ARM-software
/
CMSIS_5
зеркало из https://github-proxy.rt-thread.io/ARM-software/CMSIS_5.git


			
				
					
						
						
							12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361136213631364136513661367136813691370137113721373137413751376137713781379138013811382138313841385138613871388138913901391139213931394139513961397139813991400140114021403140414051406140714081409141014111412141314141415141614171418141914201421142214231424142514261427142814291430143114321433143414351436143714381439144014411442144314441445144614471448144914501451145214531454145514561457145814591460146114621463146414651466146714681469147014711472147314741475147614771478147914801481148214831484148514861487148814891490149114921493149414951496149714981499150015011502150315041505150615071508150915101511151215131514151515161517151815191520152115221523152415251526152715281529153015311532153315341535153615371538153915401541154215431544154515461547154815491550155115521553155415551556155715581559156015611562156315641565156615671568156915701571157215731574157515761577157815791580158115821583158415851586158715881589159015911592159315941595159615971598159916001601160216031604160516061607160816091610161116121613161416151616161716181619162016211622162316241625162616271628162916301631163216331634163516361637163816391640164116421643164416451646164716481649165016511652165316541655165616571658165916601661166216631664166516661667166816691670167116721673167416751676167716781679168016811682168316841685168616871688168916901691169216931694169516961697169816991700170117021703170417051706170717081709171017111712171317141715171617171718171917201721172217231724172517261727172817291730173117321733173417351736173717381739174017411742174317441745174617471748174917501751175217531754175517561757175817591760176117621763176417651766176717681769177017711772177317741775177617771778177917801781178217831784178517861787178817891790179117921793179417951796179717981799180018011802180318041805180618071808180918101811181218131814181518161817181818191820182118221823182418251826182718281829183018311832183318341835183618371838183918401841184218431844184518461847184818491850185118521853185418551856185718581859186018611862186318641865186618671868186918701871187218731874187518761877187818791880188118821883188418851886188718881889189018911892189318941895189618971898189919001901190219031904190519061907190819091910191119121913191419151916191719181919192019211922192319241925192619271928192919301931193219331934193519361937193819391940194119421943194419451946194719481949195019511952195319541955195619571958195919601961196219631964196519661967196819691970197119721973197419751976197719781979198019811982198319841985198619871988198919901991199219931994199519961997199819992000200120022003200420052006200720082009201020112012201320142015201620172018201920202021202220232024202520262027202820292030203120322033203420352036203720382039204020412042204320442045204620472048204920502051205220532054205520562057205820592060206120622063206420652066206720682069207020712072207320742075207620772078207920802081208220832084208520862087208820892090209120922093209420952096209720982099210021012102210321042105210621072108210921102111211221132114211521162117211821192120212121222123212421252126212721282129213021312132213321342135213621372138213921402141214221432144214521462147214821492150215121522153215421552156215721582159216021612162216321642165216621672168216921702171217221732174217521762177217821792180218121822183218421852186218721882189219021912192219321942195219621972198219922002201220222032204220522062207220822092210221122122213221422152216221722182219222022212222222322242225222622272228222922302231223222332234223522362237223822392240224122422243224422452246224722482249225022512252225322542255225622572258225922602261226222632264226522662267226822692270227122722273227422752276227722782279228022812282228322842285228622872288228922902291229222932294229522962297229822992300230123022303230423052306230723082309231023112312231323142315231623172318231923202321232223232324232523262327232823292330233123322333233423352336233723382339234023412342234323442345234623472348234923502351235223532354235523562357235823592360236123622363236423652366236723682369237023712372237323742375237623772378237923802381238223832384238523862387238823892390239123922393239423952396239723982399240024012402240324042405240624072408240924102411241224132414241524162417241824192420242124222423242424252426242724282429243024312432243324342435243624372438243924402441244224432444244524462447244824492450245124522453245424552456245724582459246024612462246324642465246624672468246924702471247224732474247524762477247824792480248124822483248424852486248724882489249024912492249324942495249624972498249925002501250225032504250525062507250825092510251125122513251425152516251725182519252025212522252325242525252625272528252925302531253225332534253525362537253825392540254125422543254425452546254725482549255025512552255325542555255625572558255925602561256225632564256525662567256825692570257125722573257425752576257725782579258025812582258325842585258625872588258925902591259225932594259525962597259825992600260126022603260426052606260726082609261026112612261326142615261626172618261926202621262226232624262526262627262826292630263126322633263426352636263726382639264026412642264326442645264626472648264926502651265226532654265526562657265826592660266126622663266426652666266726682669267026712672267326742675267626772678267926802681268226832684268526862687268826892690269126922693269426952696269726982699270027012702270327042705270627072708270927102711271227132714271527162717271827192720272127222723272427252726272727282729273027312732273327342735273627372738273927402741274227432744274527462747274827492750275127522753275427552756275727582759276027612762276327642765276627672768276927702771277227732774277527762777277827792780278127822783278427852786278727882789279027912792279327942795279627972798279928002801280228032804280528062807280828092810281128122813281428152816281728182819282028212822282328242825282628272828282928302831283228332834283528362837283828392840284128422843284428452846284728482849285028512852285328542855285628572858285928602861286228632864286528662867286828692870287128722873287428752876287728782879288028812882288328842885288628872888288928902891289228932894289528962897289828992900290129022903290429052906290729082909291029112912291329142915291629172918291929202921292229232924292529262927292829292930293129322933293429352936293729382939294029412942294329442945294629472948294929502951295229532954295529562957295829592960296129622963296429652966296729682969297029712972297329742975297629772978297929802981298229832984298529862987298829892990299129922993299429952996299729982999300030013002300330043005300630073008300930103011301230133014301530163017301830193020302130223023302430253026302730283029303030313032303330343035303630373038303930403041304230433044304530463047304830493050305130523053305430553056305730583059306030613062306330643065306630673068306930703071307230733074307530763077307830793080308130823083308430853086308730883089309030913092309330943095309630973098309931003101310231033104310531063107310831093110311131123113311431153116311731183119312031213122312331243125312631273128312931303131313231333134313531363137313831393140314131423143314431453146314731483149315031513152315331543155315631573158315931603161316231633164316531663167316831693170317131723173317431753176317731783179318031813182318331843185318631873188318931903191319231933194319531963197319831993200320132023203320432053206320732083209321032113212321332143215321632173218321932203221322232233224322532263227322832293230323132323233323432353236323732383239324032413242324332443245324632473248324932503251325232533254325532563257325832593260326132623263326432653266326732683269327032713272327332743275327632773278327932803281328232833284328532863287328832893290329132923293329432953296329732983299330033013302330333043305330633073308330933103311331233133314331533163317331833193320332133223323332433253326332733283329333033313332333333343335333633373338333933403341334233433344334533463347334833493350335133523353335433553356335733583359336033613362336333643365336633673368336933703371337233733374337533763377337833793380338133823383338433853386338733883389339033913392339333943395339633973398339934003401340234033404340534063407340834093410341134123413341434153416341734183419342034213422342334243425342634273428342934303431343234333434343534363437343834393440344134423443344434453446344734483449345034513452345334543455345634573458345934603461346234633464346534663467346834693470347134723473347434753476347734783479348034813482348334843485348634873488348934903491349234933494349534963497349834993500350135023503350435053506350735083509351035113512351335143515351635173518351935203521352235233524352535263527352835293530353135323533353435353536353735383539354035413542354335443545354635473548354935503551355235533554355535563557355835593560356135623563356435653566356735683569357035713572357335743575357635773578357935803581358235833584358535863587358835893590359135923593359435953596359735983599360036013602360336043605360636073608360936103611361236133614361536163617361836193620362136223623362436253626362736283629363036313632363336343635363636373638363936403641364236433644364536463647364836493650365136523653365436553656365736583659
							<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
<meta http-equiv="X-UA-Compatible" content="IE=9"/>
<title>Convolution Functions</title>
<title>CMSIS-NN: Convolution Functions</title>
<link href="tabs.css" rel="stylesheet" type="text/css"/>
<link href="cmsis.css" rel="stylesheet" type="text/css" />
<script type="text/javascript" src="jquery.js"></script>
<script type="text/javascript" src="dynsections.js"></script>
<script type="text/javascript" src="printComponentTabs.js"></script>
<link href="navtree.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript" src="resize.js"></script>
<script type="text/javascript" src="navtree.js"></script>
<script type="text/javascript">
  $(document).ready(initResizable);
  $(window).load(resizeHeight);
</script>
<link href="search/search.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript" src="search/search.js"></script>
<script type="text/javascript">
  $(document).ready(function() { searchBox.OnSelectItem(0); });
</script>
</head>
<body>
<div id="top"><!-- do not remove this div, it is closed by doxygen! -->
<div id="titlearea">
<table cellspacing="0" cellpadding="0">
 <tbody>
 <tr style="height: 46px;">
  <td id="projectlogo"><img alt="Logo" src="CMSIS_Logo_Final.png"/></td>
  <td style="padding-left: 0.5em;">
   <div id="projectname">CMSIS-NN
   &#160;<span id="projectnumber">Version 1.3.0</span>
   </div>
   <div id="projectbrief">CMSIS NN Software Library</div>
  </td>
 </tr>
 </tbody>
</table>
</div>
<!-- end header part -->
<div id="CMSISnav" class="tabs1">
    <ul class="tablist">
      <script type="text/javascript">
		<!--
		writeComponentTabs.call(this);
		//-->
      </script>
	  </ul>
</div>
<!-- Generated by Doxygen 1.8.6 -->
<script type="text/javascript">
var searchBox = new SearchBox("searchBox", "search",false,'Search');
</script>
  <div id="navrow1" class="tabs">
    <ul class="tablist">
      <li><a href="index.html"><span>Main&#160;Page</span></a></li>
      <li><a href="pages.html"><span>Usage&#160;and&#160;Description</span></a></li>
      <li><a href="modules.html"><span>Reference</span></a></li>
      <li>
        <div id="MSearchBox" class="MSearchBoxInactive">
        <span class="left">
          <img id="MSearchSelect" src="search/mag_sel.png"
               onmouseover="return searchBox.OnSearchSelectShow()"
               onmouseout="return searchBox.OnSearchSelectHide()"
               alt=""/>
          <input type="text" id="MSearchField" value="Search" accesskey="S"
               onfocus="searchBox.OnSearchFieldFocus(true)" 
               onblur="searchBox.OnSearchFieldFocus(false)" 
               onkeyup="searchBox.OnSearchFieldChange(event)"/>
          </span><span class="right">
            <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
          </span>
        </div>
      </li>
    </ul>
  </div>
</div><!-- top -->
<div id="side-nav" class="ui-resizable side-nav-resizable">
  <div id="nav-tree">
    <div id="nav-tree-contents">
      <div id="nav-sync" class="sync"></div>
    </div>
  </div>
  <div id="splitbar" style="-moz-user-select:none;" 
       class="ui-resizable-handle">
  </div>
</div>
<script type="text/javascript">
$(document).ready(function(){initNavTree('group__NNConv.html','');});
</script>
<div id="doc-content">
<!-- window showing the filter options -->
<div id="MSearchSelectWindow"
     onmouseover="return searchBox.OnSearchSelectShow()"
     onmouseout="return searchBox.OnSearchSelectHide()"
     onkeydown="return searchBox.OnSearchSelectKey(event)">
<a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(0)"><span class="SelectionMark">&#160;</span>All</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(1)"><span class="SelectionMark">&#160;</span>Data Structures</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(2)"><span class="SelectionMark">&#160;</span>Namespaces</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(3)"><span class="SelectionMark">&#160;</span>Files</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(4)"><span class="SelectionMark">&#160;</span>Functions</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(5)"><span class="SelectionMark">&#160;</span>Variables</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(6)"><span class="SelectionMark">&#160;</span>Enumerations</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(7)"><span class="SelectionMark">&#160;</span>Enumerator</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(8)"><span class="SelectionMark">&#160;</span>Macros</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(9)"><span class="SelectionMark">&#160;</span>Groups</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(10)"><span class="SelectionMark">&#160;</span>Pages</a></div>

