mirror of
https://github.com/pound-emu/ballistic.git
synced 2026-01-31 01:15:21 +01:00
316 lines
17 KiB
XML
316 lines
17 KiB
XML
<?xml version="1.0" encoding="utf-8"?>
|
|
<?xml-stylesheet type="text/xsl" encoding="UTF-8" href="iform.xsl" version="1.0"?>
|
|
<!DOCTYPE instructionsection PUBLIC "-//ARM//DTD instructionsection //EN" "iform-p.dtd">
|
|
<!-- Copyright (c) 2010-2022 Arm Limited or its affiliates. All rights reserved. -->
|
|
<!-- This document is Non-Confidential. This document may only be used and distributed in accordance with the terms of the agreement entered into by Arm and the party that Arm delivered this document to. -->
|
|
|
|
<instructionsection id="sdot_za32_zzw" title="SDOT (2-way, multiple vectors)" type="instruction">
|
|
<docvars>
|
|
<docvar key="instr-class" value="mortlach2" />
|
|
<docvar key="isa" value="A64" />
|
|
<docvar key="mnemonic" value="SDOT" />
|
|
</docvars>
|
|
<heading>SDOT (2-way, multiple vectors)</heading>
|
|
<desc>
|
|
<brief>Multi-vector signed integer dot-product</brief>
|
|
<description>
|
|
<para>The instruction operates on two or four ZA single-vector groups.</para>
|
|
<para>The signed integer dot product instruction computes the dot product of two signed 16-bit integer values held in each 32-bit element of the two or four first source vectors and two signed 16-bit integer values in the corresponding 32-bit element of the two or four second source vectors. The widened dot product result is destructively added to the corresponding 32-bit element of the two or four ZA single-vector groups. The vector numbers forming the single-vector group within each half or each quarter of the ZA array are selected by the sum of the vector select register and immediate offset, modulo half or quarter the number of ZA array vectors.</para>
|
|
<para>The <arm-defined-word>vector group</arm-defined-word> symbol, <syntax>VGx2</syntax> or <syntax>VGx4</syntax>, indicates that the ZA operand consists of two or four ZA single-vector groups respectively. The <arm-defined-word>vector group</arm-defined-word> symbol is preferred for disassembly, but optional in assembler source code.</para>
|
|
<para>This instruction is unpredicated.</para>
|
|
</description>
|
|
<status>Green</status>
|
|
<predicated>False</predicated>
|
|
<sm_policy>SM_1_only</sm_policy>
|
|
</desc>
|
|
<alias_list howmany="0"></alias_list>
|
|
<classes>
|
|
<classesintro count="2">
|
|
<txt>It has encodings from 2 classes:</txt>
|
|
<a href="#iclass_sme_vgx2_single">Two ZA single-vectors</a>
|
|
<txt> and </txt>
|
|
<a href="#iclass_sme_vgx4_single">Four ZA single-vectors</a>
|
|
</classesintro>
|
|
<iclass name="Two ZA single-vectors" oneof="2" id="iclass_sme_vgx2_single" no_encodings="1" isa="A64">
|
|
<docvars>
|
|
<docvar key="instr-class" value="mortlach2" />
|
|
<docvar key="isa" value="A64" />
|
|
<docvar key="mnemonic" value="SDOT" />
|
|
<docvar key="sme-multireg" value="sme-vgx2-single" />
|
|
</docvars>
|
|
<iclassintro count="1"></iclassintro>
|
|
<arch_variants>
|
|
<arch_variant name="FEAT_SME2" feature="FEAT_SME2" />
|
|
</arch_variants>
|
|
<regdiagram form="32" psname="SDOT-ZA32.ZZW-2x2" tworows="1">
|
|
<box hibit="31" width="2" settings="2">
|
|
<c>1</c>
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="29" width="9" settings="9">
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>1</c>
|
|
<c>1</c>
|
|
<c>1</c>
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="20" width="4" name="Zm" usename="1">
|
|
<c colspan="4"></c>
|
|
</box>
|
|
<box hibit="16" width="2" settings="2">
|
|
<c>0</c>
|
|
<c>0</c>
|
|
</box>
|
|
<box hibit="14" width="2" name="Rv" usename="1">
|
|
<c colspan="2"></c>
|
|
</box>
|
|
<box hibit="12" width="3" settings="3">
|
|
<c>1</c>
|
|
<c>0</c>
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="9" width="4" name="Zn" usename="1">
|
|
<c colspan="4"></c>
|
|
</box>
|
|
<box hibit="5" settings="1">
|
|
<c>0</c>
|
|
</box>
|
