Signed Multiply Accumulate Dual performs two signed 16 x 16-bit multiplications. It adds the products to a 32-bit accumulate operand.
Optionally, the instruction can exchange the halfwords of the second operand before performing the arithmetic. This produces top x bottom and bottom x top multiplication.
This instruction sets PSTATE.Q to 1 if the accumulate operation overflows. Overflow cannot occur during the multiplications.
It has encodings from the following instruction sets: A32 ( A1 ) and T32 ( T1 ) .
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 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 0 | Rd | != 1111 | Rm | 0 | 0 | M | 1 | Rn | |||||||||||||||
cond | Ra |
if Ra == '1111' then SEE "SMUAD"; constant d = UInt(Rd); constant n = UInt(Rn); constant m = UInt(Rm); constant a = UInt(Ra); constant m_swap = (M == '1'); if d == 15 || n == 15 || m == 15 then UNPREDICTABLE;
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 | 1 | 1 | 0 | 1 | 1 | 0 | 0 | 1 | 0 | Rn | != 1111 | Rd | 0 | 0 | 0 | M | Rm | ||||||||||||
Ra |
if Ra == '1111' then SEE "SMUAD"; constant d = UInt(Rd); constant n = UInt(Rn); constant m = UInt(Rm); constant a = UInt(Ra); constant m_swap = (M == '1'); // Armv8-A removes UNPREDICTABLE for R13 if d == 15 || n == 15 || m == 15 then UNPREDICTABLE;
For more information about the constrained unpredictable behavior of this instruction, see Architectural Constraints on UNPREDICTABLE behaviors.
<c> |
<q> |
<Rd> |
Is the general-purpose destination register, encoded in the "Rd" field. |
<Rn> |
Is the first general-purpose source register, encoded in the "Rn" field. |
<Rm> |
Is the second general-purpose source register, encoded in the "Rm" field. |
<Ra> |
Is the third general-purpose source register holding the addend, encoded in the "Ra" field. |
if ConditionPassed() then EncodingSpecificOperations(); constant operand2 = if m_swap then ROR(R[m],16) else R[m]; constant product1 = SInt(R[n]<15:0>) * SInt(operand2<15:0>); constant product2 = SInt(R[n]<31:16>) * SInt(operand2<31:16>); constant result = product1 + product2 + SInt(R[a]); R[d] = result<31:0>; if result != SInt(result<31:0>) then // Signed overflow PSTATE.Q = '1';
Internal version only: isa v01_31, pseudocode v2024-03_rel ; Build timestamp: 2024-03-25T10:05
Copyright © 2010-2024 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.