<!-- iframe showing the search results (closed by default) -->
<div id="MSearchResultsWindow">
<iframe src="javascript:void(0)" frameborder="0" 
        name="MSearchResults" id="MSearchResults">
</iframe>
</div>

<div class="header">
  <div class="summary">
<a href="#func-members">Functions</a>  </div>
  <div class="headertitle">
<div class="title">Convolution Functions<div class="ingroups"><a class="el" href="group__groupNN.html">Neural Network Functions</a></div></div>  </div>
</div><!--header-->
<div class="contents">
<table class="memberdecls">
<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="func-members"></a>
Functions</h2></td></tr>
<tr class="memitem:ga2d9c129189832572cf0bc97ff9e73e54"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga2d9c129189832572cf0bc97ff9e73e54">arm_convolve_1_x_n_s8</a> (const q7_t *input, const uint16_t input_x, const uint16_t input_ch, const uint16_t input_batches, const q7_t *kernel, const uint16_t output_ch, const uint16_t kernel_x, const uint16_t pad_x, const uint16_t stride_x, const int32_t *bias, q7_t *output, const int32_t *output_shift, const int32_t *output_mult, const int32_t out_offset, const int32_t input_offset, const int32_t out_activation_min, const int32_t out_activation_max, const uint16_t output_x, q15_t *buffer_a)</td></tr>
<tr class="memdesc:ga2d9c129189832572cf0bc97ff9e73e54"><td class="mdescLeft">&#160;</td><td class="mdescRight">1xn convolution  <a href="#ga2d9c129189832572cf0bc97ff9e73e54">More...</a><br/></td></tr>
<tr class="separator:ga2d9c129189832572cf0bc97ff9e73e54"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gab34cb94c592df768774d8bcac03d7e5c"><td class="memItemLeft" align="right" valign="top">int32_t&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gab34cb94c592df768774d8bcac03d7e5c">arm_convolve_1_x_n_s8_get_buffer_size</a> (const uint16_t input_ch, const uint16_t kernel_x, const uint16_t kernel_y)</td></tr>
<tr class="memdesc:gab34cb94c592df768774d8bcac03d7e5c"><td class="mdescLeft">&#160;</td><td class="mdescRight">Get the required additional buffer size for 1xn convolution.  <a href="#gab34cb94c592df768774d8bcac03d7e5c">More...</a><br/></td></tr>
<tr class="separator:gab34cb94c592df768774d8bcac03d7e5c"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga110adcfdaab356c750c6270aa5e05f29"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga110adcfdaab356c750c6270aa5e05f29">arm_convolve_1x1_HWC_q7_fast_nonsquare</a> (const q7_t *Im_in, const uint16_t dim_im_in_x, const uint16_t dim_im_in_y, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel_x, const uint16_t dim_kernel_y, const uint16_t padding_x, const uint16_t padding_y, const uint16_t stride_x, const uint16_t stride_y, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out_x, const uint16_t dim_im_out_y, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga110adcfdaab356c750c6270aa5e05f29"><td class="mdescLeft">&#160;</td><td class="mdescRight">Fast Q7 version of 1x1 convolution (non-sqaure shape)  <a href="#ga110adcfdaab356c750c6270aa5e05f29">More...</a><br/></td></tr>
<tr class="separator:ga110adcfdaab356c750c6270aa5e05f29"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gadfcad0bae34328dc806dc88cde173faa"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gadfcad0bae34328dc806dc88cde173faa">arm_convolve_1x1_s8_fast</a> (const q7_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_ch, const uint16_t input_batches, const q7_t *kernel, const uint16_t output_ch, const uint16_t pad_x, const uint16_t pad_y, const uint16_t stride_x, const uint16_t stride_y, const int32_t *bias, q7_t *output, const int32_t *output_shift, const int32_t *output_mult, const int32_t out_offset, const int32_t input_offset, const int32_t out_activation_min, const int32_t out_activation_max, const uint16_t output_x, const uint16_t output_y, q15_t *buffer_a)</td></tr>
<tr class="memdesc:gadfcad0bae34328dc806dc88cde173faa"><td class="mdescLeft">&#160;</td><td class="mdescRight">Fast s8 version for 1x1 convolution (non-square shape)  <a href="#gadfcad0bae34328dc806dc88cde173faa">More...</a><br/></td></tr>
<tr class="separator:gadfcad0bae34328dc806dc88cde173faa"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga09ed5df0234c7eb334a7264dd9e4c833"><td class="memItemLeft" align="right" valign="top">int32_t&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga09ed5df0234c7eb334a7264dd9e4c833">arm_convolve_1x1_s8_fast_get_buffer_size</a> (const uint16_t input_ch)</td></tr>
<tr class="memdesc:ga09ed5df0234c7eb334a7264dd9e4c833"><td class="mdescLeft">&#160;</td><td class="mdescRight">Get the required buffer size for the fast 1x1 convolution (non-square shape) s8 convolution function.  <a href="#ga09ed5df0234c7eb334a7264dd9e4c833">More...</a><br/></td></tr>
<tr class="separator:ga09ed5df0234c7eb334a7264dd9e4c833"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga55701f213b198084b52eab53097f1f58"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga55701f213b198084b52eab53097f1f58">arm_convolve_HWC_q15_basic</a> (const q15_t *Im_in, const uint16_t dim_im_in, const uint16_t ch_im_in, const q15_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel, const uint16_t padding, const uint16_t stride, const q15_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q15_t *Im_out, const uint16_t dim_im_out, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga55701f213b198084b52eab53097f1f58"><td class="mdescLeft">&#160;</td><td class="mdescRight">Basic Q15 convolution function.  <a href="#ga55701f213b198084b52eab53097f1f58">More...</a><br/></td></tr>
<tr class="separator:ga55701f213b198084b52eab53097f1f58"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga4efb1ccbbaa7dd936961989dcb443f50"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga4efb1ccbbaa7dd936961989dcb443f50">arm_convolve_HWC_q15_fast</a> (const q15_t *Im_in, const uint16_t dim_im_in, const uint16_t ch_im_in, const q15_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel, const uint16_t padding, const uint16_t stride, const q15_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q15_t *Im_out, const uint16_t dim_im_out, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga4efb1ccbbaa7dd936961989dcb443f50"><td class="mdescLeft">&#160;</td><td class="mdescRight">Fast Q15 convolution function.  <a href="#ga4efb1ccbbaa7dd936961989dcb443f50">More...</a><br/></td></tr>
<tr class="separator:ga4efb1ccbbaa7dd936961989dcb443f50"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga614ec3b71eb96e29952ec3f09e7b9c3c"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga614ec3b71eb96e29952ec3f09e7b9c3c">arm_convolve_HWC_q15_fast_nonsquare</a> (const q15_t *Im_in, const uint16_t dim_im_in_x, const uint16_t dim_im_in_y, const uint16_t ch_im_in, const q15_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel_x, const uint16_t dim_kernel_y, const uint16_t padding_x, const uint16_t padding_y, const uint16_t stride_x, const uint16_t stride_y, const q15_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q15_t *Im_out, const uint16_t dim_im_out_x, const uint16_t dim_im_out_y, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga614ec3b71eb96e29952ec3f09e7b9c3c"><td class="mdescLeft">&#160;</td><td class="mdescRight">Fast Q15 convolution function (non-sqaure shape)  <a href="#ga614ec3b71eb96e29952ec3f09e7b9c3c">More...</a><br/></td></tr>
<tr class="separator:ga614ec3b71eb96e29952ec3f09e7b9c3c"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga210ae8d8fc1d12ee15b41f1fa6947681"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga210ae8d8fc1d12ee15b41f1fa6947681">arm_convolve_HWC_q7_basic</a> (const q7_t *Im_in, const uint16_t dim_im_in, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel, const uint16_t padding, const uint16_t stride, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga210ae8d8fc1d12ee15b41f1fa6947681"><td class="mdescLeft">&#160;</td><td class="mdescRight">Basic Q7 convolution function.  <a href="#ga210ae8d8fc1d12ee15b41f1fa6947681">More...</a><br/></td></tr>
<tr class="separator:ga210ae8d8fc1d12ee15b41f1fa6947681"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga4501fa22c0836002aa47ccc313dce252"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga4501fa22c0836002aa47ccc313dce252">arm_convolve_HWC_q7_basic_nonsquare</a> (const q7_t *Im_in, const uint16_t dim_im_in_x, const uint16_t dim_im_in_y, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel_x, const uint16_t dim_kernel_y, const uint16_t padding_x, const uint16_t padding_y, const uint16_t stride_x, const uint16_t stride_y, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out_x, const uint16_t dim_im_out_y, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga4501fa22c0836002aa47ccc313dce252"><td class="mdescLeft">&#160;</td><td class="mdescRight">Basic Q7 convolution function (non-sqaure shape)  <a href="#ga4501fa22c0836002aa47ccc313dce252">More...</a><br/></td></tr>
