mirror of
https://github.com/ml-explore/mlx.git
synced 2025-09-21 20:58:08 +08:00
rebase
This commit is contained in:
123
docs/build/html/sort_8h_source.html
vendored
123
docs/build/html/sort_8h_source.html
vendored
@@ -3,7 +3,7 @@
|
||||
<head>
|
||||
<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
|
||||
<meta http-equiv="X-UA-Compatible" content="IE=11"/>
|
||||
<meta name="generator" content="Doxygen 1.12.0"/>
|
||||
<meta name="generator" content="Doxygen 1.13.1"/>
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1"/>
|
||||
<title>MLX: mlx/backend/metal/kernels/sort.h Source File</title>
|
||||
<link href="tabs.css" rel="stylesheet" type="text/css"/>
|
||||
@@ -11,11 +11,18 @@
|
||||
<script type="text/javascript" src="dynsections.js"></script>
|
||||
<script type="text/javascript" src="clipboard.js"></script>
|
||||
<link href="navtree.css" rel="stylesheet" type="text/css"/>
|
||||
<script type="text/javascript" src="navtreedata.js"></script>
|
||||
<script type="text/javascript" src="navtree.js"></script>
|
||||
<script type="text/javascript" src="resize.js"></script>
|
||||
<script type="text/javascript" src="cookie.js"></script>
|
||||
<link href="search/search.css" rel="stylesheet" type="text/css"/>
|
||||
<script type="text/javascript" src="search/searchdata.js"></script>
|
||||
<script type="text/javascript" src="search/search.js"></script>
|
||||
<script type="text/javascript">
|
||||
/* @license magnet:?xt=urn:btih:d3d9a9a6595521f9666a5e94cc830dab83b65699&dn=expat.txt MIT */
|
||||
$(function() { init_search(); });
|
||||
/* @license-end */
|
||||
</script>
|
||||
<link href="doxygen.css" rel="stylesheet" type="text/css" />
|
||||
</head>
|
||||
<body>
|
||||
@@ -28,12 +35,24 @@
|
||||
<div id="projectname">MLX
|
||||
</div>
|
||||
</td>
|
||||
<td> <div id="MSearchBox" class="MSearchBoxInactive">
|
||||
<span class="left">
|
||||
<span id="MSearchSelect" onmouseover="return searchBox.OnSearchSelectShow()" onmouseout="return searchBox.OnSearchSelectHide()"> </span>
|
||||
<input type="text" id="MSearchField" value="" placeholder="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.svg" alt=""/></a>
|
||||
</span>
|
||||
</div>
|
||||
</td>
|
||||
</tr>
|
||||
</tbody>
|
||||
</table>
|
||||
</div>
|
||||
<!-- end header part -->
|
||||
<!-- Generated by Doxygen 1.12.0 -->
|
||||
<!-- Generated by Doxygen 1.13.1 -->
|
||||
<script type="text/javascript">
|
||||
/* @license magnet:?xt=urn:btih:d3d9a9a6595521f9666a5e94cc830dab83b65699&dn=expat.txt MIT */
|
||||
var searchBox = new SearchBox("searchBox", "search/",'.html');
|
||||
@@ -44,22 +63,23 @@ var searchBox = new SearchBox("searchBox", "search/",'.html');
|
||||
$(function() { codefold.init(0); });
|
||||
/* @license-end */
|
||||
</script>
|
||||
<script type="text/javascript" src="menudata.js"></script>
|
||||
<script type="text/javascript" src="menu.js"></script>
|
||||
</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">
|
||||
/* @license magnet:?xt=urn:btih:d3d9a9a6595521f9666a5e94cc830dab83b65699&dn=expat.txt MIT */
|
||||
$(function() {
|
||||
initMenu('',true,false,'search.php','Search',false);
|
||||
$(function() { init_search(); });
|
||||
});
|
||||
/* @license-end */
|
||||
</script>
|
||||
<div id="main-nav"></div>
|
||||
<script type="text/javascript">
|
||||
/* @license magnet:?xt=urn:btih:d3d9a9a6595521f9666a5e94cc830dab83b65699&dn=expat.txt MIT */
|
||||
$(function(){ initResizable(false); });
|
||||
$(function(){initNavTree('sort_8h_source.html',''); initResizable(true); });
|
||||
/* @license-end */
|
||||
</script>
|
||||
<div id="doc-content">
|
||||
<!-- window showing the filter options -->
|
||||
<div id="MSearchSelectWindow"
|
||||
onmouseover="return searchBox.OnSearchSelectShow()"
|
||||
@@ -81,12 +101,6 @@ $(function(){ initResizable(false); });
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<div id="nav-path" class="navpath">
|
||||
<ul>
|
||||
<li class="navelem"><a class="el" href="dir_938ab0ecf10b8b860ff766c820f665fd.html">mlx</a></li><li class="navelem"><a class="el" href="dir_1d446c9bd3c99228254c9484e0bc5c06.html">backend</a></li><li class="navelem"><a class="el" href="dir_d0c977ea65824390717cdb7efc36c157.html">metal</a></li><li class="navelem"><a class="el" href="dir_70a37effa88bcbd6b791977fa1e64356.html">kernels</a></li> </ul>
|
||||
</div>
|
||||
</div><!-- top -->
|
||||
<div id="doc-content">
|
||||
<div class="header">
|
||||
<div class="headertitle"><div class="title">sort.h</div></div>
|
||||
</div><!--header-->
|
||||
@@ -100,7 +114,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00007" name="l00007"></a><span class="lineno"> 7</span> </div>
|
||||
<div class="line"><a id="l00008" name="l00008"></a><span class="lineno"> 8</span><span class="comment">// Based on GPU merge sort algorithm at</span></div>
|
||||
<div class="line"><a id="l00009" name="l00009"></a><span class="lineno"> 9</span><span class="comment">// https://github.com/NVIDIA/cccl/tree/main/cub/cub</span></div>
|
||||
<div class="line"><a id="l00010" name="l00010"></a><span class="lineno"> 10</span> </div>
|
||||
<div class="line"><a id="l00010" name="l00010"></a><span class="lineno"> 10</span></div>
|
||||
<div class="line"><a id="l00012" name="l00012"></a><span class="lineno"> 12</span><span class="comment">// Thread-level sort</span></div>
|
||||
<div class="line"><a id="l00014" name="l00014"></a><span class="lineno"> 14</span> </div>
|
||||
<div class="line"><a id="l00015" name="l00015"></a><span class="lineno"> 15</span><span class="keyword">template</span> <<span class="keyword">typename</span> T></div>
|
||||
@@ -115,7 +129,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00022" name="l00022"></a><span class="lineno"> 22</span><span class="keyword">template</span> <<span class="keyword">typename</span> T></div>
|
||||
<div class="foldopen" id="foldopen00023" data-start="{" data-end="};">
|
||||
<div class="line"><a id="l00023" name="l00023"></a><span class="lineno"><a class="line" href="struct_less_than.html"> 23</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct_less_than.html">LessThan</a> {</div>
|
||||
<div class="line"><a id="l00024" name="l00024"></a><span class="lineno"><a class="line" href="struct_less_than.html#abf97a6b0163048e4ba96460939dbd3a3"> 24</a></span> <span class="keyword">static</span> <span class="keyword">constexpr</span> constant T <a class="code hl_variable" href="struct_less_than.html#abf97a6b0163048e4ba96460939dbd3a3">init</a> = <a class="code hl_struct" href="struct_limits.html">Limits<T>::max</a>;</div>
