Multi-vector polynomial multiply long and accumulate vectors
Polynomial multiply over [0, 1] the corresponding even-numbered elements of the first and second source vectors, and bitwise exclusive-OR the result with the overlapping double-width elements of the first destination vector. Perform the same operation with odd-numbered elements of the source vectors, writing to the second destination vector. This instruction is unpredicated.
This instruction is legal in Streaming SVE mode if both FEAT_SSVE_AES and FEAT_SVE_AES2 are implemented.
| 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 | 1 | 0 | 0 | 1 | Zm | 1 | 1 | 1 | 1 | 1 | 1 | Zn | Zda | 0 | |||||||||||
| size | |||||||||||||||||||||||||||||||
if !IsFeatureImplemented(FEAT_SVE_AES2) then EndOfDecode(Decode_UNDEF); constant integer esize = 128; constant integer n = UInt(Zn); constant integer m = UInt(Zm); constant integer da = UInt(Zda:'0');
| <Zda1> |
Is the name of the first scalable vector register of the destination multi-vector group, encoded as "Zda" times 2. |
| <Zda2> |
Is the name of the second scalable vector register of the destination multi-vector group, encoded as "Zda" times 2 plus 1. |
| <Zn> |
Is the name of the first source scalable vector register, encoded in the "Zn" field. |
| <Zm> |
Is the name of the second source scalable vector register, encoded in the "Zm" field. |
if IsFeatureImplemented(FEAT_SSVE_AES) then CheckSVEEnabled(); else CheckNonStreamingSVEEnabled(); constant integer VL = CurrentVL; constant integer elements = VL DIV esize; constant bits(VL) operand1 = Z[n, VL]; constant bits(VL) operand2 = Z[m, VL]; bits(VL) result_lo = Z[da + 0, VL]; bits(VL) result_hi = Z[da + 1, VL]; for e = 0 to elements-1 constant bits(esize DIV 2) element1_lo = Elem[operand1, 2*e + 0, esize DIV 2]; constant bits(esize DIV 2) element2_lo = Elem[operand2, 2*e + 0, esize DIV 2]; constant bits(esize DIV 2) element1_hi = Elem[operand1, 2*e + 1, esize DIV 2]; constant bits(esize DIV 2) element2_hi = Elem[operand2, 2*e + 1, esize DIV 2]; constant bits(esize) product_lo = PolynomialMult(element1_lo, element2_lo); constant bits(esize) product_hi = PolynomialMult(element1_hi, element2_hi); Elem[result_lo, e, esize] = Elem[result_lo, e, esize] EOR product_lo; Elem[result_hi, e, esize] = Elem[result_hi, e, esize] EOR product_hi; Z[da + 0, VL] = result_lo; Z[da + 1, VL] = result_hi;
Internal version only: aarchmrs v2024-12_rel, pseudocode v2024-12_rel ; Build timestamp: 2024-12-15T22:18
Copyright © 2010-2024 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.