| <?xml version="1.0" encoding="UTF-8"?> |
| <!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook MathML Module V1.1b1//EN" |
| "http://www.oasis-open.org/docbook/xml/mathml/1.1CR1/dbmathml.dtd"> |
| <refentry> |
| <refentryinfo> |
| <keywordset> |
| <keyword>clEnqueueNDRangeKernel</keyword> |
| </keywordset> |
| </refentryinfo> |
| |
| <refmeta> |
| <refentrytitle>clEnqueueNDRangeKernel</refentrytitle> |
| |
| <refmiscinfo> |
| <copyright> |
| <year>2007-2010</year> |
| <holder>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.</holder> |
| </copyright> |
| </refmiscinfo> |
| <manvolnum>3</manvolnum> |
| </refmeta> |
| |
| <!-- ================================ SYNOPSIS --> |
| |
| <refnamediv id="clEnqueueNDRangeKernel"> |
| <refname>clEnqueueNDRangeKernel</refname> |
| |
| <refpurpose> |
| Enqueues a command to execute a kernel on a device. |
| </refpurpose> |
| </refnamediv> |
| |
| <refsynopsisdiv xmlns:xlink="http://www.w3.org/1999/xlink"><title></title> |
| <funcsynopsis> |
| <funcprototype> |
| <funcdef> |
| <link xlink:href="scalarDataTypes.html">cl_int</link> |
| <function> |
| clEnqueueNDRangeKernel |
| </function> |
| </funcdef> |
| |
| <paramdef><link xlink:href="abstractDataTypes.html">cl_command_queue</link><parameter>command_queue</parameter></paramdef> |
| <paramdef><link xlink:href="abstractDataTypes.html">cl_kernel</link><parameter>kernel</parameter></paramdef> |
| <paramdef><link xlink:href="scalarDataTypes.html">cl_uint</link><parameter>work_dim</parameter></paramdef> |
| <paramdef>const <link xlink:href="scalarDataTypes.html">size_t</link><parameter>*global_work_offset</parameter></paramdef> |
| <paramdef>const <link xlink:href="scalarDataTypes.html">size_t</link><parameter>*global_work_size</parameter></paramdef> |
| <paramdef>const <link xlink:href="scalarDataTypes.html">size_t</link><parameter>*local_work_size</parameter></paramdef> |
| <paramdef><link xlink:href="scalarDataTypes.html">cl_uint</link><parameter>num_events_in_wait_list</parameter></paramdef> |
| <paramdef>const <link xlink:href="abstractDataTypes.html">cl_event</link><parameter>*event_wait_list</parameter></paramdef> |
| <paramdef><link xlink:href="abstractDataTypes.html">cl_event</link><parameter>*event</parameter></paramdef> |
| |
| </funcprototype> |
| </funcsynopsis> |
| </refsynopsisdiv> |
| |
| |
| <!-- ================================ PARAMETERS --> |
| |
| <refsect1 id="parameters"> |
| <title>Parameters</title> |
| <variablelist> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| command_queue |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| A valid command-queue. The kernel will be queued for execution on the device associated with <varname>command_queue</varname>. |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| kernel |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| A valid kernel object. The OpenCL context associated with <varname>kernel</varname> and <varname>command_queue</varname> must be the same. |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| work_dim |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| The number of dimensions used to specify the global work-items and work-items |
| in the work-group. <varname>work_dim</varname> must be greater than zero and less than or |
| equal to CL_DEVICE_MAX_WORK_ITEM_DIMENSIONS. |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| global_work_offset |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| <varname>global_work_offset</varname> can be used to specify an array |
| of <varname>work_dim</varname> unsigned values that describe the offset used to calculate |
| the global ID of a work-item. If <varname>global_work_offset</varname> is NULL, |
| the global IDs start at offset (0, 0, ... 0). |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| global_work_size |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| Points to an array of <varname>work_dim</varname> unsigned values that describe the number of global work-items in <varname>work_dim</varname> dimensions that will execute the kernel function. The total number of global work-items is computed as <varname>global_work_size</varname>[0] *...* <varname>global_work_size</varname>[<varname>work_dim</varname> - 1]. </para> |
| <para> |