|
<box hibit="4" name="U" usename="1" settings="1">
|
|
<c>0</c>
|
|
</box>
|
|
<box hibit="3" settings="1">
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="2" width="3" name="off3" usename="1">
|
|
<c colspan="3"></c>
|
|
</box>
|
|
</regdiagram>
|
|
<encoding name="sdot_za32_zzw_2x2" oneofinclass="1" oneof="2" label="">
|
|
<docvars>
|
|
<docvar key="instr-class" value="mortlach2" />
|
|
<docvar key="isa" value="A64" />
|
|
<docvar key="mnemonic" value="SDOT" />
|
|
<docvar key="sme-multireg" value="sme-vgx2-single" />
|
|
</docvars>
|
|
<asmtemplate><text>SDOT ZA.S[</text><a link="sa_wv" hover="32-bit vector select register W8-W11 (field "Rv")"><Wv></a><text>, </text><a link="sa_offs" hover="Vector select offset [0-7] (field "off3")"><offs></a><a>{, VGx2}</a><text>], </text><text>{</text><text> </text><a link="sa_zn1" hover="First scalable vector register of a multi-vector sequence (field Zn)"><Zn1></a><text>.H-</text><a link="sa_zn2" hover="Second scalable vector register of a multi-vector sequence (field Zn)"><Zn2></a><text>.H </text><text>}</text><text>, </text><text>{</text><text> </text><a link="sa_zm1" hover="First scalable vector register of a multi-vector sequence (field Zm)"><Zm1></a><text>.H-</text><a link="sa_zm2" hover="Second scalable vector register of a multi-vector sequence (field Zm)"><Zm2></a><text>.H </text><text>}</text></asmtemplate>
|
|
</encoding>
|
|
<ps_section howmany="1">
|
|
<ps name="SDOT-ZA32.ZZW-2x2" mylink="SDOT-ZA32.ZZW-2x2" enclabels="" sections="1" secttype="noheading">
|
|
<pstext mayhavelinks="1" section="Decode" rep_section="decode">if !<a link="impl-aarch64.HaveSME2.0" file="shared_pseudocode.xml" hover="function: boolean HaveSME2()">HaveSME2</a>() then UNDEFINED;
|
|
integer v = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>('010':Rv);
|
|
constant integer esize = 32;
|
|
integer n = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>(Zn:'0');
|
|
integer m = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>(Zm:'0');
|
|
integer offset = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>(off3);
|
|
constant integer nreg = 2;</pstext>
|
|
</ps>
|
|
</ps_section>
|
|
</iclass>
|
|
<iclass name="Four ZA single-vectors" oneof="2" id="iclass_sme_vgx4_single" no_encodings="1" isa="A64">
|
|
<docvars>
|
|
<docvar key="instr-class" value="mortlach2" />
|
|
<docvar key="isa" value="A64" />
|
|
<docvar key="mnemonic" value="SDOT" />
|
|
<docvar key="sme-multireg" value="sme-vgx4-single" />
|
|
</docvars>
|
|
<iclassintro count="1"></iclassintro>
|
|
<arch_variants>
|
|
<arch_variant name="FEAT_SME2" feature="FEAT_SME2" />
|
|
</arch_variants>
|
|
<regdiagram form="32" psname="SDOT-ZA32.ZZW-4x4" tworows="1">
|
|
<box hibit="31" width="2" settings="2">
|
|
<c>1</c>
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="29" width="9" settings="9">
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>0</c>
|
|
<c>1</c>
|
|
<c>1</c>
|
|
<c>1</c>
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="20" width="3" name="Zm" usename="1">
|
|
<c colspan="3"></c>
|
|
</box>
|
|
<box hibit="17" width="3" settings="3">
|
|
<c>0</c>
|
|
<c>1</c>
|
|
<c>0</c>
|
|
</box>
|
|
<box hibit="14" width="2" name="Rv" usename="1">
|
|
<c colspan="2"></c>
|
|
</box>
|
|
<box hibit="12" width="3" settings="3">
|
|
<c>1</c>
|
|
<c>0</c>
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="9" width="3" name="Zn" usename="1">
|
|
<c colspan="3"></c>
|
|
</box>
|
|
<box hibit="6" width="2" settings="2">
|
|
<c>0</c>
|
|
<c>0</c>
|
|
</box>
|
|
<box hibit="4" name="U" usename="1" settings="1">
|
|
<c>0</c>
|
|
</box>
|
|
<box hibit="3" settings="1">
|
|
<c>1</c>
|
|
</box>
|
|
<box hibit="2" width="3" name="off3" usename="1">
|
|
<c colspan="3"></c>
|
|
</box>
|
|
</regdiagram>
|
|
<encoding name="sdot_za32_zzw_4x4" oneofinclass="1" oneof="2" label="">
|
|
<docvars>
|
|
<docvar key="instr-class" value="mortlach2" />
|
|
<docvar key="isa" value="A64" />
