blob: 83c268b41650f5a7c2f823e389d3f5615f82a4e6 [file] [log] [blame]
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "xhtml1-transitional.dtd">
<!-- saved from url=(0013)about:internet -->
<?xml-stylesheet type="text/xsl" href="mathml.xsl"?><html xmlns="http://www.w3.org/1999/xhtml" xmlns:pref="http://www.w3.org/2002/Math/preference" xmlns:xlink="http://www.w3.org/1999/xlink" pref:renderer="mathplayer-dl">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<style xmlns="" type="text/css">
/* This style sets a margin around the entire page */
html, body {
margin: 10px;
}
p {
font: normal 16px verdana, sans-serif;
margin: 0;
padding-bottom:12px;
}
h1 {
font: bold 25px verdana, sans-serif;
margin-top: 0;
margin-bottom: 3px;
padding-top: 0;
padding-bottom: 0;
}
h2 {
font: bold 19px verdana, sans-serif;
margin-top: 28px;
margin-bottom: 3px;
padding-top: 0;
padding-bottom: 0;
}
h3 {
font: bold 19px verdana, sans-serif !important;
margin-top: 28px;
margin-bottom: 3px;
padding-top: 0;
padding-bottom: 0;
}
li {
font: normal 16px verdana, sans-serif;
margin-top: 0;
margin-bottom: 18px;
padding-top: 0;
padding-bottom: 0;
}
.pdparam {
font: italic 16px verdana, sans-serif;
}
.term {
font: italic 16px verdana, sans-serif;
font-weight: normal;
}
.type {
font: normal 16px verdana, sans-serif !important;
}
.parameter {
font-style: italic;
}
a:link, a:visited {
color: blue;
text-decoration: none;
font: normal 16px;
}
a:hover {
background-color: #FFFF99;
font: normal 16px;
}
div.funcsynopsis {
text-align: left;
background-color: #e6e6e6;
font: normal 16px verdana, sans-serif;
padding-top: 10px;
padding-bottom: 10px;
}
div.funcsynopsis table {
border-collapse: separate;
font: normal 16px verdana, sans-serif;
}
div.funcsynopsis td {
background-color: #e6e6e6;
border: 0 solid #000;
padding: 1px;
font: normal 16px verdana, sans-serif;
}
div.refsect1 {
font-family: verdana, sans-serif;
font-size: 16px;
}
code.constant {
font: normal 16px courier new, monospace !important;
}
span.errorname {
font: normal 16px verdana, sans-serif !important;
}
code.function {
font: bold 16px verdana, sans-serif !important;
}
b.fsfunc {
font: bold 16px verdana, sans-serif !important;
}
code.varname {
font: italic 16px verdana, sans-serif;
}
code.replaceable {
font: italic 16px courier new, monospace;
}
code.funcdef {
font: normal 16px verdana, sans-serif !important;
}
.citerefentry {
font: normal 16px verdana, sans-serif !important;
}
.parameter {
font-style: italic;
}
code.fsfunc {
font: normal 16px verdana, sans-serif !important;
}
/* PARAMETER: This style controls spacing between the terms in Parameter section */
dt {
margin-top: 15px;
}
/* TABLES: These styles apply to all tables OTHER than the Synopsis and Example tables */
div.refsect1 table {
width: 100%;
margin-top: 10px;
background-color: #FFF;
border-collapse: collapse;
border-color: #000;
border-width: 1px;
font: normal 16px verdana, sans-serif;
}
div.refsect1 th {
border-collapse: collapse;
border-color: #000;
border-width: 1px;
font: bold 16px verdana, sans-serif;
}
div.refsect1 td {
background-color: #FFF;
padding: 5px;
vertical-align: text-top;
border-collapse: collapse;
border-color: #000;
border-width: 1px;
font: normal 16px verdana, sans-serif;
}
div.refsect1 p{
font: normal 16px verdana, sans-serif;
margin-top: 8px;
margin-bottom: 8px;
padding-top: 0;
padding-bottom: 0;
}
/* EXAMPLE: These styles apply only to the Example section */
div.refsect2 {
font: normal 16px courier new, monospace !important;
}
div.refsect2 table {
margin-top: 0;
background-color: #e6e6e6;
width: 100%;
border: 0 solid #000;
padding: 2px;
font: normal 16px courier new, monospace !important;
}
div.refsect2 td {
background-color: #e6e6e6;
font: normal 16px courier new, monospace !important;
white-space:pre;
}
/* COPYRIGHT: This style formats the text of the copyright statement at the bottom of the page */
div.refsect3 {
font: normal 11px verdana, sans-serif;
margin-top: 50px;
margin-bottom: 20px;
padding-top: 0;
padding-bottom: 0;
}
</style>
<title>Vector Data Types</title>
<meta name="generator" content="DocBook XSL Stylesheets V1.79.1" />
<meta name="keywords" content=" &#10; Vector Data Types&#10; " />
</head>
<body>
<div class="refentry">
<a id="idm1"></a>
<div class="titlepage"></div>
<div xmlns="" class="refnamediv">
<a xmlns="http://www.w3.org/1999/xhtml" id="Vector Data Types"></a>
<h1>Vector Data Types</h1>
<p>
Built-in vector data types.