| The values specified in <varname>global_work_size</varname> + corresponding values specified in <varname>global_work_offset</varname> cannot exceed the range given by the <constant>sizeof(size_t)</constant> for the device on which the kernel execution will be enqueued. The <constant>sizeof(size_t)</constant> for a device can be determined using <constant>CL_DEVICE_ADDRESS_BITS</constant> in the table of OpenCL Device Queries for <citerefentry><refentrytitle>clGetDeviceInfo</refentrytitle></citerefentry>. If, for example, <constant>CL_DEVICE_ADDRESS_BITS</constant> = 32, i.e. the device uses a 32-bit address space, size_t is a 32-bit unsigned integer and <varname>global_work_size</varname> values must be in the range 1 .. 2^32 - 1. Values outside this range return a <constant>CL_OUT_OF_RESOURCES</constant> error. |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| local_work_size |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| Points to an array of <varname>work_dim</varname> unsigned values that describe the number of work-items that make up a work-group (also referred to as the size of the work-group) that will execute the kernel specified by <varname>kernel</varname>. The total number of work-items in a work-group is computed as <varname>local_work_size</varname>[0] *... * <varname>local_work_size</varname>[<varname>work_dim</varname> - 1]. The total number of work-items in the work-group must be less than or equal to the |
| <constant>CL_DEVICE_MAX_WORK_GROUP_SIZE</constant> value specified in table of OpenCL Device Queries for <citerefentry><refentrytitle>clGetDeviceInfo</refentrytitle></citerefentry> and the number of work-items specified in <varname>local_work_size</varname>[0],... <varname>local_work_size</varname>[<varname>work_dim</varname> - 1] must be less than or |
| equal to the corresponding values specified by <constant>CL_DEVICE_MAX_WORK_ITEM_SIZES</constant>[0],.... |
| <constant>CL_DEVICE_MAX_WORK_ITEM_SIZES</constant>[<varname>work_dim</varname> - 1]. The explicitly specified <varname>local_work_size</varname> will be used to determine how to break the global work-items specified by <varname>global_work_size</varname> into appropriate work-group instances. If <varname>local_work_size</varname> is specified, the values specified in <varname>global_work_size</varname>[0],... <varname>global_work_size</varname>[<varname>work_dim</varname> - 1] must be evenly divisible by the corresponding values specified in <varname>local_work_size</varname>[0],... <varname>local_work_size</varname>[<varname>work_dim</varname> - 1].</para> |
| |
| <para>The work-group size to be used for <varname>kernel</varname> can also be specified in the program source using the <constant><citerefentry href="attribute"><refentrytitle>__attribute__</refentrytitle></citerefentry>((reqd_work_group_size(X, Y, Z)))</constant>qualifier. In this case the size of work group specified by <varname>local_work_size</varname> must match the value specified by the <constant>reqd_work_group_size</constant> <citerefentry href="attribute"><refentrytitle>__attribute__</refentrytitle></citerefentry> qualifier.</para> |
| |
| <para><varname>local_work_size</varname> can also be a NULL value in which case the OpenCL implementation will determine how to be break the global work-items into appropriate work-group instances.</para> |
| <para>See note for more information. |
| </para> |
| <para> |
| These work-group instances are executed in parallel across multiple compute units or concurrently on the same compute unit. |
| </para> |
| <para> |
| Each work-item is uniquely identified by a global identifier. The global ID, which can be read inside the kernel, is computed using the value given by <varname>global_work_size</varname> and <varname>global_work_offset</varname>. In addition, a work-item is also identified within a work-group by a unique local ID. The local ID, which can also be read by the kernel, is computed using the value given by <varname>local_work_size</varname>. The starting local ID is always (0, 0, ... 0). |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| event_wait_list |
| </varname> and |
| <varname> <!-- parameter name --> |
| num_events_in_wait_list |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| Specify events that need to complete before this |
| particular command can be executed. If <varname>event_wait_list</varname> is NULL, then this particular command does not wait on any event to complete. If <varname>event_wait_list</varname> is NULL, <varname>num_events_in_wait_list</varname> |
| must be 0. If <varname>event_wait_list</varname> is not NULL, the list of events pointed to by <varname>event_wait_list</varname> must be valid and <varname>num_events_in_wait_list</varname> must be greater than 0. The events specified in |
| <varname>event_wait_list</varname> act as synchronization points. The context associated with events in <varname>event_wait_list</varname> and <varname>command_queue</varname> must be the same. |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| <varlistentry> |
| <term> |
| <varname> <!-- parameter name --> |
| event |
| </varname> |