|
|
<docvar key="mnemonic" value="SDOT" />
|
|
<docvar key="sme-multireg" value="sme-vgx4-single" />
|
|
</docvars>
|
|
<asmtemplate><text>SDOT ZA.S[</text><a link="sa_wv" hover="32-bit vector select register W8-W11 (field "Rv")"><Wv></a><text>, </text><a link="sa_offs" hover="Vector select offset [0-7] (field "off3")"><offs></a><a>{, VGx4}</a><text>], </text><text>{</text><text> </text><a link="sa_zn1_1" hover="First scalable vector register of a multi-vector sequence (field Zn)"><Zn1></a><text>.H-</text><a link="sa_zn4" hover="Fourth scalable vector register of a multi-vector sequence (field Zn)"><Zn4></a><text>.H </text><text>}</text><text>, </text><text>{</text><text> </text><a link="sa_zm1_1" hover="First scalable vector register of a multi-vector sequence (field Zm)"><Zm1></a><text>.H-</text><a link="sa_zm4" hover="Fourth scalable vector register of a multi-vector sequence (field Zm)"><Zm4></a><text>.H </text><text>}</text></asmtemplate>
|
|
</encoding>
|
|
<ps_section howmany="1">
|
|
<ps name="SDOT-ZA32.ZZW-4x4" mylink="SDOT-ZA32.ZZW-4x4" enclabels="" sections="1" secttype="noheading">
|
|
<pstext mayhavelinks="1" section="Decode" rep_section="decode">if !<a link="impl-aarch64.HaveSME2.0" file="shared_pseudocode.xml" hover="function: boolean HaveSME2()">HaveSME2</a>() then UNDEFINED;
|
|
integer v = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>('010':Rv);
|
|
constant integer esize = 32;
|
|
integer n = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>(Zn:'00');
|
|
integer m = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>(Zm:'00');
|
|
integer offset = <a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>(off3);
|
|
constant integer nreg = 4;</pstext>
|
|
</ps>
|
|
</ps_section>
|
|
</iclass>
|
|
</classes>
|
|
<explanations scope="all">
|
|
<explanation enclist="sdot_za32_zzw_2x2, sdot_za32_zzw_4x4" symboldefcount="1">
|
|
<symbol link="sa_wv"><Wv></symbol>
|
|
<account encodedin="Rv">
|
|
<intro>
|
|
<para>Is the 32-bit name of the vector select register W8-W11, encoded in the "Rv" field.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_2x2, sdot_za32_zzw_4x4" symboldefcount="1">
|
|
<symbol link="sa_offs"><offs></symbol>
|
|
<account encodedin="off3">
|
|
<intro>
|
|
<para>Is the vector select offset, in the range 0 to 7, encoded in the "off3" field.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_2x2" symboldefcount="1">
|
|
<symbol link="sa_zn1"><Zn1></symbol>
|
|
<account encodedin="Zn">
|
|
<docvars>
|
|
<docvar key="sme-multireg" value="sme-vgx2-single" />
|
|
</docvars>
|
|
<intro>
|
|
<para>For the two ZA single-vectors variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zn" times 2.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_4x4" symboldefcount="2">
|
|
<symbol link="sa_zn1_1"><Zn1></symbol>
|
|
<account encodedin="Zn">
|
|
<docvars>
|
|
<docvar key="sme-multireg" value="sme-vgx4-single" />
|
|
</docvars>
|
|
<intro>
|
|
<para>For the four ZA single-vectors variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zn" times 4.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_4x4" symboldefcount="1">
|
|
<symbol link="sa_zn4"><Zn4></symbol>
|
|
<account encodedin="Zn">
|
|
<intro>
|
|
<para>Is the name of the fourth scalable vector register of a multi-vector sequence, encoded as "Zn" times 4 plus 3.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_2x2" symboldefcount="1">
|
|
<symbol link="sa_zn2"><Zn2></symbol>
|
|
<account encodedin="Zn">
|
|
<intro>
|
|
<para>Is the name of the second scalable vector register of a multi-vector sequence, encoded as "Zn" times 2 plus 1.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_2x2" symboldefcount="1">
|
|
<symbol link="sa_zm1"><Zm1></symbol>
|
|
<account encodedin="Zm">
|
|
<docvars>
|
|
<docvar key="sme-multireg" value="sme-vgx2-single" />
|
|
</docvars>
|
|
<intro>
|
|