|
||||
<div class="line"><a id="l00024" name="l00024"></a><span class="lineno"><a class="line" href="struct_less_than.html#abf97a6b0163048e4ba96460939dbd3a3"> 24</a></span> <span class="keyword">static</span> <span class="keyword">constexpr</span> constant T <a class="code hl_variable" href="struct_less_than.html#abf97a6b0163048e4ba96460939dbd3a3">init</a> = <a class="code hl_variable" href="struct_limits.html#a2f0673b6f9da89ce1d64f9f3d74f50a8">Limits<T>::max</a>;</div>
|
||||
<div class="line"><a id="l00025" name="l00025"></a><span class="lineno"> 25</span> </div>
|
||||
<div class="foldopen" id="foldopen00026" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00026" name="l00026"></a><span class="lineno"><a class="line" href="struct_less_than.html#a2798eb377b411c93a4ed30cf35caade2"> 26</a></span> METAL_FUNC <span class="keywordtype">bool</span> <a class="code hl_function" href="struct_less_than.html#a2798eb377b411c93a4ed30cf35caade2">operator()</a>(T a, T b) {</div>
|
||||
@@ -137,13 +151,13 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00038" name="l00038"></a><span class="lineno"><a class="line" href="struct_thread_sort.html#ad9ab3e6b47f7e9b91c0f3b773596986d"> 38</a></span> <span class="keyword">static</span> METAL_FUNC <span class="keywordtype">void</span> <a class="code hl_function" href="struct_thread_sort.html#ad9ab3e6b47f7e9b91c0f3b773596986d">sort</a>(</div>
|
||||
<div class="line"><a id="l00039" name="l00039"></a><span class="lineno"> 39</span> thread val_t (&vals)[N_PER_THREAD],</div>
|
||||
<div class="line"><a id="l00040" name="l00040"></a><span class="lineno"> 40</span> thread idx_t (&idxs)[N_PER_THREAD]) {</div>
|
||||
<div class="line"><a id="l00041" name="l00041"></a><span class="lineno"> 41</span> CompareOp <a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>;</div>
|
||||
<div class="line"><a id="l00041" name="l00041"></a><span class="lineno"> 41</span> CompareOp op;</div>
|
||||
<div class="line"><a id="l00042" name="l00042"></a><span class="lineno"> 42</span> </div>
|
||||
<div class="line"><a id="l00043" name="l00043"></a><span class="lineno"> 43</span> <a class="code hl_define" href="sort_8h.html#ad34b622323cebef136669fedd7229515">MLX_MTL_LOOP_UNROLL</a></div>
|
||||
<div class="line"><a id="l00044" name="l00044"></a><span class="lineno"> 44</span> <span class="keywordflow">for</span> (<span class="keywordtype">short</span> i = 0; i < N_PER_THREAD; ++i) {</div>
|
||||
<div class="line"><a id="l00045" name="l00045"></a><span class="lineno"> 45</span> <a class="code hl_define" href="sort_8h.html#ad34b622323cebef136669fedd7229515">MLX_MTL_LOOP_UNROLL</a></div>
|
||||
<div class="line"><a id="l00046" name="l00046"></a><span class="lineno"> 46</span> <span class="keywordflow">for</span> (<span class="keywordtype">short</span> j = i & 1; j < N_PER_THREAD - 1; j += 2) {</div>
|
||||
<div class="line"><a id="l00047" name="l00047"></a><span class="lineno"> 47</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>(vals[j + 1], vals[j])) {</div>
|
||||
<div class="line"><a id="l00047" name="l00047"></a><span class="lineno"> 47</span> <span class="keywordflow">if</span> (op(vals[j + 1], vals[j])) {</div>
|
||||
<div class="line"><a id="l00048" name="l00048"></a><span class="lineno"> 48</span> <a class="code hl_function" href="sort_8h.html#a6e8c2da4975a8001fd5ddf211a3058b7">thread_swap</a>(vals[j + 1], vals[j]);</div>
|
||||
<div class="line"><a id="l00049" name="l00049"></a><span class="lineno"> 49</span> <a class="code hl_function" href="sort_8h.html#a6e8c2da4975a8001fd5ddf211a3058b7">thread_swap</a>(idxs[j + 1], idxs[j]);</div>
|
||||
<div class="line"><a id="l00050" name="l00050"></a><span class="lineno"> 50</span> }</div>
|
||||
@@ -153,7 +167,7 @@ $(function(){ initResizable(false); });
|
||||
</div>
|
||||
<div class="line"><a id="l00054" name="l00054"></a><span class="lineno"> 54</span>};</div>
|
||||
</div>
|
||||
<div class="line"><a id="l00055" name="l00055"></a><span class="lineno"> 55</span> </div>
|
||||
<div class="line"><a id="l00055" name="l00055"></a><span class="lineno"> 55</span></div>
|
||||
<div class="line"><a id="l00057" name="l00057"></a><span class="lineno"> 57</span><span class="comment">// Threadgroup-level sort</span></div>
|
||||
<div class="line"><a id="l00059" name="l00059"></a><span class="lineno"> 59</span> </div>
|
||||
<div class="line"><a id="l00060" name="l00060"></a><span class="lineno"> 60</span><span class="keyword">template</span> <</div>
|
||||
@@ -165,7 +179,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00066" name="l00066"></a><span class="lineno"> 66</span> <span class="keyword">typename</span> CompareOp></div>
|
||||
<div class="foldopen" id="foldopen00067" data-start="{" data-end="};">
|
||||
<div class="line"><a id="l00067" name="l00067"></a><span class="lineno"><a class="line" href="struct_block_merge_sort.html"> 67</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct_block_merge_sort.html">BlockMergeSort</a> {</div>
|
||||
<div class="line"><a id="l00068" name="l00068"></a><span class="lineno"><a class="line" href="struct_block_merge_sort.html#ad2474d16721f4ceb954125728a0e2ea2"> 68</a></span> <span class="keyword">using </span><a class="code hl_struct" href="struct_thread_sort.html">thread_sort_t</a> =</div>
|
||||
<div class="line"><a id="l00068" name="l00068"></a><span class="lineno"><a class="line" href="struct_block_merge_sort.html#ad2474d16721f4ceb954125728a0e2ea2"> 68</a></span> <span class="keyword">using </span><a class="code hl_typedef" href="struct_block_merge_sort.html#ad2474d16721f4ceb954125728a0e2ea2">thread_sort_t</a> =</div>
|
||||
<div class="line"><a id="l00069" name="l00069"></a><span class="lineno"> 69</span> <a class="code hl_struct" href="struct_thread_sort.html">ThreadSort<val_t, idx_t, ARG_SORT, N_PER_THREAD, CompareOp></a>;</div>
|
||||
<div class="foldopen" id="foldopen00070" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00070" name="l00070"></a><span class="lineno"><a class="line" href="struct_block_merge_sort.html#ab2300cbecb23f3433bad888924c831ca"> 70</a></span> <span class="keyword">static</span> METAL_FUNC <span class="keywordtype">int</span> <a class="code hl_function" href="struct_block_merge_sort.html#ab2300cbecb23f3433bad888924c831ca">merge_partition</a>(</div>