<tr class="separator:ga4501fa22c0836002aa47ccc313dce252"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gae00d3c1285907d59657369fc98bcc83f"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gae00d3c1285907d59657369fc98bcc83f">arm_convolve_HWC_q7_fast</a> (const q7_t *Im_in, const uint16_t dim_im_in, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel, const uint16_t padding, const uint16_t stride, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:gae00d3c1285907d59657369fc98bcc83f"><td class="mdescLeft">&#160;</td><td class="mdescRight">Fast Q7 convolution function.  <a href="#gae00d3c1285907d59657369fc98bcc83f">More...</a><br/></td></tr>
<tr class="separator:gae00d3c1285907d59657369fc98bcc83f"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gabc6d6b991024e9e5c5cdbd7489de88ef"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gabc6d6b991024e9e5c5cdbd7489de88ef">arm_convolve_HWC_q7_fast_nonsquare</a> (const q7_t *Im_in, const uint16_t dim_im_in_x, const uint16_t dim_im_in_y, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel_x, const uint16_t dim_kernel_y, const uint16_t padding_x, const uint16_t padding_y, const uint16_t stride_x, const uint16_t stride_y, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out_x, const uint16_t dim_im_out_y, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:gabc6d6b991024e9e5c5cdbd7489de88ef"><td class="mdescLeft">&#160;</td><td class="mdescRight">Fast Q7 convolution function (non-sqaure shape)  <a href="#gabc6d6b991024e9e5c5cdbd7489de88ef">More...</a><br/></td></tr>
<tr class="separator:gabc6d6b991024e9e5c5cdbd7489de88ef"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga98f2ead67d7cbdf558b0cd8a3b8fc148"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga98f2ead67d7cbdf558b0cd8a3b8fc148">arm_convolve_HWC_q7_RGB</a> (const q7_t *Im_in, const uint16_t dim_im_in, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel, const uint16_t padding, const uint16_t stride, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga98f2ead67d7cbdf558b0cd8a3b8fc148"><td class="mdescLeft">&#160;</td><td class="mdescRight">Q7 convolution function for RGB image.  <a href="#ga98f2ead67d7cbdf558b0cd8a3b8fc148">More...</a><br/></td></tr>
<tr class="separator:ga98f2ead67d7cbdf558b0cd8a3b8fc148"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga6a0b2161ece62078228ead2d85bbcf28"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga6a0b2161ece62078228ead2d85bbcf28">arm_convolve_s8</a> (const q7_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_ch, const uint16_t input_batches, const q7_t *kernel, const uint16_t output_ch, const uint16_t kernel_x, const uint16_t kernel_y, const uint16_t pad_x, const uint16_t pad_y, const uint16_t stride_x, const uint16_t stride_y, const int32_t *bias, q7_t *output, const int32_t *output_shift, const int32_t *output_mult, const int32_t out_offset, const int32_t input_offset, const int32_t out_activation_min, const int32_t out_activation_max, const uint16_t output_x, const uint16_t output_y, q15_t *buffer_a)</td></tr>
<tr class="memdesc:ga6a0b2161ece62078228ead2d85bbcf28"><td class="mdescLeft">&#160;</td><td class="mdescRight">Basic s8 convolution function.  <a href="#ga6a0b2161ece62078228ead2d85bbcf28">More...</a><br/></td></tr>
<tr class="separator:ga6a0b2161ece62078228ead2d85bbcf28"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gaaf83450fd2d89432d9759df99b05737b"><td class="memItemLeft" align="right" valign="top">int32_t&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gaaf83450fd2d89432d9759df99b05737b">arm_convolve_s8_get_buffer_size</a> (const uint16_t input_ch, const uint16_t kernel_x, const uint16_t kernel_y)</td></tr>
<tr class="memdesc:gaaf83450fd2d89432d9759df99b05737b"><td class="mdescLeft">&#160;</td><td class="mdescRight">Get the required buffer size for s8 convolution function.  <a href="#gaaf83450fd2d89432d9759df99b05737b">More...</a><br/></td></tr>
<tr class="separator:gaaf83450fd2d89432d9759df99b05737b"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gae86b03224e6edf28f54247594b6c8bf9"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gae86b03224e6edf28f54247594b6c8bf9">arm_depthwise_conv_3x3_s8</a> (const int8_t *input, const int32_t input_x, const int32_t input_y, const int32_t input_ch, const int8_t *kernel, const int32_t output_ch, const int32_t pad_x, const int32_t pad_y, const int32_t stride_x, const int32_t stride_y, const int32_t *bias, int8_t *output, const int32_t *output_shift, const int32_t *output_mult, const int32_t output_x, const int32_t output_y, const int32_t output_offset, const int32_t input_offset, const int32_t output_activation_min, const int32_t output_activation_max, const int32_t dilation_x, const int32_t dilation_y, int16_t *buffer_a)</td></tr>
<tr class="memdesc:gae86b03224e6edf28f54247594b6c8bf9"><td class="mdescLeft">&#160;</td><td class="mdescRight">Optimized s8 depthwise convolution function for 3x3 kernel size with constraint that in_channel equals out_channel.  <a href="#gae86b03224e6edf28f54247594b6c8bf9">More...</a><br/></td></tr>
<tr class="separator:gae86b03224e6edf28f54247594b6c8bf9"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga06d7783f6788faad5165fd0ae583bc59"><td class="memItemLeft" align="right" valign="top">static void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga06d7783f6788faad5165fd0ae583bc59">depthwise_conv_s8_mult_4</a> (const int8_t *input, const int32_t input_x, const int32_t input_y, const int32_t input_ch, const int8_t *kernel, const int32_t output_ch, const int32_t ch_mult, const int32_t kernel_x, const int32_t kernel_y, const int32_t pad_x, const int32_t pad_y, const int32_t stride_x, const int32_t stride_y, const int32_t *bias, int8_t *output, const int32_t *output_shift, const int32_t *output_mult, const int32_t output_x, const int32_t output_y, const int32_t output_offset, const int32_t input_offset, const int32_t output_activation_min, const int32_t output_activation_max)</td></tr>
<tr class="separator:ga06d7783f6788faad5165fd0ae583bc59"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga00ec3089e06c8953b4e12582583261c9"><td class="memItemLeft" align="right" valign="top">static void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga00ec3089e06c8953b4e12582583261c9">depthwise_conv_s8_generic</a> (const q7_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_ch, const q7_t *kernel, const uint16_t output_ch, const uint16_t ch_mult, const uint16_t kernel_x, const uint16_t kernel_y, const uint16_t pad_x, const uint16_t pad_y, const uint16_t stride_x, const uint16_t stride_y, const int32_t *bias, q7_t *output, const int32_t *output_shift, const int32_t *output_mult, const uint16_t output_x, const uint16_t output_y, const int32_t output_offset, const int32_t input_offset, const int32_t output_activation_min, const int32_t output_activation_max)</td></tr>
<tr class="separator:ga00ec3089e06c8953b4e12582583261c9"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gae7e0c94c970bfd59d9c5e2331ce84fe7"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gae7e0c94c970bfd59d9c5e2331ce84fe7">arm_depthwise_conv_s8</a> (const q7_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_ch, const q7_t *kernel, const uint16_t output_ch, const uint16_t ch_mult, const uint16_t kernel_x, const uint16_t kernel_y, const uint16_t pad_x, const uint16_t pad_y, const uint16_t stride_x, const uint16_t stride_y, const int32_t *bias, q7_t *output, const int32_t *output_shift, const int32_t *output_mult, const uint16_t output_x, const uint16_t output_y, const int32_t output_offset, const int32_t input_offset, const int32_t output_activation_min, const int32_t output_activation_max, const uint16_t dilation_x, const uint16_t dilation_y, q15_t *buffer_a)</td></tr>
<tr class="memdesc:gae7e0c94c970bfd59d9c5e2331ce84fe7"><td class="mdescLeft">&#160;</td><td class="mdescRight">Basic s8 depthwise convolution function.  <a href="#gae7e0c94c970bfd59d9c5e2331ce84fe7">More...</a><br/></td></tr>
