FMAX (multiple vectors)

Multi-vector floating-point maximum

Determine the maximum of floating-point elements of the two or four second source vectors and the corresponding floating-point elements of the two or four first source vectors and destructively place the results in the corresponding elements of the two or four first source vectors. If either element value is NaN then the result is NaN.

This instruction follows SME2 floating-point numerical behaviors corresponding to instructions that place their results in one or more SVE Z vectors.

This instruction is unpredicated.

It has encodings from 2 classes: Two registers and Four registers

Two registers
(FEAT_SME2)

313029282726252423222120191817161514131211109876543210
11000001!= 001Zm010110001000Zdn0
size

FMAX { <Zdn1>.<T>-<Zdn2>.<T> }, { <Zdn1>.<T>-<Zdn2>.<T> }, { <Zm1>.<T>-<Zm2>.<T> }

if !HaveSME2() then UNDEFINED; constant integer esize = 8 << UInt(size); integer dn = UInt(Zdn:'0'); integer m = UInt(Zm:'0'); constant integer nreg = 2;

Four registers
(FEAT_SME2)

313029282726252423222120191817161514131211109876543210
11000001!= 001Zm0010111001000Zdn00
size

FMAX { <Zdn1>.<T>-<Zdn4>.<T> }, { <Zdn1>.<T>-<Zdn4>.<T> }, { <Zm1>.<T>-<Zm4>.<T> }

if !HaveSME2() then UNDEFINED; constant integer esize = 8 << UInt(size); integer dn = UInt(Zdn:'00'); integer m = UInt(Zm:'00'); constant integer nreg = 4;

Assembler Symbols

<Zdn1>

For the two registers variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zdn" times 2.

For the four registers variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zdn" times 4.

<T>

Is the size specifier, encoded in size:

size <T>
01 H
10 S
11 D
<Zdn4>

Is the name of the fourth scalable vector register of a multi-vector sequence, encoded as "Zdn" times 4 plus 3.

<Zdn2>

Is the name of the second scalable vector register of a multi-vector sequence, encoded as "Zdn" times 2 plus 1.

<Zm1>

For the two registers variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zm" times 2.

For the four registers variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zm" times 4.

<Zm4>

Is the name of the fourth scalable vector register of a multi-vector sequence, encoded as "Zm" times 4 plus 3.

<Zm2>

Is the name of the second scalable vector register of a multi-vector sequence, encoded as "Zm" times 2 plus 1.

Operation

CheckStreamingSVEEnabled(); constant integer VL = CurrentVL; constant integer elements = VL DIV esize; array [0..3] of bits(VL) results; for r = 0 to nreg-1 bits(VL) operand1 = Z[dn+r, VL]; bits(VL) operand2 = Z[m+r, VL]; for e = 0 to elements-1 bits(esize) element1 = Elem[operand1, e, esize]; bits(esize) element2 = Elem[operand2, e, esize]; Elem[results[r], e, esize] = FPMax(element1, element2, FPCR[]); for r = 0 to nreg-1 Z[dn+r, VL] = results[r];


Internal version only: isa v33.53, AdvSIMD v29.11, pseudocode v2022-09_rel, sve v2022-09_rel ; Build timestamp: 2022-09-30T16:37

Copyright © 2010-2022 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.