</p>
</div>
<div class="refsect1">
<a id="description"></a>
<h2>Description</h2>
<p>
The <span class="type">char</span>, <span class="type">unsigned char</span>, <span class="type">short</span>, <span class="type">unsigned short</span>, <span class="type">integer</span>, <span class="type">unsigned integer</span>, <span class="type">long</span>, <span class="type">unsigned long</span>, and <span class="type">float</span> vector data types are supported. The vector data type is defined with the type name i.e. <span class="type">char</span>, <span class="type">uchar</span>, <span class="type">short</span>, <span class="type">ushort</span>, <span class="type">int</span>, <span class="type">uint</span>, <span class="type">float</span>, <span class="type">long</span>, and <span class="type">ulong</span> followed by a literal value <span class="emphasis"><em>n</em></span> that defines
the number of elements in the vector. Supported values of <span class="emphasis"><em>n</em></span> are 2, 4, 8, and 16.
</p>
<p>
The built-in vector data types are also declared as appropriate types in the OpenCL API (and header files) that can be used by an application. The following table describes the built-in vector data type in the OpenCL C programming language and the corresponding data type available to the application:
</p>
<div class="informaltable">
<table class="informaltable" border="1">
<colgroup>
<col align="left" class="col1" />
<col align="left" class="col2" />
<col align="left" class="col3" />
</colgroup>
<thead>
<tr>
<th align="left">Type in OpenCL Language</th>
<th align="left">Description</th>
<th align="left">API type for application</th>
</tr>
</thead>
<tbody>
<tr>
<td align="left">
<span class="type">char<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 8-bit signed two's complement integer vector.
</td>
<td align="left">
<span class="type">cl_char<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">uchar<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 8-bit unsigned integer vector.
</td>
<td align="left">
<span class="type">cl_uchar<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">short<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 16-bit signed two's complement integer vector.
</td>
<td align="left">
<span class="type">cl_short<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">ushort<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 16-bit unsigned integer vector.
</td>
<td align="left">
<span class="type">cl_ushort<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">int<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 32-bit signed two's complement integer vector.
</td>
<td align="left">
<span class="type">cl_int</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">uint<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 32-bit unsigned integer vector.
</td>
<td align="left">
<span class="type">cl_uint<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">long<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 64-bit signed two's complement integer vector.
</td>
<td align="left">
<span class="type">cl_long<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">ulong<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A 64-bit unsigned integer vector.
</td>
<td align="left">
<span class="type">cl_ulong<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">float<em class="replaceable"><code>n</code></em></span>
</td>
<td align="left">
A float vector
</td>
<td align="left">
<span class="type">cl_float<em class="replaceable"><code>n</code></em></span>
</td>
</tr>
</tbody>
</table>
</div>
</div>
<div class="refsect1">
<a id="notes"></a>
<h2>Notes</h2>
<p>
Built-in vector data types are supported by the OpenCL implementation even if the underlying compute device does not support any or all of the vector data types. These are to be converted by the device compiler to appropriate instructions that use underlying built-in types supported natively by the compute device.
</p>
</div>
<div class="refsect1">
<a id="optional"></a>
<h2>Optional Double Precision and Half Floating Point</h2>
<p>
OpenCL 1.0 adds support for double precision and half floating-point as optional extensions. The <span class="type">double</span> data type must confirm to the IEEE-754 double precision storage format.