<tr class="separator:gae7e0c94c970bfd59d9c5e2331ce84fe7"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga3acc8b0cd5bd5d975fdf5a615261f0b1"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga3acc8b0cd5bd5d975fdf5a615261f0b1">arm_depthwise_conv_s8_opt</a> (const q7_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_ch, const q7_t *kernel, const uint16_t output_ch, const uint16_t kernel_x, const uint16_t kernel_y, const uint16_t pad_x, const uint16_t pad_y, const uint16_t stride_x, const uint16_t stride_y, const int32_t *bias, q7_t *output, const int32_t *output_shift, const int32_t *output_mult, const uint16_t output_x, const uint16_t output_y, const int32_t output_offset, const int32_t input_offset, const int32_t output_activation_min, const int32_t output_activation_max, const uint16_t dilation_x, const uint16_t dilation_y, q15_t *buffer_a)</td></tr>
<tr class="memdesc:ga3acc8b0cd5bd5d975fdf5a615261f0b1"><td class="mdescLeft">&#160;</td><td class="mdescRight">Optimized s8 depthwise convolution function with constraint that in_channel equals out_channel.  <a href="#ga3acc8b0cd5bd5d975fdf5a615261f0b1">More...</a><br/></td></tr>
<tr class="separator:ga3acc8b0cd5bd5d975fdf5a615261f0b1"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gaa53d31632c765b9b4bbd490d0ae3c0c1"><td class="memItemLeft" align="right" valign="top">int32_t&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gaa53d31632c765b9b4bbd490d0ae3c0c1">arm_depthwise_conv_s8_opt_get_buffer_size</a> (const uint16_t input_ch, const uint16_t kernel_x, const uint16_t kernel_y)</td></tr>
<tr class="memdesc:gaa53d31632c765b9b4bbd490d0ae3c0c1"><td class="mdescLeft">&#160;</td><td class="mdescRight">Get the required buffer size for optimized s8 depthwise convolution function with constraint that in_channel equals out_channel.  <a href="#gaa53d31632c765b9b4bbd490d0ae3c0c1">More...</a><br/></td></tr>
<tr class="separator:gaa53d31632c765b9b4bbd490d0ae3c0c1"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga02a296cb4f6361a70c3ecf1ef1238292"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga02a296cb4f6361a70c3ecf1ef1238292">arm_depthwise_conv_u8_basic_ver1</a> (const uint8_t *input, const uint16_t input_x, const uint16_t input_y, const uint16_t input_ch, const uint8_t *kernel, const uint16_t kernel_x, const uint16_t kernel_y, const int16_t ch_mult, const int16_t pad_x, const int16_t pad_y, const int16_t stride_x, const int16_t stride_y, const int16_t dilation_x, const int16_t dilation_y, const int32_t *bias, const int32_t input_offset, const int32_t filter_offset, const int32_t output_offset, uint8_t *output, const uint16_t output_x, const uint16_t output_y, const int32_t output_activation_min, const int32_t output_activation_max, const int32_t out_shift, const int32_t out_mult)</td></tr>
<tr class="memdesc:ga02a296cb4f6361a70c3ecf1ef1238292"><td class="mdescLeft">&#160;</td><td class="mdescRight">uint8 depthwise convolution function with asymmetric quantization for even number of channel multiplier and input channels. Unless specified otherwise, arguments are mandatory. Both square and non-square inputs are accepted.  <a href="#ga02a296cb4f6361a70c3ecf1ef1238292">More...</a><br/></td></tr>
<tr class="separator:ga02a296cb4f6361a70c3ecf1ef1238292"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:gad3d21b3bc6dbd6f3b97d01104349cb0a"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#gad3d21b3bc6dbd6f3b97d01104349cb0a">arm_depthwise_separable_conv_HWC_q7</a> (const q7_t *Im_in, const uint16_t dim_im_in, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel, const uint16_t padding, const uint16_t stride, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:gad3d21b3bc6dbd6f3b97d01104349cb0a"><td class="mdescLeft">&#160;</td><td class="mdescRight">Q7 depthwise separable convolution function.  <a href="#gad3d21b3bc6dbd6f3b97d01104349cb0a">More...</a><br/></td></tr>
<tr class="separator:gad3d21b3bc6dbd6f3b97d01104349cb0a"><td class="memSeparator" colspan="2">&#160;</td></tr>
<tr class="memitem:ga32ac508c5467813a84f74f96655dc697"><td class="memItemLeft" align="right" valign="top">arm_status&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group__NNConv.html#ga32ac508c5467813a84f74f96655dc697">arm_depthwise_separable_conv_HWC_q7_nonsquare</a> (const q7_t *Im_in, const uint16_t dim_im_in_x, const uint16_t dim_im_in_y, const uint16_t ch_im_in, const q7_t *wt, const uint16_t ch_im_out, const uint16_t dim_kernel_x, const uint16_t dim_kernel_y, const uint16_t padding_x, const uint16_t padding_y, const uint16_t stride_x, const uint16_t stride_y, const q7_t *bias, const uint16_t bias_shift, const uint16_t out_shift, q7_t *Im_out, const uint16_t dim_im_out_x, const uint16_t dim_im_out_y, q15_t *bufferA, q7_t *bufferB)</td></tr>
<tr class="memdesc:ga32ac508c5467813a84f74f96655dc697"><td class="mdescLeft">&#160;</td><td class="mdescRight">Q7 depthwise separable convolution function (non-square shape)  <a href="#ga32ac508c5467813a84f74f96655dc697">More...</a><br/></td></tr>
<tr class="separator:ga32ac508c5467813a84f74f96655dc697"><td class="memSeparator" colspan="2">&#160;</td></tr>
</table>
<a name="details" id="details"></a><h2 class="groupheader">Description</h2>
<p>Collection of convolution, depthwise convolution functions and their variants.</p>
<p>The convolution is implemented in 2 steps: im2col and GEMM</p>
<p>im2col is a process of converting each patch of image data into a column. After im2col, the convolution is computed as matrix-matrix multiplication.</p>
<p>To reduce the memory footprint, the im2col is performed partially. Each iteration, only a few column (i.e., patches) are generated and computed with GEMM kernels similar to CMSIS-DSP arm_mat_mult functions. </p>
<h2 class="groupheader">Function Documentation</h2>
<a class="anchor" id="ga2d9c129189832572cf0bc97ff9e73e54"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_1_x_n_s8 </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_batches</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_activation_max</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>buffer_a</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input</td><td>pointer to input tensor. Format: [N, H, W, in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_x</td><td>input tensor dimension x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_batches</td><td>argument is not used. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel</td><td>pointer to kernel weights. Format: [out_ch, H, W, in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_ch</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>kernel width along x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_x</td><td>padding along x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>stride along x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to per channel bias. Range : int32 </td></tr>
    <tr><td class="paramdir">[out]</td><td class="paramname">output</td><td>pointer to output tensor. Format: [H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_shift</td><td>pointer to per output channel requantization shift parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_mult</td><td>pointer to per output channel requantization multiplier parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_offset</td><td>output tensor offset. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_offset</td><td>input tensor offset. Range: -127 to 128 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_activation_min</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_activation_max</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_x</td><td>output tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">buffer_a</td><td>pointer to buffer space used for input optimization and is necessary when ARM_MATH_DSP is defined but not ARM_MATH_MVEI. Required space: 2 * input_ch * sizeof(q15_t) bytes Use <a class="el" href="group__NNConv.html#gab34cb94c592df768774d8bcac03d7e5c" title="Get the required additional buffer size for 1xn convolution. ">arm_convolve_1_x_n_s8_get_buffer_size()</a> to get the size </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> if argument constraints fail. or, <code>ARM_MATH_SUCCESS</code> on successful completion.</dd></dl>
<ul>
<li>Supported framework : TensorFlow Lite Micro</li>
<li>The following constrains on the arguments apply<ol type="1">
<li>input_batches equals 1</li>
<li>ouput_x is a multiple of 4</li>
<li>Explicit constraints(since it is for 1xN convolution) -## input_y equals 1 -## output_y equals 1 -## kernel_y equals 1 <dl class="todo"><dt><b><a class="el" href="todo.html#_todo000001">Todo:</a></b></dt><dd>Remove constraint on output_x to make the function generic.</dd></dl>
</li>
</ol>
</li>
</ul>