<para>For the two ZA single-vectors variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zm" times 2.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_4x4" symboldefcount="2">
|
|
<symbol link="sa_zm1_1"><Zm1></symbol>
|
|
<account encodedin="Zm">
|
|
<docvars>
|
|
<docvar key="sme-multireg" value="sme-vgx4-single" />
|
|
</docvars>
|
|
<intro>
|
|
<para>For the four ZA single-vectors variant: is the name of the first scalable vector register of a multi-vector sequence, encoded as "Zm" times 4.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_4x4" symboldefcount="1">
|
|
<symbol link="sa_zm4"><Zm4></symbol>
|
|
<account encodedin="Zm">
|
|
<intro>
|
|
<para>Is the name of the fourth scalable vector register of a multi-vector sequence, encoded as "Zm" times 4 plus 3.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
<explanation enclist="sdot_za32_zzw_2x2" symboldefcount="1">
|
|
<symbol link="sa_zm2"><Zm2></symbol>
|
|
<account encodedin="Zm">
|
|
<intro>
|
|
<para>Is the name of the second scalable vector register of a multi-vector sequence, encoded as "Zm" times 2 plus 1.</para>
|
|
</intro>
|
|
</account>
|
|
</explanation>
|
|
</explanations>
|
|
<ps_section howmany="1">
|
|
<ps name="SDOT-ZA32.ZZW-2x2" mylink="execute" enclabels="" sections="1" secttype="Operation">
|
|
<pstext mayhavelinks="1" section="Execute" rep_section="execute"><a link="impl-aarch64.CheckStreamingSVEAndZAEnabled.0" file="shared_pseudocode.xml" hover="function: CheckStreamingSVEAndZAEnabled()">CheckStreamingSVEAndZAEnabled</a>();
|
|
constant integer VL = <a link="impl-aarch64.CurrentVL.read.none" file="shared_pseudocode.xml" hover="accessor: integer CurrentVL">CurrentVL</a>;
|
|
constant integer elements = VL DIV esize;
|
|
integer vectors = VL DIV 8;
|
|
integer vstride = vectors DIV nreg;
|
|
bits(32) vbase = <a link="impl-aarch64.X.read.2" file="shared_pseudocode.xml" hover="accessor: bits(width) X[integer n, integer width]">X</a>[v, 32];
|
|
integer vec = (<a link="impl-shared.UInt.1" file="shared_pseudocode.xml" hover="function: integer UInt(bits(N) x)">UInt</a>(vbase) + offset) MOD vstride;
|
|
bits(VL) result;
|
|
|
|
for r = 0 to nreg-1
|
|
bits(VL) operand1 = <a link="impl-aarch64.Z.read.2" file="shared_pseudocode.xml" hover="accessor: bits(width) Z[integer n, integer width]">Z</a>[n+r, VL];
|
|
bits(VL) operand2 = <a link="impl-aarch64.Z.read.2" file="shared_pseudocode.xml" hover="accessor: bits(width) Z[integer n, integer width]">Z</a>[m+r, VL];
|
|
bits(VL) operand3 = <a link="impl-aarch64.ZAvector.read.2" file="shared_pseudocode.xml" hover="accessor: bits(width) ZAvector[integer index, integer width]">ZAvector</a>[vec, VL];
|
|
for e = 0 to elements-1
|
|
bits(esize) sum = <a link="impl-shared.Elem.read.3" file="shared_pseudocode.xml" hover="accessor: bits(size) Elem[bits(N) vector, integer e, integer size]">Elem</a>[operand3, e, esize];
|
|
for i = 0 to 1
|
|
integer element1 = <a link="impl-shared.SInt.1" file="shared_pseudocode.xml" hover="function: integer SInt(bits(N) x)">SInt</a>(<a link="impl-shared.Elem.read.3" file="shared_pseudocode.xml" hover="accessor: bits(size) Elem[bits(N) vector, integer e, integer size]">Elem</a>[operand1, 2 * e + i, esize DIV 2]);
|
|
integer element2 = <a link="impl-shared.SInt.1" file="shared_pseudocode.xml" hover="function: integer SInt(bits(N) x)">SInt</a>(<a link="impl-shared.Elem.read.3" file="shared_pseudocode.xml" hover="accessor: bits(size) Elem[bits(N) vector, integer e, integer size]">Elem</a>[operand2, 2 * e + i, esize DIV 2]);
|
|
sum = sum + element1 * element2;
|
|
<a link="impl-shared.Elem.write.3" file="shared_pseudocode.xml" hover="accessor: Elem[bits(N) &vector, integer e, integer size] = bits(size) value">Elem</a>[result, e, esize] = sum;
|
|
<a link="impl-aarch64.ZAvector.write.2" file="shared_pseudocode.xml" hover="accessor: ZAvector[integer index, integer width] = bits(width) value">ZAvector</a>[vec, VL] = result;
|
|
vec = vec + vstride;</pstext>
|
|
</ps>
|
|
</ps_section>
|
|
</instructionsection>
|