zhangyiss/mlx
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx.git synced 2025-11-04 02:28:13 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
Files
fdbf300746827016f535de089991f6545cd961bc
mlx/docs/build/html/kernels_2gemv__masked_8h_source.html
CircleCI Docs fdbf300746 rebase
2025-10-17 19:14:59 +00:00

982 lines
153 KiB
HTML
Raw Blame History

<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "https://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" lang="en-US">
<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="viewport" content="width=device-width, initial-scale=1"/>
<title>MLX: mlx/backend/metal/kernels/gemv_masked.h Source File</title>
<link href="tabs.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="clipboard.js"></script>
<link href="navtree.css" rel="stylesheet" type="text/css"/>
<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>
<link href="doxygen.css" rel="stylesheet" type="text/css" />
</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 id="projectrow">
<td id="projectalign">
<div id="projectname">MLX
</div>
</td>
</tr>
</tbody>
</table>
</div>
<!-- end header part -->
<!-- Generated by Doxygen 1.12.0 -->
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:d3d9a9a6595521f9666a5e94cc830dab83b65699&amp;dn=expat.txt MIT */
var searchBox = new SearchBox("searchBox", "search/",'.html');
/* @license-end */
</script>
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:d3d9a9a6595521f9666a5e94cc830dab83b65699&amp;dn=expat.txt MIT */
$(function() { codefold.init(0); });
/* @license-end */
</script>
<script type="text/javascript" src="menudata.js"></script>
<script type="text/javascript" src="menu.js"></script>
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:d3d9a9a6595521f9666a5e94cc830dab83b65699&amp;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&amp;dn=expat.txt MIT */
$(function(){ initResizable(false); });
/* @license-end */
</script>
<!-- window showing the filter options -->
<div id="MSearchSelectWindow"
onmouseover="return searchBox.OnSearchSelectShow()"
onmouseout="return searchBox.OnSearchSelectHide()"
onkeydown="return searchBox.OnSearchSelectKey(event)">
</div>
<!-- iframe showing the search results (closed by default) -->
<div id="MSearchResultsWindow">
<div id="MSearchResults">
<div class="SRPage">
<div id="SRIndex">
<div id="SRResults"></div>
<div class="SRStatus" id="Loading">Loading...</div>
<div class="SRStatus" id="Searching">Searching...</div>
<div class="SRStatus" id="NoMatches">No Matches</div>
</div>
</div>
</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">gemv_masked.h</div></div>
</div><!--header-->
<div class="contents">
<a href="kernels_2gemv__masked_8h.html">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a id="l00001" name="l00001"></a><span class="lineno"> 1</span><span class="comment">// Copyright © 2023-2024 Apple Inc.</span></div>
<div class="line"><a id="l00002" name="l00002"></a><span class="lineno"> 2</span> </div>
<div class="line"><a id="l00003" name="l00003"></a><span class="lineno"> 3</span><span class="preprocessor">#include &quot;<a class="code" href="backend_2metal_2kernels_2steel_2utils_8h.html">mlx/backend/metal/kernels/steel/utils.h</a>&quot;</span></div>
<div class="line"><a id="l00004" name="l00004"></a><span class="lineno"> 4</span> </div>
<div class="line"><a id="l00005" name="l00005"></a><span class="lineno"> 5</span><span class="keyword">using namespace </span><a class="code hl_namespace" href="namespacemetal.html">metal</a>;</div>
<div class="line"><a id="l00006" name="l00006"></a><span class="lineno"> 6</span> </div>
<div class="line"><a id="l00007" name="l00007"></a><span class="lineno"><a class="line" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd"> 7</a></span><span class="preprocessor">#define MLX_MTL_CONST static constant constexpr const</span></div>
<div class="line"><a id="l00008" name="l00008"></a><span class="lineno"><a class="line" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3"> 8</a></span><span class="preprocessor">#define MLX_MTL_PRAGMA_UNROLL _Pragma(&quot;clang loop unroll(full)&quot;)</span></div>
<div class="line"><a id="l00009" name="l00009"></a><span class="lineno"> 9</span> </div>
<div class="foldopen" id="foldopen00010" data-start="{" data-end="};">
