This is a collection of Intel®’ IA32® Software Developer's Manuals (URL of the day) and AMD' AMD64 Architecture Programmer's Manual together with the related specifications, application notes, white papers, and change logs. The collection aims to keep all available revisions. It was originally created by Michal Necasek, see OS/2 Museum.

If you have a public document, related to the IA32® specifications and missing from the collection, please mail it to me. The content of this URL and all sub-ULRs is available for convenient bulk download by rsync x86docs password "" (empty).

VFMSL (vector) -- AArch32

VFMSL (vector)

Vector Floating-point Multiply-Subtract Long from accumulator (vector). This instruction negates the values in the vector of one SIMD&FP register, multiplies these with the corresponding values in another vector, and accumulates the product to the corresponding vector element of the destination SIMD&FP register. The instruction does not round the result of the multiply before the accumulation.

Depending on settings in the CPACR, NSACR, HCPTR, and FPEXC 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.

In Armv8.2 and Armv8.3, this is an OPTIONAL instruction. From Armv8.4 it is mandatory for all implementations to support it.


Note

ID_ISAR6.FHM indicates whether this instruction is supported.


It has encodings from the following instruction sets: A32 ( A1 ) and T32 ( T1 ) .

A1
(FEAT_FHM)

313029282726252423222120191817161514131211109876543210
111111001D10VnVd1000NQM1Vm
Sop2op3op4U

Encoding for the 64-bit SIMD vector variant

Applies when (Q == 0)

VFMSL{<q>}.F16 <Dd>, <Sn>, <Sm>

Encoding for the 128-bit SIMD vector variant

Applies when (Q == 1)

VFMSL{<q>}.F16 <Qd>, <Dn>, <Dm>

Decode for all variants of this encoding

if !IsFeatureImplemented(FEAT_FHM) then Undefined(); end; if Q == '1' && Vd[0] == '1' then Undefined(); end; let d : integer = UInt(D::Vd); let n : integer = if Q == '1' then UInt(N::Vn) else UInt(Vn::N); let m : integer = if Q == '1' then UInt(M::Vm) else UInt(Vm::M); let esize : integer{} = 32; let datasize : integer{} = 32 << UInt(Q); let sub_op : boolean = S == '1'; let regs : integer = if Q == '0' then 1 else 2;

T1
(FEAT_FHM)

15141312111098765432101514131211109876543210
111111001D10VnVd1000NQM1Vm
Sop2op3op4U

Encoding for the 64-bit SIMD vector variant

Applies when (Q == 0)

VFMSL{<q>}.F16 <Dd>, <Sn>, <Sm>

Encoding for the 128-bit SIMD vector variant

Applies when (Q == 1)

VFMSL{<q>}.F16 <Qd>, <Dn>, <Dm>

Decode for all variants of this encoding

if InITBlock() then UnpredictableProcedure(); end; if !IsFeatureImplemented(FEAT_FHM) then Undefined(); end; if Q == '1' && Vd[0] == '1' then Undefined(); end; let d : integer = UInt(D::Vd); let n : integer = if Q == '1' then UInt(N::Vn) else UInt(Vn::N); let m : integer = if Q == '1' then UInt(M::Vm) else UInt(Vm::M); let esize : integer{} = 32; let regs : integer = 1 << UInt(Q); let datasize : integer{} = 32 << UInt(Q); let sub_op : boolean = S == '1';

Assembler Symbols

<q>

See Standard assembler syntax fields.

<Dd>

Is the 64-bit name of the SIMD&FP destination register, encoded in the "D:Vd" field.

<Sn>

Is the 32-bit name of the first SIMD&FP source register, encoded in the "Vn:N" field.

<Sm>

Is the 32-bit name of the second SIMD&FP source register, encoded in the "Vm:M" field.

<Qd>

Is the 128-bit name of the SIMD&FP destination register, encoded in the "D:Vd" field as <Qd>*2.

<Dn>

Is the 64-bit name of the first SIMD&FP source register, encoded in the "N:Vn" field.

<Dm>

Is the 64-bit name of the second SIMD&FP source register, encoded in the "M:Vm" field.

Operation

CheckAdvSIMDEnabled(); let fpcr : FPCR_Type = StandardFPCR(); var operand1 : bits(datasize); var operand2 : bits(datasize); var operand3 : bits(64); var result : bits(64); var element1 : bits(esize DIV 2); var element2 : bits(esize DIV 2); if Q=='0' then operand1 = S(n)[datasize-1:0]; operand2 = S(m)[datasize-1:0]; else operand1 = D(n)[datasize-1:0]; operand2 = D(m)[datasize-1:0]; end; for r = 0 to regs-1 do operand3 = D(d+r); for e = 0 to 1 do element1 = operand1[(2*r+e)*:(esize DIV 2)]; element2 = operand2[(2*r+e)*:(esize DIV 2)]; if sub_op then element1 = FPNeg{esize DIV 2}(element1, fpcr); end; result[e*:esize] = FPMulAddH(operand3[e*:esize], element1, element2, fpcr); end; D(d+r) = result; end;


2026-03_rel 2026-03-26 20:48:11

Copyright © 2010-2026 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.