Signed integer indexed dot product

The signed integer indexed dot product instruction computes the dot product of a group of four signed 8-bit or 16-bit integer values held in each 32-bit or 64-bit element of the first source vector multiplied by a group of four signed 8-bit or 16-bit integer values in an indexed 32-bit or 64-bit element of the second source vector, and then destructively adds the widened dot product to the corresponding 32-bit or 64-bit element of the destination vector.

The groups within the second source vector are specified using an immediate index which selects the same group position within each 128-bit vector segment. The index range is from 0 to one less than the number of groups per 128-bit segment, encoded in 1 to 2 bits depending on the size of the group. This instruction is unpredicated.

It has encodings from 2 classes: 32-bit and 64-bit

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 |

0 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 0 | 1 | i2 | Zm | 0 | 0 | 0 | 0 | 0 | 0 | Zn | Zda | |||||||||||

size<1> | size<0> | U |

if !IsFeatureImplemented(FEAT_SVE) && !IsFeatureImplemented(FEAT_SME) then UNDEFINED; constant integer esize = 32; constant integer index = UInt(i2); constant integer n = UInt(Zn); constant integer m = UInt(Zm); constant integer da = UInt(Zda);

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 |

0 | 1 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 1 | 1 | i1 | Zm | 0 | 0 | 0 | 0 | 0 | 0 | Zn | Zda | |||||||||||

size<1> | size<0> | U |

if !IsFeatureImplemented(FEAT_SVE) && !IsFeatureImplemented(FEAT_SME) then UNDEFINED; constant integer esize = 64; constant integer index = UInt(i1); constant integer n = UInt(Zn); constant integer m = UInt(Zm); constant integer da = UInt(Zda);

<Zda> |
Is the name of the third source and destination scalable vector register, encoded in the "Zda" field. |

<Zn> |
Is the name of the first source scalable vector register, encoded in the "Zn" field. |

CheckSVEEnabled(); constant integer VL = CurrentVL; constant integer PL = VL DIV 8; constant integer elements = VL DIV esize; constant integer eltspersegment = 128 DIV esize; constant bits(VL) operand1 = Z[n, VL]; constant bits(VL) operand2 = Z[m, VL]; constant bits(VL) operand3 = Z[da, VL]; bits(VL) result; for e = 0 to elements-1 constant integer segmentbase = e - (e MOD eltspersegment); constant integer s = segmentbase + index; bits(esize) res = Elem[operand3, e, esize]; for i = 0 to 3 constant integer element1 = SInt(Elem[operand1, 4 * e + i, esize DIV 4]); constant integer element2 = SInt(Elem[operand2, 4 * s + i, esize DIV 4]); res = res + element1 * element2; Elem[result, e, esize] = res; Z[da, VL] = result;

This instruction might be immediately preceded in program order by a MOVPRFX instruction. The MOVPRFX must conform to all of the following requirements, otherwise the behavior of the MOVPRFX and this instruction is constrained unpredictable:

- The MOVPRFX must be unpredicated.
- The MOVPRFX must specify the same destination register as this instruction.
- The destination register must not refer to architectural register state referenced by any other source operand register of this instruction.

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.