Zdn1>.<T>-<Zdn2>.<T> }, { <Zdn1>.<T>-<Zdn2>.<T> }, <Zm>.<T> A64 11000001 | size | 10 | Zm | 101000 | 0100 | 1 | Zdn | 0
0xC120A920 FMAXNM { <Zdn1>.<T>-<Zdn4>.<T> }, { <Zdn1>.<T>-<Zdn4>.<T> }, <Zm>.<T> A64 11000001 | size | 10 | Zm | 101010 | 0100 | 1 | Zdn | 0 | 0
0xC120B120 FMAXNM { <Zdn1>.<T>-<Zdn2>.<T> }, { <Zdn1>.<T>-<Zdn2>.<T> }, { <Zm1>.<T>-<Zm2>.<T> } A64 11000001 | size | 1 | Zm | 0101100 | 010 | 0 | 1 | Zdn | 0
0xC120B920 FMAXNM { <Zdn1>.<T>-<Zdn4>.<T> }, { <Zdn1>.<T>-<Zdn4>.<T> }, { <Zm1>.<T>-<Zm4>.<T> } A64 11000001 | size | 1 | Zm | 00101110 | 010 | 0 | 1 | Zdn | 0 | 0
0x651C8000 FMAXNM <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <const> A64 01100101 | size | 011 | 10 | 0 | 100 | Pg | 0000 | i1 | Zdn
0x65048000 FMAXNM <Zdn>.<T>, <Pg>/M, <Zdn>.<T>, <Zm>.<T> A64 01100101 | size | 00 | 010 | 0 | 100 | Pg | Zm | Zdn

Description

Floating-point Maximum Number (scalar). This instruction compares the first and second source SIMD&FP register values, and writes the larger of the two floating-point values to the destination SIMD&FP register. Regardless of the value of FPCR.AH, the behavior is as follows: This instruction can generate a floating-point exception. Depending on the settings in FPCR, the exception results in either a flag being set in FPSR, or a synchronous exception being generated. For more information, see Floating-point exception traps. Depending on the settings in the CPACR_EL1, CPTR_EL2, and CPTR_EL3 registers, and the current Security state and Exception level, an attempt to execute the instruction might be trapped.

Operation

CheckFPEnabled64();
bits(esize) operand1 = V[n, esize];
bits(esize) operand2 = V[m, esize];

boolean merge = IsMerging(FPCR);
bits(128) result = if merge then V[n, 128] else Zeros(128);

Elem[result, 0, esize] = FPMaxNum(operand1, operand2, FPCR);
V[d, 128] = result;