<p>References <a class="el" href="group__NNConv.html#ga6a0b2161ece62078228ead2d85bbcf28">arm_convolve_s8()</a>, <a class="el" href="group__NNBasicMath.html#ga5224f2f155a59498c97f4d62ae726ea5">arm_nn_mat_mul_core_1x_s8()</a>, <a class="el" href="group__NNBasicMath.html#gae893fd2b0fadf44bcd2870ca2c2c3634">arm_nn_mat_mul_core_4x_s8()</a>, <a class="el" href="arm__convolve__1__x__n__s8_8c.html#a5fa9a687139f39c5033eb566a9e7b6e1">DIM_Y</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#ad935f1ff1a50822e317bdb321ce991ad">MAX</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#adcd021ac91d43a62b2cdecf9a5b971a7">MIN</a>.</p>

</div>
</div>
<a class="anchor" id="gab34cb94c592df768774d8bcac03d7e5c"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">int32_t arm_convolve_1_x_n_s8_get_buffer_size </td>
          <td>(</td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>filter/kernel width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_y</td><td>filter/kernel height </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns required buffer size(bytes) </dd></dl>

</div>
</div>
<a class="anchor" id="ga110adcfdaab356c750c6270aa5e05f29"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_1x1_HWC_q7_fast_nonsquare </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_x</td><td>input tensor dimention x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_y</td><td>input tensor dimention y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_x</td><td>filter kernel size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_y</td><td>filter kernel size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_x</td><td>padding size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_y</td><td>padding size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_x</td><td>output tensor dimension x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_y</td><td>output tensor dimension y </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p>This function is optimized for convolution with 1x1 kernel size (i.e., dim_kernel_x=1 and dim_kernel_y=1). It can be used for the second half of MobileNets [1] after depthwise separable convolution.</p>
<p>This function is the version with full list of optimization tricks, but with some contraints: ch_im_in is multiple of 4 ch_im_out is multiple of 2</p>
<p>[1] MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications <a href="https://arxiv.org/abs/1704.04861">https://arxiv.org/abs/1704.04861</a> </p>

<p>References <a class="el" href="arm__nnfunctions_8h.html#aefe9c7ce9a65060a244b06dffe74c4b3">arm_nn_mat_mult_kernel_q7_q15_reordered()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, <a class="el" href="group__nndata__convert.html#gaba8fd446d5f54760b406ee63b25d1aee">arm_q7_to_q15_reordered_no_shift()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

</div>
</div>
<a class="anchor" id="gadfcad0bae34328dc806dc88cde173faa"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_1x1_s8_fast </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_batches</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_activation_max</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>buffer_a</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input</td><td>pointer to input tensor. Format: [N, H, W, in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_x</td><td>input tensor dimension x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_y</td><td>input tensor dimension y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_batches</td><td>number of input batches </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel</td><td>pointer to kernel weights. Format: [out_ch, H, W, in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_ch</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_x</td><td>padding size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_y</td><td>padding size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to per channel bias. Range : int32 </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">output</td><td>pointer to output tensor. Format: [H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_shift</td><td>pointer to per output channel requantization shift parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_mult</td><td>pointer to per output channel requantization multiplier parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_offset</td><td>output tensor offset. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_offset</td><td>input tensor offset. Range: -127 to 128 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_activation_min</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_activation_max</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_x</td><td>output tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_y</td><td>output tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">buffer_a</td><td>pointer to buffer space used if required by the implementation Use <a class="el" href="group__NNConv.html#ga09ed5df0234c7eb334a7264dd9e4c833" title="Get the required buffer size for the fast 1x1 convolution (non-square shape) s8 convolution function...">arm_convolve_1x1_s8_fast_get_buffer_size()</a> to get the size </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> if argument constraints fail. or, <code>ARM_MATH_SUCCESS</code> on successful completion.</dd></dl>
<ul>
<li>Supported framework : TensorFlow Lite Micro</li>
<li>The following constrains on the arguments apply<ol type="1">
<li>input_ch is a multiple of 4</li>
<li>padding equals 0</li>
<li>Stride equals 1</li>
<li>kernel dimension is 1x1 (Not provided in the argument list) </li>
</ol>
</li>
</ul>

<p>References <a class="el" href="group__NNBasicMath.html#ga5224f2f155a59498c97f4d62ae726ea5">arm_nn_mat_mul_core_1x_s8()</a>, <a class="el" href="group__NNBasicMath.html#gae893fd2b0fadf44bcd2870ca2c2c3634">arm_nn_mat_mul_core_4x_s8()</a>, <a class="el" href="group__NNBasicMath.html#ga360cdaa16c399e3bfbeaaa7296c326e1">arm_nn_mat_mult_nt_t_s8()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a6a13b7a567485da5fc7f0d311318886d">arm_nn_requantize()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#ad935f1ff1a50822e317bdb321ce991ad">MAX</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#adcd021ac91d43a62b2cdecf9a5b971a7">MIN</a>.</p>

</div>
</div>
<a class="anchor" id="ga09ed5df0234c7eb334a7264dd9e4c833"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">int32_t arm_convolve_1x1_s8_fast_get_buffer_size </td>
          <td>(</td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em></td><td>)</td>
          <td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns required buffer size </dd></dl>

</div>
</div>
<a class="anchor" id="ga55701f213b198084b52eab53097f1f58"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q15_basic </td>
          <td>(</td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in</td><td>input tensor dimention </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel</td><td>filter kernel size </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding</td><td>padding sizes </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride</td><td>convolution stride </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out</td><td>output tensor dimension </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns <code>ARM_MATH_SUCCESS</code></dd></dl>
<p><b>Buffer size:</b></p>
<p>bufferA size: ch_im_in*dim_kernel*dim_kernel</p>
<p>bufferB size: 0</p>
<p>This basic version is designed to work for any input tensor and weight dimension. </p>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

</div>
</div>
<a class="anchor" id="ga4efb1ccbbaa7dd936961989dcb443f50"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q15_fast </td>
          <td>(</td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in</td><td>input tensor dimention </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel</td><td>filter kernel size </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding</td><td>padding sizes </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride</td><td>convolution stride </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out</td><td>output tensor dimension </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p><b>Buffer size:</b></p>
<p>bufferA size: 2*ch_im_in*dim_kernel*dim_kernel</p>
<p>bufferB size: 0</p>
<p><b>Input dimension constraints:</b></p>
<p>ch_im_in is multiple of 2</p>
<p>ch_im_out is multipe of 2 </p>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

</div>
</div>
<a class="anchor" id="ga614ec3b71eb96e29952ec3f09e7b9c3c"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q15_fast_nonsquare </td>
          <td>(</td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q15_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_x</td><td>input tensor dimention x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_y</td><td>input tensor dimention y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_x</td><td>filter kernel size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_y</td><td>filter kernel size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_x</td><td>padding size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_y</td><td>padding size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_x</td><td>output tensor dimension x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_y</td><td>output tensor dimension y </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p><b>Buffer size:</b></p>
<p>bufferA size: 2*ch_im_in*dim_kernel*dim_kernel</p>
<p>bufferB size: 0</p>
<p><b>Input dimension constraints:</b></p>
<p>ch_im_in is multiple of 2</p>
<p>ch_im_out is multipe of 2 </p>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

</div>
</div>
<a class="anchor" id="ga210ae8d8fc1d12ee15b41f1fa6947681"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q7_basic </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in</td><td>input tensor dimention </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel</td><td>filter kernel size </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding</td><td>padding sizes </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride</td><td>convolution stride </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out</td><td>output tensor dimension </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns <code>ARM_MATH_SUCCESS</code></dd></dl>
<p><b>Buffer size:</b></p>
<p>bufferA size: 2*ch_im_in*dim_kernel*dim_kernel</p>
<p>bufferB size: 0</p>
<p>This basic version is designed to work for any input tensor and weight dimension. </p>