|
||||
@@ -174,7 +188,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00073" name="l00073"></a><span class="lineno"> 73</span> <span class="keywordtype">short</span> A_sz,</div>
|
||||
<div class="line"><a id="l00074" name="l00074"></a><span class="lineno"> 74</span> <span class="keywordtype">short</span> B_sz,</div>
|
||||
<div class="line"><a id="l00075" name="l00075"></a><span class="lineno"> 75</span> <span class="keywordtype">short</span> sort_md) {</div>
|
||||
<div class="line"><a id="l00076" name="l00076"></a><span class="lineno"> 76</span> CompareOp <a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>;</div>
|
||||
<div class="line"><a id="l00076" name="l00076"></a><span class="lineno"> 76</span> CompareOp op;</div>
|
||||
<div class="line"><a id="l00077" name="l00077"></a><span class="lineno"> 77</span> </div>
|
||||
<div class="line"><a id="l00078" name="l00078"></a><span class="lineno"> 78</span> <span class="keywordtype">short</span> A_st = <a class="code hl_function" href="namespacemetal.html#a853c80479ab2264d9c4587c7bcac767b">max</a>(0, sort_md - B_sz);</div>
|
||||
<div class="line"><a id="l00079" name="l00079"></a><span class="lineno"> 79</span> <span class="keywordtype">short</span> A_ed = <a class="code hl_function" href="namespacemetal.html#a6653b28c9473087141eddce39878d4d3">min</a>(sort_md, A_sz);</div>
|
||||
@@ -184,7 +198,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00083" name="l00083"></a><span class="lineno"> 83</span> <span class="keyword">auto</span> a = As[md];</div>
|
||||
<div class="line"><a id="l00084" name="l00084"></a><span class="lineno"> 84</span> <span class="keyword">auto</span> b = Bs[sort_md - 1 - md];</div>
|
||||
<div class="line"><a id="l00085" name="l00085"></a><span class="lineno"> 85</span> </div>
|
||||
<div class="line"><a id="l00086" name="l00086"></a><span class="lineno"> 86</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>(b, a)) {</div>
|
||||
<div class="line"><a id="l00086" name="l00086"></a><span class="lineno"> 86</span> <span class="keywordflow">if</span> (op(b, a)) {</div>
|
||||
<div class="line"><a id="l00087" name="l00087"></a><span class="lineno"> 87</span> A_ed = md;</div>
|
||||
<div class="line"><a id="l00088" name="l00088"></a><span class="lineno"> 88</span> } <span class="keywordflow">else</span> {</div>
|
||||
<div class="line"><a id="l00089" name="l00089"></a><span class="lineno"> 89</span> A_st = md + 1;</div>
|
||||
@@ -205,14 +219,14 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00102" name="l00102"></a><span class="lineno"> 102</span> <span class="keywordtype">short</span> B_sz,</div>
|
||||
<div class="line"><a id="l00103" name="l00103"></a><span class="lineno"> 103</span> thread val_t (&vals)[N_PER_THREAD],</div>
|
||||
<div class="line"><a id="l00104" name="l00104"></a><span class="lineno"> 104</span> thread idx_t (&idxs)[N_PER_THREAD]) {</div>
|
||||
<div class="line"><a id="l00105" name="l00105"></a><span class="lineno"> 105</span> CompareOp <a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>;</div>
|
||||
<div class="line"><a id="l00105" name="l00105"></a><span class="lineno"> 105</span> CompareOp op;</div>
|
||||
<div class="line"><a id="l00106" name="l00106"></a><span class="lineno"> 106</span> <span class="keywordtype">short</span> a_idx = 0;</div>
|
||||
<div class="line"><a id="l00107" name="l00107"></a><span class="lineno"> 107</span> <span class="keywordtype">short</span> b_idx = 0;</div>
|
||||
<div class="line"><a id="l00108" name="l00108"></a><span class="lineno"> 108</span> </div>
|
||||
<div class="line"><a id="l00109" name="l00109"></a><span class="lineno"> 109</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> i = 0; i < N_PER_THREAD; ++i) {</div>
|
||||
<div class="line"><a id="l00110" name="l00110"></a><span class="lineno"> 110</span> <span class="keyword">auto</span> a = As[a_idx];</div>
|
||||
<div class="line"><a id="l00111" name="l00111"></a><span class="lineno"> 111</span> <span class="keyword">auto</span> b = Bs[b_idx];</div>
|
||||
<div class="line"><a id="l00112" name="l00112"></a><span class="lineno"> 112</span> <span class="keywordtype">bool</span> pred = (b_idx < B_sz) && (a_idx >= A_sz || <a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>(b, a));</div>
|
||||
<div class="line"><a id="l00112" name="l00112"></a><span class="lineno"> 112</span> <span class="keywordtype">bool</span> pred = (b_idx < B_sz) && (a_idx >= A_sz || op(b, a));</div>
|
||||
<div class="line"><a id="l00113" name="l00113"></a><span class="lineno"> 113</span> </div>
|
||||
<div class="line"><a id="l00114" name="l00114"></a><span class="lineno"> 114</span> vals[i] = pred ? b : a;</div>
|
||||
<div class="line"><a id="l00115" name="l00115"></a><span class="lineno"> 115</span> idxs[i] = pred ? Bs_idx[b_idx] : As_idx[a_idx];</div>
|
||||
@@ -313,7 +327,7 @@ $(function(){ initResizable(false); });
|
||||
</div>
|
||||
<div class="line"><a id="l00208" name="l00208"></a><span class="lineno"> 208</span>};</div>
|
||||
</div>
|
||||
<div class="line"><a id="l00209" name="l00209"></a><span class="lineno"> 209</span> </div>
|
||||
<div class="line"><a id="l00209" name="l00209"></a><span class="lineno"> 209</span></div>
|
||||
<div class="line"><a id="l00211" name="l00211"></a><span class="lineno"> 211</span><span class="comment">// Kernel sort</span></div>
|
||||
<div class="line"><a id="l00213" name="l00213"></a><span class="lineno"> 213</span> </div>
|
||||
<div class="line"><a id="l00214" name="l00214"></a><span class="lineno"> 214</span><span class="keyword">template</span> <</div>
|
||||
@@ -327,7 +341,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00221" name="l00221"></a><span class="lineno"><a class="line" href="struct_kernel_merge_sort.html"> 221</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct_kernel_merge_sort.html">KernelMergeSort</a> {</div>
|
||||
<div class="line"><a id="l00222" name="l00222"></a><span class="lineno"><a class="line" href="struct_kernel_merge_sort.html#a4e3f09896275956fc4c23e1f157dca3b"> 222</a></span> <span class="keyword">using </span><a class="code hl_typedef" href="struct_kernel_merge_sort.html#a4e3f09896275956fc4c23e1f157dca3b">val_t</a> = T;</div>
|
||||