| </term> |
| |
| <listitem> |
| <para> <!-- parameter description --> |
| Returns an event object that identifies this particular kernel execution instance. Event objects are unique and can be used to identify a particular kernel execution instance later on. If <varname>event</varname> is NULL, no event will be created for this kernel execution instance and therefore it will not be possible for the application to query or queue a wait for this particular kernel execution instance. |
| </para> |
| </listitem> |
| </varlistentry> |
| |
| </variablelist> |
| </refsect1> |
| |
| |
| <!-- ================================ NOTES --> |
| |
| <!-- |
| <refsect1 id="notes"><title>Notes</title> |
| <para> |
| |
| </para> |
| |
| </refsect1> --> |
| |
| |
| <!-- ================================ ERRORS --> |
| |
| |
| <refsect1 id="errors"><title>Errors</title> |
| <para> |
| Returns <errorname>CL_SUCCESS</errorname> if the kernel execution was successfully queued. |
| Otherwise, it returns one of the following errors: |
| </para> |
| |
| <itemizedlist mark="disc"> |
| <listitem> |
| <errorname>CL_INVALID_PROGRAM_EXECUTABLE</errorname> if there is no successfully built program |
| executable available for device associated with <varname>command_queue</varname>. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_COMMAND_QUEUE</errorname> if <varname>command_queue</varname> is not a |
| valid command-queue. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_KERNEL</errorname> if <varname>kernel</varname> is not a valid kernel object. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_CONTEXT</errorname> if context associated with <varname>command_queue</varname> |
| and <varname>kernel</varname> is not the same or if the context associated |
| with <varname>command_queue</varname> and events in <varname>event_wait_list</varname> are not the same. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_KERNEL_ARGS</errorname> if the kernel argument values have not been specified. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_WORK_DIMENSION</errorname> if <varname>work_dim</varname> is not a valid value |
| (i.e. a value between 1 and 3). |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_GLOBAL_WORK_SIZE</errorname> if <varname>global_work_size</varname> is NULL, or |
| if any of the values specified in <varname>global_work_size</varname>[0], ...<varname>global_work_size</varname> |
| [<varname>work_dim</varname> - 1] are 0 or exceed the range given by the <code>sizeof(size_t)</code> for the |
| device on which the kernel execution will be enqueued. |
| </listitem> |
| <listitem> |
| <errorname>CL_INVALID_GLOBAL_OFFSET</errorname> if the value specified in <varname>global_work_size</varname> |
| + the corresponding values in <varname>global_work_offset</varname> for any dimensions is greater |
| than the <code>sizeof(size_t)</code> for the device on which the kernel execution will be enqueued. |
| </listitem> |
| <listitem> |
| <errorname>CL_INVALID_WORK_GROUP_SIZE</errorname> if <varname>local_work_size</varname> is |
| specified and number of work-items specified by <varname>global_work_size</varname> is not evenly |
| divisable by size of work-group given by <varname>local_work_size</varname> or does not match the |
| work-group size specified for <varname>kernel</varname> using |
| the |
| <errorname><citerefentry href="functionQualifiers"><refentrytitle>__attribute__</refentrytitle></citerefentry> |
| ((reqd_work_group_size(X, Y, Z)))</errorname> qualifier in program source. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_WORK_GROUP_SIZE</errorname> if <varname>local_work_size</varname> is specified and the total number of work-items in the work-group computed as <varname>local_work_size</varname>[0] *... <varname>local_work_size</varname>[<varname>work_dim</varname> - 1] is greater than the value specified by <errorname>CL_DEVICE_MAX_WORK_GROUP_SIZE</errorname> in the table of OpenCL Device Queries for <citerefentry><refentrytitle>clGetDeviceInfo</refentrytitle></citerefentry>. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_WORK_GROUP_SIZE</errorname> if <varname>local_work_size</varname> is NULL and the <errorname><citerefentry href="functionQualifiers"><refentrytitle>__attribute__</refentrytitle></citerefentry>((reqd_work_group_size(X, Y, Z)))</errorname> qualifier is used to declare the work-group size for <varname>kernel</varname> in the program source. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_WORK_ITEM_SIZE</errorname> if the number of work-items specified in any of <varname>local_work_size</varname>[0], ... <varname>local_work_size</varname>[<varname>work_dim</varname> - 1] is greater than the corresponding values specified by <errorname>CL_DEVICE_MAX_WORK_ITEM_SIZES</errorname>[0], .... |
