Vector Fused Multiply Subtract negates the elements of one vector and multiplies them with the corresponding elements of another vector, adds the products to the corresponding elements of the destination vector, and places the results in the destination vector. The instruction does not round the result of the multiply before the addition.
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.
It has encodings from the following instruction sets: A32 ( A1 and A2 ) and T32 ( T1 and T2 ) .
31 | 30 | 29 | 28 | 27 | 26 | 25 | 24 | 23 | 22 | 21 | 20 | 19 | 18 | 17 | 16 | 15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
1 | 1 | 1 | 1 | 0 | 0 | 1 | 0 | 0 | D | 1 | sz | Vn | Vd | 1 | 1 | 0 | 0 | N | Q | M | 1 | Vm | |||||||||
op |
if Q == '1' && (Vd<0> == '1' || Vn<0> == '1' || Vm<0> == '1') then UNDEFINED; if sz == '1' && !HaveFP16Ext() then UNDEFINED; advsimd = TRUE; op1_neg = (op == '1'); integer esize; integer elements; case sz of when '0' esize = 32; elements = 2; when '1' esize = 16; elements = 4; d = UInt(D:Vd); n = UInt(N:Vn); m = UInt(M:Vm); regs = if Q == '0' then 1 else 2;
31 | 30 | 29 | 28 | 27 | 26 | 25 | 24 | 23 | 22 | 21 | 20 | 19 | 18 | 17 | 16 | 15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
!= 1111 | 1 | 1 | 1 | 0 | 1 | D | 1 | 0 | Vn | Vd | 1 | 0 | size | N | 1 | M | 0 | Vm | |||||||||||||
cond | op |
if FPSCR.Len != '000' || FPSCR.Stride != '00' then UNDEFINED; if size == '00' || (size == '01' && !HaveFP16Ext()) then UNDEFINED; if size == '01' && cond != '1110' then UNPREDICTABLE; advsimd = FALSE; op1_neg = (op == '1'); integer esize; integer d; integer n; integer m; case size of when '01' esize = 16; d = UInt(Vd:D); n = UInt(Vn:N); m = UInt(Vm:M); when '10' esize = 32; d = UInt(Vd:D); n = UInt(Vn:N); m = UInt(Vm:M); when '11' esize = 64; d = UInt(D:Vd); n = UInt(N:Vn); m = UInt(M:Vm); boolean floating_point = boolean UNKNOWN; integer regs = integer UNKNOWN; integer elements = integer UNKNOWN;
If size == '01' && cond != '1110', then one of the following behaviors must occur:
15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 | 15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 0 | D | 1 | sz | Vn | Vd | 1 | 1 | 0 | 0 | N | Q | M | 1 | Vm | |||||||||
op |
if Q == '1' && (Vd<0> == '1' || Vn<0> == '1' || Vm<0> == '1') then UNDEFINED; if sz == '1' && !HaveFP16Ext() then UNDEFINED; if sz == '1' && InITBlock() then UNPREDICTABLE; advsimd = TRUE; op1_neg = (op == '1'); integer esize; integer elements; case sz of when '0' esize = 32; elements = 2; when '1' esize = 16; elements = 4; d = UInt(D:Vd); n = UInt(N:Vn); m = UInt(M:Vm); regs = if Q == '0' then 1 else 2;
If sz == '1' && InITBlock(), then one of the following behaviors must occur:
15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 | 15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
1 | 1 | 1 | 0 | 1 | 1 | 1 | 0 | 1 | D | 1 | 0 | Vn | Vd | 1 | 0 | size | N | 1 | M | 0 | Vm | ||||||||||
op |
if FPSCR.Len != '000' || FPSCR.Stride != '00' then UNDEFINED; if size == '00' || (size == '01' && !HaveFP16Ext()) then UNDEFINED; if size == '01' && InITBlock() then UNPREDICTABLE; advsimd = FALSE; op1_neg = (op == '1'); integer esize; integer d; integer n; integer m; case size of when '01' esize = 16; d = UInt(Vd:D); n = UInt(Vn:N); m = UInt(Vm:M); when '10' esize = 32; d = UInt(Vd:D); n = UInt(Vn:N); m = UInt(Vm:M); when '11' esize = 64; d = UInt(D:Vd); n = UInt(N:Vn); m = UInt(M:Vm); boolean floating_point = boolean UNKNOWN; integer regs = integer UNKNOWN; integer elements = integer UNKNOWN;
If size == '01' && InITBlock(), then one of the following behaviors must occur:
<c> |
For encoding A1: see Standard assembler syntax fields. This encoding must be unconditional. |
For encoding A2, T1 and T2: see Standard assembler syntax fields. |
<q> |
<dt> |
Is the data type for the elements of the vectors,
encoded in
|
<Qd> |
Is the 128-bit name of the SIMD&FP destination register, encoded in the "D:Vd" field as <Qd>*2. |
<Qn> |
Is the 128-bit name of the first SIMD&FP source register, encoded in the "N:Vn" field as <Qn>*2. |
<Qm> |
Is the 128-bit name of the second SIMD&FP source register, encoded in the "M:Vm" field as <Qm>*2. |
<Dd> |
Is the 64-bit name of the SIMD&FP destination register, encoded in the "D:Vd" field. |
<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. |
<Sd> |
Is the 32-bit name of the SIMD&FP destination register, encoded in the "Vd:D" 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. |
if ConditionPassed() then EncodingSpecificOperations(); CheckAdvSIMDOrVFPEnabled(TRUE, advsimd); if advsimd then // Advanced SIMD instruction for r = 0 to regs-1 for e = 0 to elements-1 bits(esize) op1 = Elem[D[n+r],e,esize]; if op1_neg then op1 = FPNeg(op1); Elem[D[d+r],e,esize] = FPMulAdd(Elem[D[d+r],e,esize], op1, Elem[D[m+r],e,esize], StandardFPSCRValue()); else // VFP instruction case esize of when 16 op16 = if op1_neg then FPNeg(S[n]<15:0>) else S[n]<15:0>; S[d] = Zeros(16) : FPMulAdd(S[d]<15:0>, op16, S[m]<15:0>, FPSCR[]); when 32 op32 = if op1_neg then FPNeg(S[n]) else S[n]; S[d] = FPMulAdd(S[d], op32, S[m], FPSCR[]); when 64 op64 = if op1_neg then FPNeg(D[n]) else D[n]; D[d] = FPMulAdd(D[d], op64, D[m], FPSCR[]);
Internal version only: isa v01_31, pseudocode v2023-06_rel, sve v2023-06_rel ; Build timestamp: 2023-07-04T18:06
Copyright © 2010-2023 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.