<div class="line"><a id="l00010" name="l00010"></a><span class="lineno"><a class="line" href="struct___no_mask.html"> 10</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct___no_mask.html">_NoMask</a> {</div>
<div class="line"><a id="l00011" name="l00011"></a><span class="lineno"><a class="line" href="struct___no_mask.html#a0c4a4557d5c97ceafe3a2c4e521cdf7e"> 11</a></span> <span class="keywordtype">char</span> <a class="code hl_variable" href="struct___no_mask.html#a0c4a4557d5c97ceafe3a2c4e521cdf7e">x</a>;</div>
<div class="line"><a id="l00012" name="l00012"></a><span class="lineno"> 12</span> </div>
<div class="foldopen" id="foldopen00013" data-start="{" data-end="}">
<div class="line"><a id="l00013" name="l00013"></a><span class="lineno"><a class="line" href="struct___no_mask.html#ad3723c1e70e46beefd283ce6317416cb"> 13</a></span> <span class="keyword">constexpr</span> METAL_FUNC <span class="keyword">operator</span> bool() {</div>
<div class="line"><a id="l00014" name="l00014"></a><span class="lineno"> 14</span> <span class="keywordflow">return</span> <span class="keyword">true</span>;</div>
<div class="line"><a id="l00015" name="l00015"></a><span class="lineno"> 15</span> }</div>
</div>
<div class="foldopen" id="foldopen00016" data-start="{" data-end="}">
<div class="line"><a id="l00016" name="l00016"></a><span class="lineno"><a class="line" href="struct___no_mask.html#aafbf8a3201e1cc1abf74dd1f1b7272cd"> 16</a></span> <span class="keyword">constexpr</span> METAL_FUNC <span class="keyword">operator</span> bool() const threadgroup {</div>
<div class="line"><a id="l00017" name="l00017"></a><span class="lineno"> 17</span> <span class="keywordflow">return</span> <span class="keyword">true</span>;</div>
<div class="line"><a id="l00018" name="l00018"></a><span class="lineno"> 18</span> }</div>
</div>
<div class="foldopen" id="foldopen00019" data-start="{" data-end="}">
<div class="line"><a id="l00019" name="l00019"></a><span class="lineno"><a class="line" href="struct___no_mask.html#a73e9612a619885cbc97cbd8f40df71e7"> 19</a></span> <span class="keyword">constexpr</span> METAL_FUNC <span class="keyword">operator</span> bool() const device {</div>
<div class="line"><a id="l00020" name="l00020"></a><span class="lineno"> 20</span> <span class="keywordflow">return</span> <span class="keyword">true</span>;</div>
<div class="line"><a id="l00021" name="l00021"></a><span class="lineno"> 21</span> }</div>
</div>
<div class="foldopen" id="foldopen00022" data-start="{" data-end="}">
<div class="line"><a id="l00022" name="l00022"></a><span class="lineno"><a class="line" href="struct___no_mask.html#a4bf336d472bc677028250f76b9cdc08c"> 22</a></span> <span class="keyword">constexpr</span> METAL_FUNC <span class="keyword">operator</span> bool() const constant {</div>
<div class="line"><a id="l00023" name="l00023"></a><span class="lineno"> 23</span> <span class="keywordflow">return</span> <span class="keyword">true</span>;</div>
<div class="line"><a id="l00024" name="l00024"></a><span class="lineno"> 24</span> }</div>
</div>
<div class="line"><a id="l00025" name="l00025"></a><span class="lineno"> 25</span>};</div>
</div>
<div class="line"><a id="l00026" name="l00026"></a><span class="lineno"> 26</span> </div>
<div class="line"><a id="l00027" name="l00027"></a><span class="lineno"><a class="line" href="kernels_2gemv__masked_8h.html#a1480c8cdff1cae1462a5a71632969bca"> 27</a></span><span class="keyword">typedef</span> <span class="keyword">struct </span><a class="code hl_struct" href="struct___no_mask.html">_NoMask</a> <a class="code hl_struct" href="struct___no_mask.html">nomask_t</a>;</div>
<div class="line"><a id="l00028" name="l00028"></a><span class="lineno"> 28</span> </div>
<div class="line"><a id="l00029" name="l00029"></a><span class="lineno"> 29</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> OutT, <span class="keyword">typename</span> InT = OutT&gt;</div>
<div class="foldopen" id="foldopen00030" data-start="{" data-end="};">
<div class="line"><a id="l00030" name="l00030"></a><span class="lineno"><a class="line" href="struct_scale_op.html"> 30</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct_scale_op.html">ScaleOp</a> {</div>
<div class="line"><a id="l00031" name="l00031"></a><span class="lineno"><a class="line" href="struct_scale_op.html#a02043fac21c68fb8d6863a01f45ede4b"> 31</a></span> OutT <a class="code hl_variable" href="struct_scale_op.html#a02043fac21c68fb8d6863a01f45ede4b">scale</a>;</div>