<div class="line"><a id="l00223" name="l00223"></a><span class="lineno"><a class="line" href="struct_kernel_merge_sort.html#a0df65b709ae7f153a2bf381179d55e00"> 223</a></span> <span class="keyword">using </span><a class="code hl_typedef" href="struct_kernel_merge_sort.html#a0df65b709ae7f153a2bf381179d55e00">idx_t</a> = uint;</div>
|
||||
<div class="line"><a id="l00224" name="l00224"></a><span class="lineno"><a class="line" href="struct_kernel_merge_sort.html#adae7850e057fc30d5328c7b3dcc998fa"> 224</a></span> <span class="keyword">using </span><a class="code hl_struct" href="struct_block_merge_sort.html">block_merge_sort_t</a> = <a class="code hl_struct" href="struct_block_merge_sort.html">BlockMergeSort</a><</div>
|
||||
<div class="line"><a id="l00224" name="l00224"></a><span class="lineno"><a class="line" href="struct_kernel_merge_sort.html#adae7850e057fc30d5328c7b3dcc998fa"> 224</a></span> <span class="keyword">using </span><a class="code hl_typedef" href="struct_kernel_merge_sort.html#adae7850e057fc30d5328c7b3dcc998fa">block_merge_sort_t</a> = <a class="code hl_struct" href="struct_block_merge_sort.html">BlockMergeSort</a><</div>
|
||||
<div class="line"><a id="l00225" name="l00225"></a><span class="lineno"> 225</span> <a class="code hl_typedef" href="struct_kernel_merge_sort.html#a4e3f09896275956fc4c23e1f157dca3b">val_t</a>,</div>
|
||||
<div class="line"><a id="l00226" name="l00226"></a><span class="lineno"> 226</span> <a class="code hl_typedef" href="struct_kernel_merge_sort.html#a0df65b709ae7f153a2bf381179d55e00">idx_t</a>,</div>
|
||||
<div class="line"><a id="l00227" name="l00227"></a><span class="lineno"> 227</span> ARG_SORT,</div>
|
||||
@@ -447,7 +461,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00336" name="l00336"></a><span class="lineno"> 336</span> <span class="keywordtype">short</span> BLOCK_THREADS,</div>
|
||||
<div class="line"><a id="l00337" name="l00337"></a><span class="lineno"> 337</span> <span class="keywordtype">short</span> N_PER_THREAD></div>
|
||||
<div class="foldopen" id="foldopen00338" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00338" name="l00338"></a><span class="lineno"><a class="line" href="sort_8h.html#a4ee3de195a6f9c33aa91ac52461808ad"> 338</a></span>[[kernel, max_total_threads_per_threadgroup(BLOCK_THREADS)]] <span class="keywordtype">void</span> <a class="code hl_function" href="sort_8h.html#a4ee3de195a6f9c33aa91ac52461808ad">block_sort_nc</a>(</div>
|
||||
<div class="line"><a id="l00338" name="l00338"></a><span class="lineno"><a class="line" href="sort_8h.html#ad272ef4422c75565a79b7f5e751bb247"> 338</a></span>[[kernel, max_total_threads_per_threadgroup(BLOCK_THREADS)]] <span class="keywordtype">void</span> <a class="code hl_function" href="sort_8h.html#ad272ef4422c75565a79b7f5e751bb247">block_sort_nc</a>(</div>
|
||||
<div class="line"><a id="l00339" name="l00339"></a><span class="lineno"> 339</span> <span class="keyword">const</span> device T* inp [[buffer(0)]],</div>
|
||||
<div class="line"><a id="l00340" name="l00340"></a><span class="lineno"> 340</span> device U* out [[buffer(1)]],</div>
|
||||
<div class="line"><a id="l00341" name="l00341"></a><span class="lineno"> 341</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>& size_sorted_axis [[buffer(2)]],</div>
|
||||
@@ -455,8 +469,8 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00343" name="l00343"></a><span class="lineno"> 343</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>& out_stride_sorted_axis [[buffer(4)]],</div>
|
||||
<div class="line"><a id="l00344" name="l00344"></a><span class="lineno"> 344</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>& nc_dim [[buffer(5)]],</div>
|
||||
<div class="line"><a id="l00345" name="l00345"></a><span class="lineno"> 345</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* nc_shape [[buffer(6)]],</div>
|
||||
<div class="line"><a id="l00346" name="l00346"></a><span class="lineno"> 346</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* in_nc_strides [[buffer(7)]],</div>
|
||||
<div class="line"><a id="l00347" name="l00347"></a><span class="lineno"> 347</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* out_nc_strides [[buffer(8)]],</div>
|
||||
<div class="line"><a id="l00346" name="l00346"></a><span class="lineno"> 346</span> <span class="keyword">const</span> constant int64_t* in_nc_strides [[buffer(7)]],</div>
|
||||
<div class="line"><a id="l00347" name="l00347"></a><span class="lineno"> 347</span> <span class="keyword">const</span> constant int64_t* out_nc_strides [[buffer(8)]],</div>
|
||||
<div class="line"><a id="l00348" name="l00348"></a><span class="lineno"> 348</span> uint3 tid [[threadgroup_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00349" name="l00349"></a><span class="lineno"> 349</span> uint3 lid [[thread_position_in_threadgroup]]) {</div>
|
||||
<div class="line"><a id="l00350" name="l00350"></a><span class="lineno"> 350</span> <span class="keyword">using </span>sort_kernel =</div>
|
||||
@@ -464,8 +478,8 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00352" name="l00352"></a><span class="lineno"> 352</span> <span class="keyword">using </span>val_t = <span class="keyword">typename</span> sort_kernel::val_t;</div>
|
||||
<div class="line"><a id="l00353" name="l00353"></a><span class="lineno"> 353</span> <span class="keyword">using </span>idx_t = <span class="keyword">typename</span> sort_kernel::idx_t;</div>
|
||||
<div class="line"><a id="l00354" name="l00354"></a><span class="lineno"> 354</span> </div>
|
||||
<div class="line"><a id="l00355" name="l00355"></a><span class="lineno"> 355</span> <span class="keyword">auto</span> in_block_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.y, nc_shape, in_nc_strides, nc_dim);</div>
|
||||
<div class="line"><a id="l00356" name="l00356"></a><span class="lineno"> 356</span> <span class="keyword">auto</span> out_block_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.y, nc_shape, out_nc_strides, nc_dim);</div>
|
||||
<div class="line"><a id="l00355" name="l00355"></a><span class="lineno"> 355</span> <span class="keyword">auto</span> in_block_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a497dd9f1a00c8a4303d8782158a0812a">elem_to_loc</a>(tid.y, nc_shape, in_nc_strides, nc_dim);</div>
|
||||
<div class="line"><a id="l00356" name="l00356"></a><span class="lineno"> 356</span> <span class="keyword">auto</span> out_block_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a497dd9f1a00c8a4303d8782158a0812a">elem_to_loc</a>(tid.y, nc_shape, out_nc_strides, nc_dim);</div>