<p>References <a class="el" href="arm__nnfunctions_8h.html#abc4fb258cfe8500ee68e812a293a80a3">arm_nn_mat_mult_kernel_q7_q15()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, <a class="el" href="group__nndata__convert.html#gae349de4dba8d253c89d45794ccf05680">arm_q7_to_q15_no_shift()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

</div>
</div>
<a class="anchor" id="ga4501fa22c0836002aa47ccc313dce252"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q7_basic_nonsquare </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<p>Basic Q7 convolution function (non-square shape)</p>
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_x</td><td>input tensor dimention x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_y</td><td>input tensor dimention y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_x</td><td>filter kernel size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_y</td><td>filter kernel size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_x</td><td>padding size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_y</td><td>padding size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_x</td><td>output tensor dimension x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_y</td><td>output tensor dimension y </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns <code>ARM_MATH_SUCCESS</code> </dd></dl>

<p>References <a class="el" href="arm__nnfunctions_8h.html#abc4fb258cfe8500ee68e812a293a80a3">arm_nn_mat_mult_kernel_q7_q15()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, <a class="el" href="group__nndata__convert.html#gae349de4dba8d253c89d45794ccf05680">arm_q7_to_q15_no_shift()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

</div>
</div>
<a class="anchor" id="gae00d3c1285907d59657369fc98bcc83f"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q7_fast </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in</td><td>input tensor dimention </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel</td><td>filter kernel size </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding</td><td>padding sizes </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride</td><td>convolution stride </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out</td><td>output tensor dimension </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p><b>Buffer size:</b></p>
<p>bufferA size: 2*ch_im_in*dim_kernel*dim_kernel</p>
<p>bufferB size: 0</p>
<p><b>Input dimension constraints:</b></p>
<p>ch_im_in is multiple of 4 ( because of the SIMD32 read and swap )</p>
<p>ch_im_out is multipe of 2 ( bacause 2x2 mat_mult kernel )</p>
<p>The im2col converts the Q7 tensor input into Q15 column, which is stored in bufferA. There is reordering happenning during this im2col process with arm_q7_to_q15_reordered_no_shift. For every four elements, the second and third elements are swapped.</p>
<p>The computation kernel arm_nn_mat_mult_kernel_q7_q15_reordered does the GEMM computation with the reordered columns.</p>
<p>To speed-up the determination of the padding condition, we split the computation into 3x3 parts, i.e., {top, mid, bottom} X {left, mid, right}. This reduces the total number of boundary condition checks and improves the data copying performance. </p>

<p>References <a class="el" href="arm__nnfunctions_8h.html#aefe9c7ce9a65060a244b06dffe74c4b3">arm_nn_mat_mult_kernel_q7_q15_reordered()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, <a class="el" href="group__nndata__convert.html#gaba8fd446d5f54760b406ee63b25d1aee">arm_q7_to_q15_reordered_no_shift()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

<p>Referenced by <a class="el" href="arm__nnexamples__cifar10_8cpp.html#ae66f6b31b5ad750f1fe042a706a4e3d4">main()</a>.</p>

</div>
</div>
<a class="anchor" id="gabc6d6b991024e9e5c5cdbd7489de88ef"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q7_fast_nonsquare </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_x</td><td>input tensor dimention x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_y</td><td>input tensor dimention y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_x</td><td>filter kernel size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_y</td><td>filter kernel size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_x</td><td>padding size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_y</td><td>padding size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_x</td><td>output tensor dimension x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_y</td><td>output tensor dimension y </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p>This function is the version with full list of optimization tricks, but with some contraints: ch_im_in is multiple of 4 ch_im_out is multiple of 2 </p>

<p>References <a class="el" href="arm__nnfunctions_8h.html#aefe9c7ce9a65060a244b06dffe74c4b3">arm_nn_mat_mult_kernel_q7_q15_reordered()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, <a class="el" href="group__nndata__convert.html#gaba8fd446d5f54760b406ee63b25d1aee">arm_q7_to_q15_reordered_no_shift()</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>.</p>

</div>
</div>
<a class="anchor" id="ga98f2ead67d7cbdf558b0cd8a3b8fc148"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_HWC_q7_RGB </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<p>Q7 version of convolution for RGB image.</p>
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in</td><td>input tensor dimention </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel</td><td>filter kernel size </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding</td><td>padding sizes </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride</td><td>convolution stride </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out</td><td>output tensor dimension </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p><b>Buffer size:</b></p>
<p>bufferA size: 2*ch_im_in*dim_kernel*dim_kernel</p>
<p>bufferB size: 0</p>
<p><b>Input dimension constraints:</b></p>
<p>ch_im_in equals 3</p>
<p>This kernel is written exclusively for convolution with ch_im_in equals 3. This applies on the first layer of CNNs which has input image with RGB format. </p>

<p>References <a class="el" href="arm__nnfunctions_8h.html#abc4fb258cfe8500ee68e812a293a80a3">arm_nn_mat_mult_kernel_q7_q15()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a31fa46f17212fbcb65d2cdae06fc08f5">arm_nn_read_q7x4()</a>, <a class="el" href="unionarm__nnword.html#a9b5e49e4e2c4b7203e07b305386bb2ba">arm_nnword::half_words</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>, and <a class="el" href="unionarm__nnword.html#a35c7b2ae25e35e0ddcd9ec0a1a6f8d18">arm_nnword::word</a>.</p>

<p>Referenced by <a class="el" href="arm__nnexamples__cifar10_8cpp.html#ae66f6b31b5ad750f1fe042a706a4e3d4">main()</a>.</p>

</div>
</div>
<a class="anchor" id="ga6a0b2161ece62078228ead2d85bbcf28"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_convolve_s8 </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_batches</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_max</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>buffer_a</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input</td><td>pointer to input tensor. Range: int8, format: [N,H,W,in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_x</td><td>input tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_y</td><td>input tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_batches</td><td>number of input batches </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel</td><td>pointer to kernel weights. Range: int8, format: [out_ch, H, W, in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_ch</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>filter/kernel width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_y</td><td>filter/kernel height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_x</td><td>padding along width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_y</td><td>padding along height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to per output channel bias. Range: int32 </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">output</td><td>pointer to output tensor. format: [H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_shift</td><td>pointer to per output channel requantization shift parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_mult</td><td>pointer to per output channel requantization multiplier parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_offset</td><td>output tensor offset. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_offset</td><td>input tensor offset. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_min</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_max</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_x</td><td>output tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_y</td><td>output tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">buffer_a</td><td>pointer to buffer space used for input optimization(partial im2col) and is necessary when ARM_MATH_DSP is defined. Required space: (2 * input_ch * kernel_x * kernel_y) * sizeof(q15_t) bytes Use <a class="el" href="group__NNConv.html#gaaf83450fd2d89432d9759df99b05737b" title="Get the required buffer size for s8 convolution function. ">arm_convolve_s8_get_buffer_size()</a> to get the size. </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns <code>ARM_MATH_SUCCESS</code></dd></dl>
<ol type="1">
<li>Supported framework: TensorFlow Lite micro</li>
<li>q7 is used as data type eventhough it is s8 data. It is done so to be consistent with existing APIs.</li>
<li>Additional memory is required for optimization. Refer to argument 'buffer_a' for details. </li>
</ol>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#a085459f6280c8f586ddd64998dece532">arm_memcpy_q7()</a>, <a class="el" href="group__NNBasicMath.html#gae893fd2b0fadf44bcd2870ca2c2c3634">arm_nn_mat_mul_core_4x_s8()</a>, <a class="el" href="arm__nnfunctions_8h.html#a2da69a593037381cd149616f2fd57cce">arm_nn_mat_mult_kernel_s8_s16()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#aaf7a3da870f2c5e86cdef0d3975e3d69">arm_nn_mat_mult_s8()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#afdda94a339b76615d3161e9fc63f4d21">arm_nn_read_q15x2_ia()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a6a13b7a567485da5fc7f0d311318886d">arm_nn_requantize()</a>, <a class="el" href="group__nndata__convert.html#gab2d48529ae05b350a015f8f1e6596a31">arm_q7_to_q15_with_offset()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#ad935f1ff1a50822e317bdb321ce991ad">MAX</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#adcd021ac91d43a62b2cdecf9a5b971a7">MIN</a>.</p>

<p>Referenced by <a class="el" href="group__NNConv.html#ga2d9c129189832572cf0bc97ff9e73e54">arm_convolve_1_x_n_s8()</a>.</p>

</div>
</div>
<a class="anchor" id="gaaf83450fd2d89432d9759df99b05737b"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">int32_t arm_convolve_s8_get_buffer_size </td>
          <td>(</td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>filter/kernel width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_y</td><td>filter/kernel height </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns required buffer size </dd></dl>

