FCM<cc> (zero)

Floating-point compare vector with zero

Compare active floating-point elements in the source vector with zero, and place the boolean results of the specified comparison in the corresponding elements of the destination predicate. Inactive elements in the destination predicate register are set to zero. Does not set the condition flags.

<cc> Comparison
EQ equal
GE greater than or equal
GT greater than
LE less than or equal
LT less than
NE not equal
UO unordered

It has encodings from 6 classes: Equal , Greater than , Greater than or equal , Less than , Less than or equal and Not equal

Equal

313029282726252423222120191817161514131211109876543210
01100101size010010001PgZn0Pd
eqltne

FCMEQ <Pd>.<T>, <Pg>/Z, <Zn>.<T>, #0.0

if !HaveSVE() && !HaveSME() then UNDEFINED; if size == '00' then UNDEFINED; constant integer esize = 8 << UInt(size); integer g = UInt(Pg); integer n = UInt(Zn); integer d = UInt(Pd); SVECmp op = Cmp_EQ;

Greater than

313029282726252423222120191817161514131211109876543210
01100101size010000001PgZn1Pd
eqltne

FCMGT <Pd>.<T>, <Pg>/Z, <Zn>.<T>, #0.0

if !HaveSVE() && !HaveSME() then UNDEFINED; if size == '00' then UNDEFINED; constant integer esize = 8 << UInt(size); integer g = UInt(Pg); integer n = UInt(Zn); integer d = UInt(Pd); SVECmp op = Cmp_GT;

Greater than or equal

313029282726252423222120191817161514131211109876543210
01100101size010000001PgZn0Pd
eqltne

FCMGE <Pd>.<T>, <Pg>/Z, <Zn>.<T>, #0.0

if !HaveSVE() && !HaveSME() then UNDEFINED; if size == '00' then UNDEFINED; constant integer esize = 8 << UInt(size); integer g = UInt(Pg); integer n = UInt(Zn); integer d = UInt(Pd); SVECmp op = Cmp_GE;

Less than

313029282726252423222120191817161514131211109876543210
01100101size010001001PgZn0Pd
eqltne

FCMLT <Pd>.<T>, <Pg>/Z, <Zn>.<T>, #0.0

if !HaveSVE() && !HaveSME() then UNDEFINED; if size == '00' then UNDEFINED; constant integer esize = 8 << UInt(size); integer g = UInt(Pg); integer n = UInt(Zn); integer d = UInt(Pd); SVECmp op = Cmp_LT;

Less than or equal

313029282726252423222120191817161514131211109876543210
01100101size010001001PgZn1Pd
eqltne

FCMLE <Pd>.<T>, <Pg>/Z, <Zn>.<T>, #0.0

if !HaveSVE() && !HaveSME() then UNDEFINED; if size == '00' then UNDEFINED; constant integer esize = 8 << UInt(size); integer g = UInt(Pg); integer n = UInt(Zn); integer d = UInt(Pd); SVECmp op = Cmp_LE;

Not equal

313029282726252423222120191817161514131211109876543210
01100101size010011001PgZn0Pd
eqltne

FCMNE <Pd>.<T>, <Pg>/Z, <Zn>.<T>, #0.0

if !HaveSVE() && !HaveSME() then UNDEFINED; if size == '00' then UNDEFINED; constant integer esize = 8 << UInt(size); integer g = UInt(Pg); integer n = UInt(Zn); integer d = UInt(Pd); SVECmp op = Cmp_NE;

Assembler Symbols

<Pd>

Is the name of the destination scalable predicate register, encoded in the "Pd" field.

<T>

Is the size specifier, encoded in size:

size <T>
00 RESERVED
01 H
10 S
11 D
<Pg>

Is the name of the governing scalable predicate register P0-P7, encoded in the "Pg" field.

<Zn>

Is the name of the source scalable vector register, encoded in the "Zn" field.

Operation

CheckSVEEnabled(); constant integer VL = CurrentVL; constant integer PL = VL DIV 8; constant integer elements = VL DIV esize; bits(PL) mask = P[g, PL]; bits(VL) operand = if AnyActiveElement(mask, esize) then Z[n, VL] else Zeros(VL); bits(PL) result; constant integer psize = esize DIV 8; for e = 0 to elements-1 if ActivePredicateElement(mask, e, esize) then bits(esize) element = Elem[operand, e, esize]; boolean res; case op of when Cmp_EQ res = FPCompareEQ(element, 0<esize-1:0>, FPCR); when Cmp_GE res = FPCompareGE(element, 0<esize-1:0>, FPCR); when Cmp_GT res = FPCompareGT(element, 0<esize-1:0>, FPCR); when Cmp_NE res = FPCompareNE(element, 0<esize-1:0>, FPCR); when Cmp_LT res = FPCompareGT(0<esize-1:0>, element, FPCR); when Cmp_LE res = FPCompareGE(0<esize-1:0>, element, FPCR); bit pbit = if res then '1' else '0'; Elem[result, e, psize] = ZeroExtend(pbit, psize); else Elem[result, e, psize] = ZeroExtend('0', psize); P[d, PL] = result;

Operational information

If FEAT_SME is implemented and the PE is in Streaming SVE mode, then any subsequent instruction which is dependent on the predicate register written by this instruction might be significantly delayed.


Internal version only: aarchmrs v2023-12_rel, pseudocode v2023-12_rel, sve v2023-12_rel ; Build timestamp: 2023-12-15T16:46

Copyright © 2010-2023 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.