Floating-point add strictly-ordered reduction, accumulating in scalar
Floating-point add a SIMD&FP scalar source and all active lanes of the vector source and place the result destructively in the SIMD&FP scalar source register. Vector elements are processed strictly in order from low to high, with the scalar source providing the initial value. Inactive elements in the source vector are ignored.
| 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 | 1 | 0 | 0 | 1 | 0 | 1 | size | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | Pg | Zm | Vdn | |||||||||||
if !HaveSVE() then UNDEFINED; if size == '00' then UNDEFINED; integer esize = 8 << UInt(size); integer g = UInt(Pg); integer dn = UInt(Vdn); integer m = UInt(Zm);
| <V> |
Is a width specifier,
encoded in
|
| <dn> |
Is the number [0-31] of the source and destination SIMD&FP register, encoded in the "Vdn" field. |
| <Pg> |
Is the name of the governing scalable predicate register P0-P7, encoded in the "Pg" field. |
| <Zm> |
Is the name of the source scalable vector register, encoded in the "Zm" field. |
| <T> |
Is the size specifier,
encoded in
|
CheckSVEEnabled(); integer elements = VL DIV esize; bits(PL) mask = P[g]; bits(esize) operand1 = V[dn]; bits(VL) operand2 = if AnyActiveElement(mask, esize) then Z[m] else Zeros(); bits(esize) result = operand1; for e = 0 to elements-1 if ElemP[mask, e, esize] == '1' then bits(esize) element = Elem[operand2, e, esize]; result = FPAdd(result, element, FPCR[]); V[dn] = result;
Internal version only: isa v32.15, AdvSIMD v29.05, pseudocode v2021-03, sve v2021-03_rc2 ; Build timestamp: 2021-03-30T21:36
Copyright © 2010-2021 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.