</div>
</div>
<a class="anchor" id="gae86b03224e6edf28f54247594b6c8bf9"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_depthwise_conv_3x3_s8 </td>
          <td>(</td>
          <td class="paramtype">const int8_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int8_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">int8_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_max</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>dilation_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>dilation_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">int16_t *&#160;</td>
          <td class="paramname"><em>buffer_a</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input</td><td>pointer to input tensor. Range: int8, format: [H,W,in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_x</td><td>input tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_y</td><td>input tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel</td><td>pointer to kernel weights. Range: int8, Format: [in_ch, H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_ch</td><td>Number of output channels. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_x</td><td>padding along width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_y</td><td>padding along height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride along width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride along height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to per output channel bias. Range: int8 </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">output</td><td>pointer to output tensor. Format: [H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_shift</td><td>pointer to per output channel requantization shift parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_mult</td><td>pointer to per output channel requantization multiplier parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_x</td><td>output tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_y</td><td>output tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_offset</td><td>offset to elements of output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_offset</td><td>offset to elements of input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_min</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_max</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_x</td><td>dilation along x. Not used. Dilation factor of 1 is used. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_y</td><td>dilation along y. Not used. Dilation factor of 1 is used. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">buffer_a</td><td>Buffer for partial im2col optimization. Not used.</td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns one of the following <code>ARM_MATH_SIZE_MISMATCH</code> - Unsupported dimension of tensors <code>ARM_MATH_ARGUMENT_ERROR</code> - Unsupported pad size along the x axis <code>ARM_MATH_SUCCESS</code> - Successful operation</dd></dl>
<p>Supported framework: TensorFlow Lite </p>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#a31fa46f17212fbcb65d2cdae06fc08f5">arm_nn_read_q7x4()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a6a13b7a567485da5fc7f0d311318886d">arm_nn_requantize()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#ad935f1ff1a50822e317bdb321ce991ad">MAX</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#adcd021ac91d43a62b2cdecf9a5b971a7">MIN</a>.</p>

</div>
</div>
<a class="anchor" id="gae7e0c94c970bfd59d9c5e2331ce84fe7"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_depthwise_conv_s8 </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_max</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dilation_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dilation_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>buffer_a</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input</td><td>pointer to input tensor. Range: int8, format: [H,W,in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_x</td><td>input tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_y</td><td>input tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel</td><td>pointer to kernel weights. Range: int8, format: [in_ch, H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_ch</td><td>Number of output channels. output_ch = ch_mult * input_ch </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_mult</td><td>channel multiplier. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>filter/kernel width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_y</td><td>filter/kernel height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_x</td><td>padding along width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_y</td><td>padding along height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride along width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride along height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to per output channel bias. Range: int32 </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">output</td><td>pointer to output tensor. Format: [H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_shift</td><td>pointer to per output channel requantization shift parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_mult</td><td>pointer to per output channel requantization multiplier parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_x</td><td>output tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_y</td><td>output tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_offset</td><td>offset to elements of output tensor. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_offset</td><td>offset to elements of input tensor. Range: -127 to 128 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_min</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_max</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_x</td><td>dilation along x. Not used. Dilation factor of 1 is used. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_y</td><td>dilation along y. Not used. Dilation factor of 1 is used. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">buffer_a</td><td>Not used.</td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns <code>ARM_MATH_SUCCESS</code></dd></dl>
<ol type="1">
<li>Supported framework: TensorFlow Lite</li>
<li>q7 is used as data type eventhough it is s8 data. It is done so to be consistent with existing APIs.</li>
<li>Optimization using DSP extension is not available for the generic case where channel multiplier is &gt; 1. </li>
</ol>

<p>References <a class="el" href="group__NNConv.html#ga00ec3089e06c8953b4e12582583261c9">depthwise_conv_s8_generic()</a>, and <a class="el" href="group__NNConv.html#ga06d7783f6788faad5165fd0ae583bc59">depthwise_conv_s8_mult_4()</a>.</p>

<p>Referenced by <a class="el" href="group__NNConv.html#ga3acc8b0cd5bd5d975fdf5a615261f0b1">arm_depthwise_conv_s8_opt()</a>.</p>

</div>
</div>
<a class="anchor" id="ga3acc8b0cd5bd5d975fdf5a615261f0b1"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_depthwise_conv_s8_opt </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_max</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dilation_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dilation_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>buffer_a</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input</td><td>pointer to input tensor. Range: int8, format: [H,W,in_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_x</td><td>input tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_y</td><td>input tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel</td><td>pointer to kernel weights. Range: int8, Format: [in_ch, H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_ch</td><td>Number of output channels. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>filter/kernel width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_y</td><td>filter/kernel height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_x</td><td>padding along width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_y</td><td>padding along height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride along width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride along height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to per output channel bias. Range: int8 </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">output</td><td>pointer to output tensor. Format: [H, W, out_ch] </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_shift</td><td>pointer to per output channel requantization shift parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_mult</td><td>pointer to per output channel requantization multiplier parameter. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_x</td><td>output tensor width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_y</td><td>output tensor height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_offset</td><td>offset to elements of output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_offset</td><td>offset to elements of input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_min</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_max</td><td>Minimum value to clamp the output to. Range: int8 </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_x</td><td>dilation along x. Not used. Dilation factor of 1 is used. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_y</td><td>dilation along y. Not used. Dilation factor of 1 is used. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">buffer_a</td><td>Buffer for partial im2col optimization. This is mandatory when ARM_MATH_DSP is defined. Required space: (2 * input_ch * kernel_x * kernel_y) * sizeof(q15_t) bytes Use <a class="el" href="group__NNConv.html#gaa53d31632c765b9b4bbd490d0ae3c0c1" title="Get the required buffer size for optimized s8 depthwise convolution function with constraint that in_...">arm_depthwise_conv_s8_opt_get_buffer_size()</a> to get the size.</td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns one of the following <code>ARM_MATH_SIZE_MISMATCH</code> - Unsupported dimension of tensors <code>ARM_MATH_SUCCESS</code> - Successful operation</dd></dl>
<dl class="section note"><dt>Note</dt><dd>If number of channels is not a multiple of 4, upto 3 elements outside the boundary will be read out for the following if MVE optimizations(Arm Helium Technology) are used.<ul>
<li>Output shift</li>
<li>Output multiplier</li>
<li>Output bias</li>
<li>kernel</li>
</ul>
</dd></dl>
<ol type="1">
<li>Supported framework: TensorFlow Lite</li>
<li>q7 is used as data type eventhough it is s8 data. It is done so to be consistent with existing APIs.</li>
<li>Reccomended when number of channels is 4 or greater. </li>
</ol>

<p>References <a class="el" href="group__NNConv.html#gae7e0c94c970bfd59d9c5e2331ce84fe7">arm_depthwise_conv_s8()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a085459f6280c8f586ddd64998dece532">arm_memcpy_q7()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a7018797f0abd9595cc0dc0aedbb5f17e">arm_memset_q7()</a>, <a class="el" href="group__NNBasicMath.html#gaf7e914e84c2fa6bcf9b0727b8c46c250">arm_nn_depthwise_conv_nt_t_padded_s8()</a>, <a class="el" href="group__NNBasicMath.html#ga7f74841dd08196377cc29ce98deb2c12">arm_nn_depthwise_conv_nt_t_s8()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a9d505dfe68f4486c2327df8a857a89ad">arm_nn_read_q15x2()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a31fa46f17212fbcb65d2cdae06fc08f5">arm_nn_read_q7x4()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a6a13b7a567485da5fc7f0d311318886d">arm_nn_requantize()</a>, <a class="el" href="group__nndata__convert.html#gab2d48529ae05b350a015f8f1e6596a31">arm_q7_to_q15_with_offset()</a>, <a class="el" href="arm__nnexamples__cifar10_8cpp.html#a1f988afc6d7fe29ded1eafbf79a70386">col_buffer</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#ad935f1ff1a50822e317bdb321ce991ad">MAX</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#adcd021ac91d43a62b2cdecf9a5b971a7">MIN</a>.</p>

</div>
</div>
<a class="anchor" id="gaa53d31632c765b9b4bbd490d0ae3c0c1"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">int32_t arm_depthwise_conv_s8_opt_get_buffer_size </td>
          <td>(</td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>filter/kernel width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_y</td><td>filter/kernel height </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns required buffer size </dd></dl>