|
||||
<div class="line"><a id="l00357" name="l00357"></a><span class="lineno"> 357</span> inp += in_block_idx;</div>
|
||||
<div class="line"><a id="l00358" name="l00358"></a><span class="lineno"> 358</span> out += out_block_idx;</div>
|
||||
<div class="line"><a id="l00359" name="l00359"></a><span class="lineno"> 359</span> </div>
|
||||
@@ -511,7 +525,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00398" name="l00398"></a><span class="lineno"> 398</span> <span class="keyword">typename</span> CompareOp = <a class="code hl_struct" href="struct_less_than.html">LessThan<val_t></a>></div>
|
||||
<div class="foldopen" id="foldopen00399" data-start="{" data-end="};">
|
||||
<div class="line"><a id="l00399" name="l00399"></a><span class="lineno"><a class="line" href="struct_kernel_multi_block_merge_sort.html"> 399</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct_kernel_multi_block_merge_sort.html">KernelMultiBlockMergeSort</a> {</div>
|
||||
<div class="line"><a id="l00400" name="l00400"></a><span class="lineno"><a class="line" href="struct_kernel_multi_block_merge_sort.html#af27e9af4b58640c0aa620bc4efc68dff"> 400</a></span> <span class="keyword">using </span><a class="code hl_struct" href="struct_block_merge_sort.html">block_merge_sort_t</a> = <a class="code hl_struct" href="struct_block_merge_sort.html">BlockMergeSort</a><</div>
|
||||
<div class="line"><a id="l00400" name="l00400"></a><span class="lineno"><a class="line" href="struct_kernel_multi_block_merge_sort.html#af27e9af4b58640c0aa620bc4efc68dff"> 400</a></span> <span class="keyword">using </span><a class="code hl_typedef" href="struct_kernel_multi_block_merge_sort.html#af27e9af4b58640c0aa620bc4efc68dff">block_merge_sort_t</a> = <a class="code hl_struct" href="struct_block_merge_sort.html">BlockMergeSort</a><</div>
|
||||
<div class="line"><a id="l00401" name="l00401"></a><span class="lineno"> 401</span> val_t,</div>
|
||||
<div class="line"><a id="l00402" name="l00402"></a><span class="lineno"> 402</span> idx_t,</div>
|
||||
<div class="line"><a id="l00403" name="l00403"></a><span class="lineno"> 403</span> ARG_SORT,</div>
|
||||
@@ -568,7 +582,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00451" name="l00451"></a><span class="lineno"> 451</span> <span class="keywordtype">int</span> A_sz,</div>
|
||||
<div class="line"><a id="l00452" name="l00452"></a><span class="lineno"> 452</span> <span class="keywordtype">int</span> B_sz,</div>
|
||||
<div class="line"><a id="l00453" name="l00453"></a><span class="lineno"> 453</span> <span class="keywordtype">int</span> sort_md) {</div>
|
||||
<div class="line"><a id="l00454" name="l00454"></a><span class="lineno"> 454</span> CompareOp <a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>;</div>
|
||||
<div class="line"><a id="l00454" name="l00454"></a><span class="lineno"> 454</span> CompareOp op;</div>
|
||||
<div class="line"><a id="l00455" name="l00455"></a><span class="lineno"> 455</span> </div>
|
||||
<div class="line"><a id="l00456" name="l00456"></a><span class="lineno"> 456</span> <span class="keywordtype">int</span> A_st = <a class="code hl_function" href="namespacemetal.html#a853c80479ab2264d9c4587c7bcac767b">max</a>(0, sort_md - B_sz);</div>
|
||||
<div class="line"><a id="l00457" name="l00457"></a><span class="lineno"> 457</span> <span class="keywordtype">int</span> A_ed = <a class="code hl_function" href="namespacemetal.html#a6653b28c9473087141eddce39878d4d3">min</a>(sort_md, A_sz);</div>
|
||||
@@ -578,7 +592,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00461" name="l00461"></a><span class="lineno"> 461</span> <span class="keyword">auto</span> a = As[md];</div>
|
||||
<div class="line"><a id="l00462" name="l00462"></a><span class="lineno"> 462</span> <span class="keyword">auto</span> b = Bs[sort_md - 1 - md];</div>
|
||||
<div class="line"><a id="l00463" name="l00463"></a><span class="lineno"> 463</span> </div>
|
||||
<div class="line"><a id="l00464" name="l00464"></a><span class="lineno"> 464</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a>(b, a)) {</div>
|
||||
<div class="line"><a id="l00464" name="l00464"></a><span class="lineno"> 464</span> <span class="keywordflow">if</span> (op(b, a)) {</div>
|
||||
<div class="line"><a id="l00465" name="l00465"></a><span class="lineno"> 465</span> A_ed = md;</div>
|
||||
<div class="line"><a id="l00466" name="l00466"></a><span class="lineno"> 466</span> } <span class="keywordflow">else</span> {</div>
|
||||
<div class="line"><a id="l00467" name="l00467"></a><span class="lineno"> 467</span> A_st = md + 1;</div>
|
||||
@@ -598,7 +612,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00479" name="l00479"></a><span class="lineno"> 479</span> <span class="keywordtype">short</span> BLOCK_THREADS,</div>
|
||||
<div class="line"><a id="l00480" name="l00480"></a><span class="lineno"> 480</span> <span class="keywordtype">short</span> N_PER_THREAD></div>
|
||||
<div class="foldopen" id="foldopen00481" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00481" name="l00481"></a><span class="lineno"><a class="line" href="sort_8h.html#aa48ff1aff1e9dc1301b6781aa0721d6b"> 481</a></span>[[kernel, max_total_threads_per_threadgroup(BLOCK_THREADS)]] <span class="keywordtype">void</span> <a class="code hl_function" href="sort_8h.html#aa48ff1aff1e9dc1301b6781aa0721d6b">mb_block_sort</a>(</div>
|
||||
<div class="line"><a id="l00481" name="l00481"></a><span class="lineno"><a class="line" href="sort_8h.html#a7b5b1180af2b8c6ab814eadd701ce4ba"> 481</a></span>[[kernel, max_total_threads_per_threadgroup(BLOCK_THREADS)]] <span class="keywordtype">void</span> <a class="code hl_function" href="sort_8h.html#a7b5b1180af2b8c6ab814eadd701ce4ba">mb_block_sort</a>(</div>
|
||||
<div class="line"><a id="l00482" name="l00482"></a><span class="lineno"> 482</span> <span class="keyword">const</span> device val_t* inp [[buffer(0)]],</div>
|
||||
<div class="line"><a id="l00483" name="l00483"></a><span class="lineno"> 483</span> device val_t* out_vals [[buffer(1)]],</div>
|
||||
<div class="line"><a id="l00484" name="l00484"></a><span class="lineno"> 484</span> device idx_t* out_idxs [[buffer(2)]],</div>
|
||||
@@ -606,7 +620,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00486" name="l00486"></a><span class="lineno"> 486</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>& stride_sorted_axis [[buffer(4)]],</div>
