vcmpe

Vector Compare Exception (VFP)

VCMPE<c>.F32 <Sd>, <Sm>

Compares values and raises exception on NaN.

Details

Compares two single-precision floating-point values in VFP registers and sets the FPSCR condition flags (N, Z, C, V) based on the comparison result. Unlike VCMP, this variant raises an Invalid Operation exception if either operand is NaN (signaling comparison). Executed in A32/T32 with VFP extension; exception behavior depends on FPSCR exception-enable bits.

Pseudocode Operation

if Sd is NaN or Sm is NaN then
  if FPSCR.IXE then raise InvalidOperationException
result ← compare(Sd, Sm)
FPSCR.N ← result.N
FPSCR.Z ← result.Z
FPSCR.C ← result.C
FPSCR.V ← result.V

Example

VCMPE.F32 s0, s2

Encoding

Binary Layout
cond
11101
D
11
0
100
Vd
10
10
1
1
M
0
Vm
 
Format VFP Compare
Opcode 0x0EB40AC0
Extension VFP (Float)

Operands

  • Sd
    Destination 32-bit floating-point register
  • Sm
    Second source 32-bit floating-point register

Reference (Arm AArch32 ISA)

Instruction Forms

Encoding Instruction ISA Bit pattern
0x0EB409C0 VCMPE{<c>}{<q>}.F16 <Sd>, <Sm> A32 cond | 11101 | D | 11 | 0 | 100 | Vd | 10 | 01 | 1 | 1 | M | 0 | Vm
0x0EB40AC0 VCMPE{<c>}{<q>}.F32 <Sd>, <Sm> A32 cond | 11101 | D | 11 | 0 | 100 | Vd | 10 | 10 | 1 | 1 | M | 0 | Vm
0x0EB40BC0 VCMPE{<c>}{<q>}.F64 <Dd>, <Dm> A32 cond | 11101 | D | 11 | 0 | 100 | Vd | 10 | 11 | 1 | 1 | M | 0 | Vm
0x0EB509C0 VCMPE{<c>}{<q>}.F16 <Sd>, #0.0 A32 cond | 11101 | D | 11 | 0 | 101 | Vd | 10 | 01 | 1 | 1 | 0 | 0 | 0000
0x0EB50AC0 VCMPE{<c>}{<q>}.F32 <Sd>, #0.0 A32 cond | 11101 | D | 11 | 0 | 101 | Vd | 10 | 10 | 1 | 1 | 0 | 0 | 0000
0x0EB50BC0 VCMPE{<c>}{<q>}.F64 <Dd>, #0.0 A32 cond | 11101 | D | 11 | 0 | 101 | Vd | 10 | 11 | 1 | 1 | 0 | 0 | 0000
0xEEB409C0 VCMPE{<c>}{<q>}.F16 <Sd>, <Sm> T32 111011101 | D | 11 | 0 | 100 | Vd | 10 | 01 | 1 | 1 | M | 0 | Vm
0xEEB40AC0 VCMPE{<c>}{<q>}.F32 <Sd>, <Sm> T32 111011101 | D | 11 | 0 | 100 | Vd | 10 | 10 | 1 | 1 | M | 0 | Vm
0xEEB40BC0 VCMPE{<c>}{<q>}.F64 <Dd>, <Dm> T32 111011101 | D | 11 | 0 | 100 | Vd | 10 | 11 | 1 | 1 | M | 0 | Vm
0xEEB509C0 VCMPE{<c>}{<q>}.F16 <Sd>, #0.0 T32 111011101 | D | 11 | 0 | 101 | Vd | 10 | 01 | 1 | 1 | 0 | 0 | 0000
0xEEB50AC0 VCMPE{<c>}{<q>}.F32 <Sd>, #0.0 T32 111011101 | D | 11 | 0 | 101 | Vd | 10 | 10 | 1 | 1 | 0 | 0 | 0000
0xEEB50BC0 VCMPE{<c>}{<q>}.F64 <Dd>, #0.0 T32 111011101 | D | 11 | 0 | 101 | Vd | 10 | 11 | 1 | 1 | 0 | 0 | 0000

Description

Vector Compare, raising Invalid Operation on NaN compares two floating-point registers, or one floating-point register and zero. It writes the result to the FPSCR flags. These are normally transferred to the PSTATE.{N, Z, C, V} Condition flags by a subsequent VMRS instruction. This instruction raises an Invalid Operation floating-point exception if either or both of the operands is any type of NaN. Depending on settings in the CPACR, NSACR, and HCPTR registers, and the Security state and PE mode in which the instruction is executed, an attempt to execute the instruction might be undefined, or trapped to Hyp mode. For more information see Enabling Advanced SIMD and floating-point support.

Operation

if ConditionPassed() then
    EncodingSpecificOperations();  CheckVFPEnabled(TRUE);
    bits(4) nzcv;
    case esize of
        when 16
            bits(16) op16 = if with_zero then FPZero('0', 16) else S[m]<15:0>;
            nzcv = FPCompare(S[d]<15:0>, op16, quiet_nan_exc, FPSCR[]);
        when 32
            bits(32) op32 = if with_zero then FPZero('0', 32) else S[m];
            nzcv = FPCompare(S[d], op32, quiet_nan_exc, FPSCR[]);
        when 64
            bits(64) op64 = if with_zero then FPZero('0', 64) else D[m];
            nzcv = FPCompare(D[d], op64, quiet_nan_exc, FPSCR[]);

    FPSCR<31:28> = nzcv; // FPSCR.<N,Z,C,V> set to nzcv