</div>
</div>
<a class="anchor" id="ga02a296cb4f6361a70c3ecf1ef1238292"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_depthwise_conv_u8_basic_ver1 </td>
          <td>(</td>
          <td class="paramtype">const uint8_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint8_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int16_t&#160;</td>
          <td class="paramname"><em>ch_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int16_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int16_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int16_t&#160;</td>
          <td class="paramname"><em>dilation_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int16_t&#160;</td>
          <td class="paramname"><em>dilation_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>filter_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">uint8_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_max</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>out_mult</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<p>uint8 depthwise convolution function with asymmetric quantization for even number of channel multiplier and input channels. Unless specified otherwise, arguments are mandatory.</p>
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">input</td><td>Pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_x</td><td>Width of input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_y</td><td>Height of input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_ch</td><td>Channels in input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel</td><td>Pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_x</td><td>Width of kernel </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">kernel_y</td><td>Height of kernel </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_mult</td><td>Number of channel multiplier </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_x</td><td>Padding sizes x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">pad_y</td><td>Padding sizes y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>Convolution stride along the width </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>Convolution stride along the height </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_x</td><td>Dilation along width. Not used and intended for future enhancement. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dilation_y</td><td>Dilation along height. Not used and intended for future enhancement. </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>Pointer to optional bias values. If no bias is availble, NULL is expected </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">input_offset</td><td>Input tensor zero offset </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">filter_offset</td><td>Kernel tensor zero offset </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_offset</td><td>Output tensor zero offset </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">output</td><td>Pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_x</td><td>Width of output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_y</td><td>Height of output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_min</td><td>Minimum value to clamp the output to. Range : {0, 255} </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">output_activation_max</td><td>Minimum value to clamp the output to. Range : {0, 255} </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>Amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_mult</td><td>Output multiplier for requantization </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns one of the following <code>ARM_MATH_SIZE_MISMATCH</code> - Not supported dimension of tensors <code>ARM_MATH_SUCCESS</code> - Successful operation <code>ARM_MATH_ARGUMENT_ERROR</code> - Implementation not available</dd></dl>
<p><b> Input constraints</b> ch_mult is multiple of 2 kernel_x is multiple of 2 </p>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#ab6dbc2fd53fae3ccdd1d0d70c8d3b491">arm_nn_divide_by_power_of_two()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a51721c10f116c9f5b8d9908367989d44">arm_nn_sat_doubling_high_mult()</a>, <a class="el" href="arm__depthwise__conv__u8__basic__ver1_8c.html#aeb98996ffa34fb40d8e91919d2ebdc56">DILATION_X</a>, <a class="el" href="arm__depthwise__conv__u8__basic__ver1_8c.html#a9371ac7b5689f9e74a18a2d548c32033">DILATION_Y</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a2953f124ae444ebdd2b2a0873ac36b7e">LEFT_SHIFT</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#a26af54489c1401b91595bf0c92ef87c4">RIGHT_SHIFT</a>.</p>

</div>
</div>
<a class="anchor" id="gad3d21b3bc6dbd6f3b97d01104349cb0a"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_depthwise_separable_conv_HWC_q7 </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in</td><td>input tensor dimention </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel</td><td>filter kernel size </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding</td><td>padding sizes </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride</td><td>convolution stride </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out</td><td>output tensor dimension </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p><b>Buffer size:</b></p>
<p>bufferA size: 2*ch_im_in*dim_kernel*dim_kernel</p>
<p>bufferB size: 0</p>
<p><b>Input dimension constraints:</b></p>
<p>ch_im_in equals ch_im_out</p>
<p>Implementation: There are 3 nested loop here: Inner loop: calculate each output value with MAC instruction over an accumulator Mid loop: loop over different output channel Outer loop: loop over different output (x, y) </p>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#a31fa46f17212fbcb65d2cdae06fc08f5">arm_nn_read_q7x4()</a>, <a class="el" href="unionarm__nnword.html#ac7cff6480a8e29d95f29b73cb1267249">arm_nnword::bytes</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>, and <a class="el" href="unionarm__nnword.html#a35c7b2ae25e35e0ddcd9ec0a1a6f8d18">arm_nnword::word</a>.</p>

</div>
</div>
<a class="anchor" id="ga32ac508c5467813a84f74f96655dc697"></a>
<div class="memitem">
<div class="memproto">
      <table class="memname">
        <tr>
          <td class="memname">arm_status arm_depthwise_separable_conv_HWC_q7_nonsquare </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>Im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_in_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_in</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>wt</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>padding_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>bias_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>out_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>Im_out</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>dim_im_out_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q15_t *&#160;</td>
          <td class="paramname"><em>bufferA</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>bufferB</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
</div><div class="memdoc">
<dl class="params"><dt>Parameters</dt><dd>
  <table class="params">
    <tr><td class="paramdir">[in]</td><td class="paramname">Im_in</td><td>pointer to input tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_x</td><td>input tensor dimention x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_in_y</td><td>input tensor dimention y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_in</td><td>number of input tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">wt</td><td>pointer to kernel weights </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">ch_im_out</td><td>number of filters, i.e., output tensor channels </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_x</td><td>filter kernel size x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_kernel_y</td><td>filter kernel size y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_x</td><td>padding sizes x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">padding_y</td><td>padding sizes y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_x</td><td>convolution stride x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">stride_y</td><td>convolution stride y </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias</td><td>pointer to bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">bias_shift</td><td>amount of left-shift for bias </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">out_shift</td><td>amount of right-shift for output </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">Im_out</td><td>pointer to output tensor </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_x</td><td>output tensor dimension x </td></tr>
    <tr><td class="paramdir">[in]</td><td class="paramname">dim_im_out_y</td><td>output tensor dimension y </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferA</td><td>pointer to buffer space for input </td></tr>
    <tr><td class="paramdir">[in,out]</td><td class="paramname">bufferB</td><td>pointer to buffer space for output </td></tr>
  </table>
  </dd>
</dl>
<dl class="section return"><dt>Returns</dt><dd>The function returns either <code>ARM_MATH_SIZE_MISMATCH</code> or <code>ARM_MATH_SUCCESS</code> based on the outcome of size checking.</dd></dl>
<p>This function is the version with full list of optimization tricks, but with some contraints: ch_im_in is equal to ch_im_out </p>

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#a31fa46f17212fbcb65d2cdae06fc08f5">arm_nn_read_q7x4()</a>, <a class="el" href="unionarm__nnword.html#ac7cff6480a8e29d95f29b73cb1267249">arm_nnword::bytes</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#a4cbd428a2b4a4f6b2a6e4219520c7ce0">NN_ROUND</a>, and <a class="el" href="unionarm__nnword.html#a35c7b2ae25e35e0ddcd9ec0a1a6f8d18">arm_nnword::word</a>.</p>

</div>
</div>
<a class="anchor" id="ga00ec3089e06c8953b4e12582583261c9"></a>
<div class="memitem">
<div class="memproto">
<table class="mlabels">
  <tr>
  <td class="mlabels-left">
      <table class="memname">
        <tr>
          <td class="memname">static void depthwise_conv_s8_generic </td>
          <td>(</td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const q7_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>ch_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">q7_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const uint16_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_max</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
  </td>
  <td class="mlabels-right">
<span class="mlabels"><span class="mlabel">static</span></span>  </td>
  </tr>
</table>
</div><div class="memdoc">

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#a6a13b7a567485da5fc7f0d311318886d">arm_nn_requantize()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#ad935f1ff1a50822e317bdb321ce991ad">MAX</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#adcd021ac91d43a62b2cdecf9a5b971a7">MIN</a>.</p>

<p>Referenced by <a class="el" href="group__NNConv.html#gae7e0c94c970bfd59d9c5e2331ce84fe7">arm_depthwise_conv_s8()</a>.</p>

</div>
</div>
<a class="anchor" id="ga06d7783f6788faad5165fd0ae583bc59"></a>
<div class="memitem">
<div class="memproto">
<table class="mlabels">
  <tr>
  <td class="mlabels-left">
      <table class="memname">
        <tr>
          <td class="memname">static void depthwise_conv_s8_mult_4 </td>
          <td>(</td>
          <td class="paramtype">const int8_t *&#160;</td>
          <td class="paramname"><em>input</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int8_t *&#160;</td>
          <td class="paramname"><em>kernel</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_ch</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>ch_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>kernel_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>kernel_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>pad_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>pad_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>stride_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>stride_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>bias</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">int8_t *&#160;</td>
          <td class="paramname"><em>output</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_shift</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t *&#160;</td>
          <td class="paramname"><em>output_mult</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_x</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_y</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>input_offset</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_min</em>, </td>
        </tr>
        <tr>
          <td class="paramkey"></td>
          <td></td>
          <td class="paramtype">const int32_t&#160;</td>
          <td class="paramname"><em>output_activation_max</em>&#160;</td>
        </tr>
        <tr>
          <td></td>
          <td>)</td>
          <td></td><td></td>
        </tr>
      </table>
  </td>
  <td class="mlabels-right">
<span class="mlabels"><span class="mlabel">static</span></span>  </td>
  </tr>
</table>
</div><div class="memdoc">

<p>References <a class="el" href="arm__nnsupportfunctions_8h.html#a6a13b7a567485da5fc7f0d311318886d">arm_nn_requantize()</a>, <a class="el" href="arm__nnsupportfunctions_8h.html#ad935f1ff1a50822e317bdb321ce991ad">MAX</a>, and <a class="el" href="arm__nnsupportfunctions_8h.html#adcd021ac91d43a62b2cdecf9a5b971a7">MIN</a>.</p>

<p>Referenced by <a class="el" href="group__NNConv.html#gae7e0c94c970bfd59d9c5e2331ce84fe7">arm_depthwise_conv_s8()</a>.</p>

</div>
</div>
</div><!-- contents -->
</div><!-- doc-content -->
<!-- start footer part -->
<div id="nav-path" class="navpath"><!-- id is needed for treeview function! -->
  <ul>
    <li class="footer">Generated on Thu Apr 9 2020 15:49:34 for CMSIS-NN Version 1.3.0 by Arm Ltd. All rights reserved.
	<!--
    <a href="http://www.doxygen.org/index.html">
    <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.6 
	-->
	</li>
  </ul>
</div>
</body>
</html>