|
||||
<div class="line"><a id="l00487" name="l00487"></a><span class="lineno"> 487</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>& nc_dim [[buffer(5)]],</div>
|
||||
<div class="line"><a id="l00488" name="l00488"></a><span class="lineno"> 488</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* nc_shape [[buffer(6)]],</div>
|
||||
<div class="line"><a id="l00489" name="l00489"></a><span class="lineno"> 489</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* nc_strides [[buffer(7)]],</div>
|
||||
<div class="line"><a id="l00489" name="l00489"></a><span class="lineno"> 489</span> <span class="keyword">const</span> constant int64_t* nc_strides [[buffer(7)]],</div>
|
||||
<div class="line"><a id="l00490" name="l00490"></a><span class="lineno"> 490</span> uint3 tid [[threadgroup_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00491" name="l00491"></a><span class="lineno"> 491</span> uint3 lid [[thread_position_in_threadgroup]]) {</div>
|
||||
<div class="line"><a id="l00492" name="l00492"></a><span class="lineno"> 492</span> <span class="keyword">using </span>sort_kernel = <a class="code hl_struct" href="struct_kernel_multi_block_merge_sort.html">KernelMultiBlockMergeSort</a><</div>
|
||||
@@ -616,7 +630,7 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00496" name="l00496"></a><span class="lineno"> 496</span> BLOCK_THREADS,</div>
|
||||
<div class="line"><a id="l00497" name="l00497"></a><span class="lineno"> 497</span> N_PER_THREAD>;</div>
|
||||
<div class="line"><a id="l00498" name="l00498"></a><span class="lineno"> 498</span> </div>
|
||||
<div class="line"><a id="l00499" name="l00499"></a><span class="lineno"> 499</span> <span class="keyword">auto</span> block_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.y, nc_shape, nc_strides, nc_dim);</div>
|
||||
<div class="line"><a id="l00499" name="l00499"></a><span class="lineno"> 499</span> <span class="keyword">auto</span> block_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a497dd9f1a00c8a4303d8782158a0812a">elem_to_loc</a>(tid.y, nc_shape, nc_strides, nc_dim);</div>
|
||||
<div class="line"><a id="l00500" name="l00500"></a><span class="lineno"> 500</span> inp += block_idx;</div>
|
||||
<div class="line"><a id="l00501" name="l00501"></a><span class="lineno"> 501</span> out_vals += tid.y * size_sorted_axis;</div>
|
||||
<div class="line"><a id="l00502" name="l00502"></a><span class="lineno"> 502</span> out_idxs += tid.y * size_sorted_axis;</div>
|
||||
@@ -818,44 +832,49 @@ $(function(){ initResizable(false); });
|
||||
<div class="line"><a id="l00694" name="l00694"></a><span class="lineno"> 694</span> }</div>
|
||||
<div class="line"><a id="l00695" name="l00695"></a><span class="lineno"> 695</span>}</div>
|
||||
</div>
|
||||
<div class="ttc" id="abackend_2metal_2kernels_2utils_8h_html_a22eaa505dbc7dd2a63a895f2e16712f5"><div class="ttname"><a href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a></div><div class="ttdeci">METAL_FUNC IdxT elem_to_loc(uint elem, constant const int *shape, constant const StrideT *strides, int ndim)</div><div class="ttdef"><b>Definition</b> utils.h:93</div></div>
|
||||
<div class="ttc" id="acommon_2binary_8h_html_a70228731d29946574b238d21fb4b360c"><div class="ttname"><a href="common_2binary_8h.html#a70228731d29946574b238d21fb4b360c">op</a></div><div class="ttdeci">Op op</div><div class="ttdef"><b>Definition</b> binary.h:129</div></div>
|
||||
<div class="ttc" id="abackend_2metal_2kernels_2utils_8h_html_a497dd9f1a00c8a4303d8782158a0812a"><div class="ttname"><a href="backend_2metal_2kernels_2utils_8h.html#a497dd9f1a00c8a4303d8782158a0812a">elem_to_loc</a></div><div class="ttdeci">METAL_FUNC IdxT elem_to_loc(IdxT elem, constant const int *shape, constant const int64_t *strides, int ndim)</div><div class="ttdef"><b>Definition</b> utils.h:93</div></div>
|
||||
<div class="ttc" id="anamespacemetal_html"><div class="ttname"><a href="namespacemetal.html">metal</a></div><div class="ttdef"><b>Definition</b> bf16_math.h:226</div></div>
|
||||
<div class="ttc" id="anamespacemetal_html_a6653b28c9473087141eddce39878d4d3"><div class="ttname"><a href="namespacemetal.html#a6653b28c9473087141eddce39878d4d3">metal::min</a></div><div class="ttdeci">METAL_FUNC bfloat16_t min(bfloat16_t x, bfloat16_t y)</div><div class="ttdef"><b>Definition</b> bf16_math.h:232</div></div>
|
||||
<div class="ttc" id="anamespacemetal_html_a853c80479ab2264d9c4587c7bcac767b"><div class="ttname"><a href="namespacemetal.html#a853c80479ab2264d9c4587c7bcac767b">metal::max</a></div><div class="ttdeci">METAL_FUNC bfloat16_t max(bfloat16_t x, bfloat16_t y)</div><div class="ttdef"><b>Definition</b> bf16_math.h:232</div></div>
|
||||
<div class="ttc" id="asort_8h_html_a0386011c52d03e60885a31e6fbd903dd"><div class="ttname"><a href="sort_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a></div><div class="ttdeci">#define MLX_MTL_CONST</div><div class="ttdef"><b>Definition</b> sort.h:3</div></div>
|
||||
<div class="ttc" id="asort_8h_html_a32cbe4163b8b0f5cb2c97b256119a4b2"><div class="ttname"><a href="sort_8h.html#a32cbe4163b8b0f5cb2c97b256119a4b2">mb_block_partition</a></div><div class="ttdeci">void mb_block_partition(device idx_t *block_partitions, const device val_t *dev_vals, const device idx_t *dev_idxs, const constant int &size_sorted_axis, const constant int &merge_tiles, const constant int &n_blocks, uint3 tid, uint3 lid, uint3 tgp_dims)</div><div class="ttdef"><b>Definition</b> sort.h:525</div></div>
|
||||
<div class="ttc" id="asort_8h_html_a4ee3de195a6f9c33aa91ac52461808ad"><div class="ttname"><a href="sort_8h.html#a4ee3de195a6f9c33aa91ac52461808ad">block_sort_nc</a></div><div class="ttdeci">void block_sort_nc(const device T *inp, device U *out, const constant int &size_sorted_axis, const constant int &in_stride_sorted_axis, const constant int &out_stride_sorted_axis, const constant int &nc_dim, const constant int *nc_shape, const constant size_t *in_nc_strides, const constant size_t *out_nc_strides, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:338</div></div>
|
||||
<div class="ttc" id="asort_8h_html_a6e8c2da4975a8001fd5ddf211a3058b7"><div class="ttname"><a href="sort_8h.html#a6e8c2da4975a8001fd5ddf211a3058b7">thread_swap</a></div><div class="ttdeci">METAL_FUNC void thread_swap(thread T &a, thread T &b)</div><div class="ttdef"><b>Definition</b> sort.h:16</div></div>
|
||||