</p>
<p>
An application that wants to use <span class="type">double</span> will need to include the <a class="citerefentry" href="cl_khr_fp64.html"><span class="citerefentry"><span class="refentrytitle">#pragma OPENCL
EXTENSION cl_khr_fp64 : enable</span></span></a> directive before any double precision data type is declared in the kernel code. This will extended the list of built-in vector and scalar data types to include the following:
</p>
<p>
</p>
<div class="informaltable">
<table class="informaltable" border="1">
<colgroup>
<col align="left" class="col1" />
<col align="left" class="col2" />
<col align="left" class="col3" />
</colgroup>
<thead>
<tr>
<th align="left">Type in OpenCL Language</th>
<th align="left">Description</th>
<th align="left">API type for application</th>
</tr>
</thead>
<tbody>
<tr>
<td align="left">
<span class="type">double</span>
</td>
<td align="left">A double precision float.</td>
<td align="left">
<span class="type">cl_double</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">double2</span>
</td>
<td align="left">A 2-component double vector.</td>
<td align="left">
<span class="type">cl_double2</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">double4</span>
</td>
<td align="left">A 4-component double vector.</td>
<td align="left">
<span class="type">cl_double4</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">double8</span>
</td>
<td align="left">An 8-component double vector.</td>
<td align="left">
<span class="type">cl_double8</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">double16</span>
</td>
<td align="left">A 16-component double vector.</td>
<td align="left">
<span class="type">cl_double16</span>
</td>
</tr>
</tbody>
</table>
</div>
<p>
</p>
<p>
An application that wants to use <span class="type">half</span> and <span class="type">half<em class="replaceable"><code>n</code></em></span> will need to include the <a class="citerefentry" href="cl_khr_fp16.html"><span class="citerefentry"><span class="refentrytitle">#pragma OPENCL
EXTENSION cl_khr_fp16 : enable</span></span></a> directive. This will extended the list of built-in vector and scalar data types to include the following:
</p>
<p>
</p>
<div class="informaltable">
<table class="informaltable" border="1">
<colgroup>
<col align="left" class="col1" />
<col align="left" class="col2" />
<col align="left" class="col3" />
</colgroup>
<thead>
<tr>
<th align="left">Type in OpenCL Language</th>
<th align="left">Description</th>
<th align="left">API type for application</th>
</tr>
</thead>
<tbody>
<tr>
<td align="left">
<span class="type">half2</span>
</td>
<td align="left">A 2-component half-precision floating-point vector.</td>
<td align="left">
<span class="type">cl_half2</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">half4</span>
</td>
<td align="left">A 4-component half-precision floating-point vector.</td>
<td align="left">
<span class="type">cl_half4</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">half8</span>
</td>
<td align="left">An 8-component half-precision floating-point vector.</td>
<td align="left">
<span class="type">cl_half8</span>
</td>
</tr>
<tr>
<td align="left">
<span class="type">half16</span>
</td>
<td align="left">A 16-component half-precision floating-point vector.</td>
<td align="left">
<span class="type">cl_half16</span>
</td>
</tr>
</tbody>
</table>
</div>
<p>
</p>
</div>
<div class="refsect1">
<a id="specification"></a>
<h2>Specification</h2>
<p>
<img src="pdficon_small1.gif" />
<a href="http://www.khronos.org/registry/cl/specs/opencl-1.0.pdf#page=129" target="OpenCL Spec">OpenCL Specification</a>
</p>
</div>
<div class="refsect1">
<a id="seealso"></a>
<h2>Also see</h2>
<p>
<a class="citerefentry" href="macroLimits.html"><span class="citerefentry"><span class="refentrytitle">Macros and Limits</span></span></a>,
<a class="citerefentry" href="scalarDataTypes.html"><span class="citerefentry"><span class="refentrytitle">Scalar Data Types</span></span></a>,
<a class="citerefentry" href="reservedDataTypes.html"><span class="citerefentry"><span class="refentrytitle">Reserved Data Types</span></span></a>,
<a class="citerefentry" href="otherDataTypes.html"><span class="citerefentry"><span class="refentrytitle">Other Data Types</span></span></a>,
<a class="citerefentry" href="abstractDataTypes.html"><span class="citerefentry"><span class="refentrytitle">Abstract Data Types</span></span></a>,
<a class="citerefentry" href="cl_khr_fp16.html"><span class="citerefentry"><span class="refentrytitle">cl_khr_fp16</span></span></a>,
<a class="citerefentry" href="cl_khr_fp64.html"><span class="citerefentry"><span class="refentrytitle">cl_khr_fp64</span></span></a>
</p>
</div>
<div xmlns="" class="refsect3" lang="en" xml:lang="en"><a xmlns="http://www.w3.org/1999/xhtml" id="Copyright"></a><h4 xmlns="http://www.w3.org/1999/xhtml"></h4><img xmlns="http://www.w3.org/1999/xhtml" src="KhronosLogo.jpg" /><p xmlns="http://www.w3.org/1999/xhtml"></p>Copyright © 2007-2009 The Khronos Group Inc.
Permission is hereby granted, free of charge, to any person obtaining a
copy of this software and/or associated documentation files (the
"Materials"), to deal in the Materials without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Materials, and to
permit persons to whom the Materials are furnished to do so, subject to
the condition that this copyright notice and permission notice shall be included
in all copies or substantial portions of the Materials.
</div>
</div>
</body>
</html>