LUTI4 (two registers)

Lookup table read with 4-bit indexes

Copy 8-bit, 16-bit or 32-bit elements from ZT0 to two destination vectors using packed 4-bit indices from a segment of the source vector register. A segment corresponds to a portion of the source vector that is consumed in order to fill the destination vector. The segment is selected by the vector segment index modulo the total number of segments.

This instruction is unpredicated.

It has encodings from 2 classes: Consecutive and Strided

Consecutive
(FEAT_SME2)

313029282726252423222120191817161514131211109876543210
110000001000101i21size00ZnZd0

LUTI4 { <Zd1>.<T>-<Zd2>.<T> }, ZT0, <Zn>[<index>]

if !IsFeatureImplemented(FEAT_SME2) then UNDEFINED; if size == '11' then UNDEFINED; constant integer esize = 8 << UInt(size); constant integer isize = 4; constant integer n = UInt(Zn); constant integer dstride = 1; constant integer d = UInt(Zd:'0'); constant integer imm = UInt(i2); constant integer nreg = 2;

Strided
(FEAT_SME2p1)

313029282726252423222120191817161514131211109876543210
110000001001101i21size00ZnD0Zd

LUTI4 { <Zd1>.<T>, <Zd2>.<T> }, ZT0, <Zn>[<index>]

if !IsFeatureImplemented(FEAT_SME2p1) then UNDEFINED; if size == '10' || size == '11' then UNDEFINED; constant integer esize = 8 << UInt(size); constant integer isize = 4; constant integer n = UInt(Zn); constant integer dstride = 8; constant integer d = UInt(D:'0':Zd); constant integer imm = UInt(i2); constant integer nreg = 2;

Assembler Symbols

<Zd1>

For the consecutive variant: is the name of the first scalable vector register of the destination multi-vector group, encoded as "Zd" times 2.

For the strided variant: is the name of the first scalable vector register Z0-Z7 or Z16-Z23 of the destination multi-vector group, encoded as "D:'0':Zd".

<T>

For the consecutive variant: is the size specifier, encoded in size:

size <T>
00 B
01 H
10 S
11 RESERVED

For the strided variant: is the size specifier, encoded in size<0>:

size<0> <T>
0 B
1 H
<Zd2>

For the consecutive variant: is the name of the second scalable vector register of the destination multi-vector group, encoded as "Zd" times 2 plus 1.

For the strided variant: is the name of the second scalable vector register Z8-Z15 or Z24-Z31 of the destination multi-vector group, encoded as "D:'1':Zd".

<Zn>

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

<index>

Is the vector segment index, in the range 0 to 3, encoded in the "i2" field.

Operation

CheckStreamingSVEEnabled(); CheckSMEZT0Enabled(); constant integer VL = CurrentVL; constant integer elements = VL DIV esize; constant integer segments = esize DIV (isize * nreg); constant integer segment = imm MOD segments; constant bits(VL) indexes = Z[n, VL]; integer dst = d; constant bits(512) table = ZT0[512]; for r = 0 to nreg-1 constant integer base = (segment * nreg + r) * elements; bits(VL) result; for e = 0 to elements-1 constant integer index = UInt(Elem[indexes, base+e, isize]); Elem[result, e, esize] = Elem[table, index, 32]<esize-1:0>; Z[dst, VL] = result; dst = dst + dstride;

Operational information

If PSTATE.DIT is 1:


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.