<div class="ttc" id="asort_8h_html_a7b5b1180af2b8c6ab814eadd701ce4ba"><div class="ttname"><a href="sort_8h.html#a7b5b1180af2b8c6ab814eadd701ce4ba">mb_block_sort</a></div><div class="ttdeci">void mb_block_sort(const device val_t *inp, device val_t *out_vals, device idx_t *out_idxs, const constant int &size_sorted_axis, const constant int &stride_sorted_axis, const constant int &nc_dim, const constant int *nc_shape, const constant int64_t *nc_strides, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:481</div></div>
|
||||
<div class="ttc" id="asort_8h_html_a93f14092416169c4449141043ac45ffd"><div class="ttname"><a href="sort_8h.html#a93f14092416169c4449141043ac45ffd">block_sort</a></div><div class="ttdeci">void block_sort(const device T *inp, device U *out, const constant int &size_sorted_axis, const constant int &in_stride_sorted_axis, const constant int &out_stride_sorted_axis, const constant int &in_stride_segment_axis, const constant int &out_stride_segment_axis, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:283</div></div>
|
||||
<div class="ttc" id="asort_8h_html_aa48ff1aff1e9dc1301b6781aa0721d6b"><div class="ttname"><a href="sort_8h.html#aa48ff1aff1e9dc1301b6781aa0721d6b">mb_block_sort</a></div><div class="ttdeci">void mb_block_sort(const device val_t *inp, device val_t *out_vals, device idx_t *out_idxs, const constant int &size_sorted_axis, const constant int &stride_sorted_axis, const constant int &nc_dim, const constant int *nc_shape, const constant size_t *nc_strides, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:481</div></div>
|
||||
<div class="ttc" id="asort_8h_html_ab381cd57f344bc7304ab580bfdc78807"><div class="ttname"><a href="sort_8h.html#ab381cd57f344bc7304ab580bfdc78807">mb_block_merge</a></div><div class="ttdeci">void mb_block_merge(const device idx_t *block_partitions, const device val_t *dev_vals_in, const device idx_t *dev_idxs_in, device val_t *dev_vals_out, device idx_t *dev_idxs_out, const constant int &size_sorted_axis, const constant int &merge_tiles, const constant int &num_tiles, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:579</div></div>
|
||||
<div class="ttc" id="asort_8h_html_aca8b6f36c9024b8406fe545765316dc0"><div class="ttname"><a href="sort_8h.html#aca8b6f36c9024b8406fe545765316dc0">zero_helper</a></div><div class="ttdeci">constant constexpr const int zero_helper</div><div class="ttdef"><b>Definition</b> sort.h:330</div></div>
|
||||
<div class="ttc" id="asort_8h_html_ad272ef4422c75565a79b7f5e751bb247"><div class="ttname"><a href="sort_8h.html#ad272ef4422c75565a79b7f5e751bb247">block_sort_nc</a></div><div class="ttdeci">void block_sort_nc(const device T *inp, device U *out, const constant int &size_sorted_axis, const constant int &in_stride_sorted_axis, const constant int &out_stride_sorted_axis, const constant int &nc_dim, const constant int *nc_shape, const constant int64_t *in_nc_strides, const constant int64_t *out_nc_strides, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:338</div></div>
|
||||
<div class="ttc" id="asort_8h_html_ad34b622323cebef136669fedd7229515"><div class="ttname"><a href="sort_8h.html#ad34b622323cebef136669fedd7229515">MLX_MTL_LOOP_UNROLL</a></div><div class="ttdeci">#define MLX_MTL_LOOP_UNROLL</div><div class="ttdef"><b>Definition</b> sort.h:4</div></div>
|
||||
<div class="ttc" id="astruct_block_merge_sort_html"><div class="ttname"><a href="struct_block_merge_sort.html">BlockMergeSort</a></div><div class="ttdef"><b>Definition</b> sort.h:67</div></div>
|
||||
<div class="ttc" id="astruct_block_merge_sort_html_ab2300cbecb23f3433bad888924c831ca"><div class="ttname"><a href="struct_block_merge_sort.html#ab2300cbecb23f3433bad888924c831ca">BlockMergeSort::merge_partition</a></div><div class="ttdeci">static METAL_FUNC int merge_partition(const threadgroup val_t *As, const threadgroup val_t *Bs, short A_sz, short B_sz, short sort_md)</div><div class="ttdef"><b>Definition</b> sort.h:70</div></div>
|
||||
<div class="ttc" id="astruct_block_merge_sort_html_ab65f190edf1851b37c39ad49ce99a43c"><div class="ttname"><a href="struct_block_merge_sort.html#ab65f190edf1851b37c39ad49ce99a43c">BlockMergeSort::merge_step</a></div><div class="ttdeci">static METAL_FUNC void merge_step(const threadgroup val_t *As, const threadgroup val_t *Bs, const threadgroup idx_t *As_idx, const threadgroup idx_t *Bs_idx, short A_sz, short B_sz, thread val_t(&vals)[N_PER_THREAD], thread idx_t(&idxs)[N_PER_THREAD])</div><div class="ttdef"><b>Definition</b> sort.h:96</div></div>
|
||||
<div class="ttc" id="astruct_block_merge_sort_html_acc970f5eb963f7f2010f5ae5ea8b8bc0"><div class="ttname"><a href="struct_block_merge_sort.html#acc970f5eb963f7f2010f5ae5ea8b8bc0">BlockMergeSort::sort</a></div><div class="ttdeci">static METAL_FUNC void sort(threadgroup val_t *tgp_vals, threadgroup idx_t *tgp_idxs, int size_sorted_axis, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:122</div></div>
|
||||
<div class="ttc" id="astruct_block_merge_sort_html_ad2474d16721f4ceb954125728a0e2ea2"><div class="ttname"><a href="struct_block_merge_sort.html#ad2474d16721f4ceb954125728a0e2ea2">BlockMergeSort::thread_sort_t</a></div><div class="ttdeci">ThreadSort< val_t, idx_t, ARG_SORT, N_PER_THREAD, CompareOp > thread_sort_t</div><div class="ttdef"><b>Definition</b> sort.h:68</div></div>
|
||||
<div class="ttc" id="astruct_kernel_merge_sort_html"><div class="ttname"><a href="struct_kernel_merge_sort.html">KernelMergeSort</a></div><div class="ttdef"><b>Definition</b> sort.h:221</div></div>
|
||||
<div class="ttc" id="astruct_kernel_merge_sort_html_a0df65b709ae7f153a2bf381179d55e00"><div class="ttname"><a href="struct_kernel_merge_sort.html#a0df65b709ae7f153a2bf381179d55e00">KernelMergeSort::idx_t</a></div><div class="ttdeci">uint idx_t</div><div class="ttdef"><b>Definition</b> sort.h:223</div></div>
|
||||
<div class="ttc" id="astruct_kernel_merge_sort_html_a4e3f09896275956fc4c23e1f157dca3b"><div class="ttname"><a href="struct_kernel_merge_sort.html#a4e3f09896275956fc4c23e1f157dca3b">KernelMergeSort::val_t</a></div><div class="ttdeci">T val_t</div><div class="ttdef"><b>Definition</b> sort.h:222</div></div>
|
||||