<div class="line"><a id="l00032" name="l00032"></a><span class="lineno"> 32</span> </div>
<div class="foldopen" id="foldopen00033" data-start="{" data-end="}">
<div class="line"><a id="l00033" name="l00033"></a><span class="lineno"><a class="line" href="struct_scale_op.html#a69f82bc925843a4e1c14dfe8ad2f3218"> 33</a></span> METAL_FUNC OutT <a class="code hl_function" href="struct_scale_op.html#a69f82bc925843a4e1c14dfe8ad2f3218">apply</a>(InT x)<span class="keyword"> const </span>{</div>
<div class="line"><a id="l00034" name="l00034"></a><span class="lineno"> 34</span> <span class="keywordflow">return</span> <span class="keyword">static_cast&lt;</span>OutT<span class="keyword">&gt;</span>(x) * <a class="code hl_variable" href="struct_scale_op.html#a02043fac21c68fb8d6863a01f45ede4b">scale</a>;</div>
<div class="line"><a id="l00035" name="l00035"></a><span class="lineno"> 35</span> }</div>
</div>
<div class="line"><a id="l00036" name="l00036"></a><span class="lineno"> 36</span>};</div>
</div>
<div class="line"><a id="l00037" name="l00037"></a><span class="lineno"> 37</span> </div>
<div class="line"><a id="l00038" name="l00038"></a><span class="lineno"> 38</span><span class="keyword">template</span> &lt;</div>
<div class="line"><a id="l00039" name="l00039"></a><span class="lineno"> 39</span> <span class="keyword">typename</span> T,</div>
<div class="line"><a id="l00040" name="l00040"></a><span class="lineno"> 40</span> <span class="keyword">typename</span> out_mask_t,</div>
<div class="line"><a id="l00041" name="l00041"></a><span class="lineno"> 41</span> <span class="keyword">typename</span> op_mask_t,</div>
<div class="line"><a id="l00042" name="l00042"></a><span class="lineno"> 42</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BM, <span class="comment">/* Threadgroup rows (in simdgroups) */</span></div>
<div class="line"><a id="l00043" name="l00043"></a><span class="lineno"> 43</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BN, <span class="comment">/* Threadgroup cols (in simdgroups) */</span></div>
<div class="line"><a id="l00044" name="l00044"></a><span class="lineno"> 44</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SM, <span class="comment">/* Simdgroup rows (in threads) */</span></div>
<div class="line"><a id="l00045" name="l00045"></a><span class="lineno"> 45</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SN, <span class="comment">/* Simdgroup cols (in threads) */</span></div>
<div class="line"><a id="l00046" name="l00046"></a><span class="lineno"> 46</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TM, <span class="comment">/* Thread rows (in elements) */</span></div>
<div class="line"><a id="l00047" name="l00047"></a><span class="lineno"> 47</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TN&gt; <span class="comment">/* Thread cols (in elements) */</span></div>
<div class="foldopen" id="foldopen00048" data-start="{" data-end="};">
<div class="line"><a id="l00048" name="l00048"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html"> 48</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct_g_e_m_v_kernel.html">GEMVKernel</a> {</div>
<div class="line"><a id="l00049" name="l00049"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a1dd943fcbf5e7be435fc36bed589a641"> 49</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a1dd943fcbf5e7be435fc36bed589a641">threadsM</a> = BM * SM;</div>
<div class="line"><a id="l00050" name="l00050"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a47bfab7d21dd18760d3e0937ad36b19d"> 50</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a47bfab7d21dd18760d3e0937ad36b19d">threadsN</a> = BN * SN;</div>
<div class="line"><a id="l00051" name="l00051"></a><span class="lineno"> 51</span> </div>
<div class="line"><a id="l00052" name="l00052"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c"> 52</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">blockM</a> = <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a1dd943fcbf5e7be435fc36bed589a641">threadsM</a> * TM;</div>
<div class="line"><a id="l00053" name="l00053"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b"> 53</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b">blockN</a> = <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a47bfab7d21dd18760d3e0937ad36b19d">threadsN</a> * TN;</div>