| <errorname>CL_DEVICE_MAX_WORK_ITEM_SIZES</errorname>[<varname>work_dim</varname> - 1]. |
| </listitem> |
| <listitem> |
| <errorname>CL_MISALIGNED_SUB_BUFFER_OFFSET</errorname> if a sub-buffer object is specified as the value |
| for an argument that is a buffer object and the <varname>offset</varname> specified when the sub-buffer object |
| is created is not aligned to CL_DEVICE_MEM_BASE_ADDR_ALIGN value for device associated with <varname>queue</varname>. |
| </listitem> |
| <listitem> |
| <errorname>CL_INVALID_IMAGE_SIZE</errorname> if an image object is specified as an argument value |
| and the image dimensions (image width, height, specified or compute row and/or slice pitch) are not |
| supported by device associated with <varname>queue</varname>. |
| </listitem> |
| <listitem> |
| <errorname>CL_OUT_OF_RESOURCES</errorname> if there is a failure to queue the execution instance |
| of <varname>kernel</varname> on the command-queue because of insufficient resources needed to execute the kernel. |
| For example, the explicitly specified <varname>local_work_size</varname> causes a failure to execute the kernel |
| because of insufficient resources such as registers or local memory. Another example would be the number of |
| read-only image args used in <varname>kernel</varname> exceed the <errorname>CL_DEVICE_MAX_READ_IMAGE_ARGS</errorname> |
| value for device or the number of write-only image args used in <varname>kernel</varname> exceed |
| the <errorname>CL_DEVICE_MAX_WRITE_IMAGE_ARGS</errorname> value for device or the number of samplers used |
| in <varname>kernel</varname> exceed <errorname>CL_DEVICE_MAX_SAMPLERS</errorname> for device. |
| </listitem> |
| <listitem> |
| <errorname>CL_OUT_OF_RESOURCES</errorname> if there is a failure to allocate resources required by the |
| OpenCL implementation on the device. |
| </listitem> |
| <listitem> |
| <errorname>CL_MEM_OBJECT_ALLOCATION_FAILURE</errorname> if there is a failure to allocate memory |
| for data store associated with image or buffer objects specified as arguments to <varname>kernel</varname>. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_INVALID_EVENT_WAIT_LIST</errorname> if <varname>event_wait_list</varname> is NULL |
| and <varname>num_events_in_wait_list</varname> > 0, or <varname>event_wait_list</varname> is not NULL |
| and <varname>num_events_in_wait_list</varname> is 0, or if event objects in <varname>event_wait_list</varname> |
| are not valid events. |
| </listitem> |
| |
| <listitem> |
| <errorname>CL_OUT_OF_HOST_MEMORY</errorname> if there is a failure to allocate resources required |
| by the OpenCL implementation on the host. |
| </listitem> |
| |
| </itemizedlist> |
| </refsect1> |
| |
| |
| <!-- ================================ EXAMPLE --> |
| <!-- DO NOT DELETE IN CASE AN EXAMPLE IS ADDED IN THE FUTURE --> |
| <!-- |
| <refsect2 id="example1"> |
| <title> |
| Example |
| </title> |
| |
| <informaltable frame="none"> |
| <tgroup cols="1" align="left" colsep="0" rowsep="0"> |
| <colspec colname="col1" colnum="1" /> |
| <tbody> |
| <row> |
| <entry> |
| Example goes here - it will be set in "code" type with white space preserved. |
| </entry> |
| </row> |
| </tbody> |
| </tgroup> |
| </informaltable> |
| </refsect2> |
| --> |
| |
| |
| <!-- ================================ SPECIFICATION --> |
| <!-- Set the "uri" attribute in the <olink /> element to the "named destination" for the PDF page |
| --> |
| <refsect1 id="specification"><title>Specification</title> |
| <para> |
| <imageobject> |
| <imagedata fileref="pdficon_small1.gif" format="gif" /> |
| </imageobject> |
| |
| <olink uri="clEnqueueNDRangeKernel">OpenCL Specification</olink> |
| </para> |
| </refsect1> |
| |
| |
| <!-- ================================ ALSO SEE --> |
| |
| <refsect1 id="seealso"><title>Also see</title> |
| <para> |
| <citerefentry><refentrytitle>clCreateCommandQueue</refentrytitle></citerefentry>, <citerefentry><refentrytitle>clGetDeviceInfo</refentrytitle></citerefentry>, |
| <citerefentry><refentrytitle>clEnqueueNativeKernel</refentrytitle></citerefentry>, |
| <citerefentry><refentrytitle>clEnqueueTask</refentrytitle></citerefentry>, |
| <citerefentry href="workItemFunctions"><refentrytitle>Work-Item Functions</refentrytitle></citerefentry> |
| </para> |
| </refsect1> |
| |
| |
| |
| <!-- ============================== COPYRIGHT --> |
| <!-- Content included from copyright.inc.xsl --> |
| |
| <refsect3 id="Copyright"><title></title> |
| <imageobject> |
| <imagedata fileref="KhronosLogo.jpg" format="jpg" /> |
| </imageobject> |
| <para /> |
| </refsect3> |
| |
| </refentry> |
| |