<div class="ttc" id="astruct_kernel_merge_sort_html_a56b644ec66f7fb5c01b280f124304be9"><div class="ttname"><a href="struct_kernel_merge_sort.html#a56b644ec66f7fb5c01b280f124304be9">KernelMergeSort::block_sort</a></div><div class="ttdeci">static METAL_FUNC void block_sort(const device T *inp, device U *out, const constant int &size_sorted_axis, const constant int &in_stride_sorted_axis, const constant int &out_stride_sorted_axis, const constant int &in_stride_segment_axis, const constant int &out_stride_segment_axis, threadgroup val_t *tgp_vals, threadgroup idx_t *tgp_idxs, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:234</div></div>
|
||||
<div class="ttc" id="astruct_kernel_merge_sort_html_a959aaf5bfb70796a525fed318f7ae8ab"><div class="ttname"><a href="struct_kernel_merge_sort.html#a959aaf5bfb70796a525fed318f7ae8ab">KernelMergeSort::N_PER_BLOCK</a></div><div class="ttdeci">static constant constexpr const short N_PER_BLOCK</div><div class="ttdef"><b>Definition</b> sort.h:232</div></div>
|
||||
<div class="ttc" id="astruct_kernel_merge_sort_html_adae7850e057fc30d5328c7b3dcc998fa"><div class="ttname"><a href="struct_kernel_merge_sort.html#adae7850e057fc30d5328c7b3dcc998fa">KernelMergeSort::block_merge_sort_t</a></div><div class="ttdeci">BlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp > block_merge_sort_t</div><div class="ttdef"><b>Definition</b> sort.h:224</div></div>
|
||||
<div class="ttc" id="astruct_kernel_multi_block_merge_sort_html"><div class="ttname"><a href="struct_kernel_multi_block_merge_sort.html">KernelMultiBlockMergeSort</a></div><div class="ttdef"><b>Definition</b> sort.h:399</div></div>
|
||||
<div class="ttc" id="astruct_kernel_multi_block_merge_sort_html_a322ed2eac315a561e0fd90af2fd577eb"><div class="ttname"><a href="struct_kernel_multi_block_merge_sort.html#a322ed2eac315a561e0fd90af2fd577eb">KernelMultiBlockMergeSort::block_sort</a></div><div class="ttdeci">static METAL_FUNC void block_sort(const device val_t *inp, device val_t *out_vals, device idx_t *out_idxs, const constant int &size_sorted_axis, const constant int &stride_sorted_axis, threadgroup val_t *tgp_vals, threadgroup idx_t *tgp_idxs, uint3 tid, uint3 lid)</div><div class="ttdef"><b>Definition</b> sort.h:410</div></div>
|
||||
<div class="ttc" id="astruct_kernel_multi_block_merge_sort_html_ab15895b4233aba0e279cc44a07a201fe"><div class="ttname"><a href="struct_kernel_multi_block_merge_sort.html#ab15895b4233aba0e279cc44a07a201fe">KernelMultiBlockMergeSort::merge_partition</a></div><div class="ttdeci">static METAL_FUNC int merge_partition(const device val_t *As, const device val_t *Bs, int A_sz, int B_sz, int sort_md)</div><div class="ttdef"><b>Definition</b> sort.h:448</div></div>
|
||||
<div class="ttc" id="astruct_kernel_multi_block_merge_sort_html_ae5113ca5852d11999ae932439af95a5c"><div class="ttname"><a href="struct_kernel_multi_block_merge_sort.html#ae5113ca5852d11999ae932439af95a5c">KernelMultiBlockMergeSort::N_PER_BLOCK</a></div><div class="ttdeci">static constant constexpr const short N_PER_BLOCK</div><div class="ttdef"><b>Definition</b> sort.h:408</div></div>
|
||||
<div class="ttc" id="astruct_kernel_multi_block_merge_sort_html_af27e9af4b58640c0aa620bc4efc68dff"><div class="ttname"><a href="struct_kernel_multi_block_merge_sort.html#af27e9af4b58640c0aa620bc4efc68dff">KernelMultiBlockMergeSort::block_merge_sort_t</a></div><div class="ttdeci">BlockMergeSort< val_t, idx_t, ARG_SORT, BLOCK_THREADS, N_PER_THREAD, CompareOp > block_merge_sort_t</div><div class="ttdef"><b>Definition</b> sort.h:400</div></div>
|
||||
<div class="ttc" id="astruct_less_than_html"><div class="ttname"><a href="struct_less_than.html">LessThan</a></div><div class="ttdef"><b>Definition</b> sort.h:23</div></div>
|
||||
<div class="ttc" id="astruct_less_than_html_a2798eb377b411c93a4ed30cf35caade2"><div class="ttname"><a href="struct_less_than.html#a2798eb377b411c93a4ed30cf35caade2">LessThan::operator()</a></div><div class="ttdeci">METAL_FUNC bool operator()(T a, T b)</div><div class="ttdef"><b>Definition</b> sort.h:26</div></div>
|
||||
<div class="ttc" id="astruct_less_than_html_abf97a6b0163048e4ba96460939dbd3a3"><div class="ttname"><a href="struct_less_than.html#abf97a6b0163048e4ba96460939dbd3a3">LessThan::init</a></div><div class="ttdeci">static constexpr constant T init</div><div class="ttdef"><b>Definition</b> sort.h:24</div></div>
|
||||
<div class="ttc" id="astruct_limits_html"><div class="ttname"><a href="struct_limits.html">Limits</a></div><div class="ttdef"><b>Definition</b> utils.h:23</div></div>
|
||||
<div class="ttc" id="astruct_limits_html_a2f0673b6f9da89ce1d64f9f3d74f50a8"><div class="ttname"><a href="struct_limits.html#a2f0673b6f9da89ce1d64f9f3d74f50a8">Limits::max</a></div><div class="ttdeci">static const constant U max</div><div class="ttdef"><b>Definition</b> utils.h:24</div></div>
|
||||
<div class="ttc" id="astruct_thread_sort_html"><div class="ttname"><a href="struct_thread_sort.html">ThreadSort</a></div><div class="ttdef"><b>Definition</b> sort.h:37</div></div>
|
||||
<div class="ttc" id="astruct_thread_sort_html_ad9ab3e6b47f7e9b91c0f3b773596986d"><div class="ttname"><a href="struct_thread_sort.html#ad9ab3e6b47f7e9b91c0f3b773596986d">ThreadSort::sort</a></div><div class="ttdeci">static METAL_FUNC void sort(thread val_t(&vals)[N_PER_THREAD], thread idx_t(&idxs)[N_PER_THREAD])</div><div class="ttdef"><b>Definition</b> sort.h:38</div></div>
|
||||
</div><!-- fragment --></div><!-- contents -->
|
||||
<!-- start footer part -->
|
||||
<hr class="footer"/><address class="footer"><small>
|
||||
Generated by <a href="https://www.doxygen.org/index.html"><img class="footer" src="doxygen.svg" width="104" height="31" alt="doxygen"/></a> 1.12.0
|
||||
</small></address>
|
||||
</div><!-- doc-content -->
|
||||
<!-- start footer part -->
|
||||
<div id="nav-path" class="navpath"><!-- id is needed for treeview function! -->
|
||||
<ul>
|
||||
<li class="navelem"><a class="el" href="dir_938ab0ecf10b8b860ff766c820f665fd.html">mlx</a></li><li class="navelem"><a class="el" href="dir_1d446c9bd3c99228254c9484e0bc5c06.html">backend</a></li><li class="navelem"><a class="el" href="dir_d0c977ea65824390717cdb7efc36c157.html">metal</a></li><li class="navelem"><a class="el" href="dir_70a37effa88bcbd6b791977fa1e64356.html">kernels</a></li><li class="navelem"><a class="el" href="sort_8h.html">sort.h</a></li>
|
||||
<li class="footer">Generated by <a href="https://www.doxygen.org/index.html"><img class="footer" src="doxygen.svg" width="104" height="31" alt="doxygen"/></a> 1.13.1 </li>
|
||||
</ul>
|
||||
</div>
|
||||
</body>
|
||||
</html>
|
||||
|
Reference in New Issue
Block a user