FMULX (by element)

Floating-point Multiply extended (by element). This instruction multiplies the floating-point values in the vector elements in the first source SIMD&FP register by the specified floating-point value in the second source SIMD&FP register, places the results in a vector, and writes the vector to the destination SIMD&FP register.

If one value is zero and the other value is infinite, the result is 2.0. In this case, the result is negative if only one of the values is negative, otherwise the result is positive.

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.

It has encodings from 4 classes: Scalar, half-precision , Scalar, single-precision and double-precision , Vector, half-precision and Vector, single-precision and double-precision

Scalar, half-precision
(FEAT_FP16)

313029282726252423222120191817161514131211109876543210
0111111100LMRm1001H0RnRd
Usizeopcode

FMULX <Hd>, <Hn>, <Vm>.H[<index>]

if !IsFeatureImplemented(FEAT_FP16) then UNDEFINED; constant integer idxdsize = 64 << UInt(H); integer n = UInt(Rn); integer m = UInt(Rm); integer d = UInt(Rd); integer index = UInt(H:L:M); constant integer esize = 16; constant integer datasize = esize; integer elements = 1; boolean mulx_op = (U == '1');

Scalar, single-precision and double-precision

313029282726252423222120191817161514131211109876543210
011111111szLMRm1001H0RnRd
Uopcode

FMULX <V><d>, <V><n>, <Vm>.<Ts>[<index>]

constant integer idxdsize = 64 << UInt(H); integer index; bit Rmhi = M; case sz:L of when '0x' index = UInt(H:L); when '10' index = UInt(H); when '11' UNDEFINED; integer d = UInt(Rd); integer n = UInt(Rn); integer m = UInt(Rmhi:Rm); constant integer esize = 32 << UInt(sz); constant integer datasize = esize; integer elements = 1; boolean mulx_op = (U == '1');

Vector, half-precision
(FEAT_FP16)

313029282726252423222120191817161514131211109876543210
0Q10111100LMRm1001H0RnRd
Usizeopcode

FMULX <Vd>.<T>, <Vn>.<T>, <Vm>.H[<index>]

if !IsFeatureImplemented(FEAT_FP16) then UNDEFINED; constant integer idxdsize = 64 << UInt(H); integer n = UInt(Rn); integer m = UInt(Rm); integer d = UInt(Rd); integer index = UInt(H:L:M); constant integer esize = 16; constant integer datasize = 64 << UInt(Q); integer elements = datasize DIV esize; boolean mulx_op = (U == '1');

Vector, single-precision and double-precision

313029282726252423222120191817161514131211109876543210
0Q1011111szLMRm1001H0RnRd
Uopcode

FMULX <Vd>.<T>, <Vn>.<T>, <Vm>.<Ts>[<index>]

constant integer idxdsize = 64 << UInt(H); integer index; bit Rmhi = M; case sz:L of when '0x' index = UInt(H:L); when '10' index = UInt(H); when '11' UNDEFINED; integer d = UInt(Rd); integer n = UInt(Rn); integer m = UInt(Rmhi:Rm); if sz:Q == '10' then UNDEFINED; constant integer esize = 32 << UInt(sz); constant integer datasize = 64 << UInt(Q); integer elements = datasize DIV esize; boolean mulx_op = (U == '1');

Assembler Symbols

<Hd>

Is the 16-bit name of the SIMD&FP destination register, encoded in the "Rd" field.

<Hn>

Is the 16-bit name of the first SIMD&FP source register, encoded in the "Rn" field.

<Vm>

For the half-precision variant: is the name of the second SIMD&FP source register, in the range V0 to V15, encoded in the "Rm" field.

For the single-precision and double-precision variant: is the name of the second SIMD&FP source register, encoded in the "M:Rm" fields.

<index>

For the half-precision variant: is the element index, in the range 0 to 7, encoded in the "H:L:M" fields.

For the single-precision and double-precision variant: is the element index, encoded in sz:L:H:

sz L <index>
0 x UInt(H:L)
1 0 UInt(H)
1 1 RESERVED
<V>

Is a width specifier, encoded in sz:

sz <V>
0 S
1 D
<d>

Is the number of the SIMD&FP destination register, encoded in the "Rd" field.

<n>

Is the number of the first SIMD&FP source register, encoded in the "Rn" field.

<Ts>

Is an element size specifier, encoded in sz:

sz <Ts>
0 S
1 D
<Vd>

Is the name of the SIMD&FP destination register, encoded in the "Rd" field.

<T>

For the half-precision variant: is an arrangement specifier, encoded in Q:

Q <T>
0 4H
1 8H

For the single-precision and double-precision variant: is an arrangement specifier, encoded in Q:sz:

Q sz <T>
0 0 2S
0 1 RESERVED
1 0 4S
1 1 2D
<Vn>

Is the name of the first SIMD&FP source register, encoded in the "Rn" field.

Operation

CheckFPAdvSIMDEnabled64(); bits(datasize) operand1 = V[n, datasize]; bits(idxdsize) operand2 = V[m, idxdsize]; bits(esize) element1; bits(esize) element2 = Elem[operand2, index, esize]; boolean merge = elements == 1 && IsMerging(FPCR); bits(128) result = if merge then V[n, 128] else Zeros(128); for e = 0 to elements-1 element1 = Elem[operand1, e, esize]; if mulx_op then Elem[result, e, esize] = FPMulX(element1, element2, FPCR); else Elem[result, e, esize] = FPMul(element1, element2, FPCR); V[d, 128] = result;


Internal version only: aarchmrs v2024-03_relA, pseudocode v2024-03_rel, sve v2024-03_rel ; Build timestamp: 2024-03-26T09:45

Copyright © 2010-2024 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.