<div class="line"><a id="l00054" name="l00054"></a><span class="lineno"> 54</span> </div>
<div class="line"><a id="l00055" name="l00055"></a><span class="lineno"> 55</span> <span class="keyword">static_assert</span>(SM * SN == 32, <span class="stringliteral">&quot;simdgroup can only have 32 threads&quot;</span>);</div>
<div class="line"><a id="l00056" name="l00056"></a><span class="lineno"> 56</span> </div>
<div class="line"><a id="l00057" name="l00057"></a><span class="lineno"> 57</span> <span class="keyword">static_assert</span>(</div>
<div class="line"><a id="l00058" name="l00058"></a><span class="lineno"> 58</span> SN == 8 || SN == 16 || SN == 32,</div>
<div class="line"><a id="l00059" name="l00059"></a><span class="lineno"> 59</span> <span class="stringliteral">&quot;gemv block must have a width of 8, 16, or 32&quot;</span>);</div>
<div class="line"><a id="l00060" name="l00060"></a><span class="lineno"> 60</span> </div>
<div class="line"><a id="l00061" name="l00061"></a><span class="lineno"> 61</span> <span class="keyword">static_assert</span>(<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b">blockN</a> &gt;= <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">blockM</a>, <span class="stringliteral">&quot;Masked gemv must have blockN &gt;= blockM&quot;</span>);</div>
<div class="line"><a id="l00062" name="l00062"></a><span class="lineno"> 62</span> </div>
<div class="line"><a id="l00063" name="l00063"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e"> 63</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> = !metal::is_same_v&lt;op_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00064" name="l00064"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#ab8b64c94f4c8f6f09c0777415589b487"> 64</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab8b64c94f4c8f6f09c0777415589b487">has_output_mask</a> = !metal::is_same_v&lt;out_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00065" name="l00065"></a><span class="lineno"> 65</span> </div>
<div class="line"><a id="l00066" name="l00066"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#ad47223ee49b3cb7bf3746a2cec45f883"> 66</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ad47223ee49b3cb7bf3746a2cec45f883">has_mul_operand_mask</a> =</div>
<div class="line"><a id="l00067" name="l00067"></a><span class="lineno"> 67</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> &amp;&amp; !metal::is_same_v&lt;op_mask_t, bool&gt;;</div>
<div class="line"><a id="l00068" name="l00068"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a0edbf2dd6a6563e7afa6dab6b670615c"> 68</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a0edbf2dd6a6563e7afa6dab6b670615c">has_mul_output_mask</a> =</div>
<div class="line"><a id="l00069" name="l00069"></a><span class="lineno"> 69</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab8b64c94f4c8f6f09c0777415589b487">has_output_mask</a> &amp;&amp; !metal::is_same_v&lt;out_mask_t, bool&gt;;</div>
<div class="line"><a id="l00070" name="l00070"></a><span class="lineno"> 70</span> </div>
<div class="line"><a id="l00071" name="l00071"></a><span class="lineno"> 71</span> <span class="comment">// - The matrix of size (M = out_vec_size, K = in_vec_size) is divided up</span></div>
<div class="line"><a id="l00072" name="l00072"></a><span class="lineno"> 72</span> <span class="comment">// into blocks of (blockM, blockN) divided among threadgroups</span></div>
<div class="line"><a id="l00073" name="l00073"></a><span class="lineno"> 73</span> <span class="comment">// - Every thread works on a block of (TM, TN)</span></div>
<div class="line"><a id="l00074" name="l00074"></a><span class="lineno"> 74</span> <span class="comment">// - We assume each threadgroup has (threadsN, threadsM, 1) threads</span></div>
<div class="line"><a id="l00075" name="l00075"></a><span class="lineno"> 75</span> <span class="comment">//</span></div>
<div class="line"><a id="l00076" name="l00076"></a><span class="lineno"> 76</span> <span class="comment">// 1. A thread loads TN elements each from mat along TM rows</span></div>
<div class="line"><a id="l00077" name="l00077"></a><span class="lineno"> 77</span> <span class="comment">// and the corresponding scalar from the vector</span></div>
<div class="line"><a id="l00078" name="l00078"></a><span class="lineno"> 78</span> <span class="comment">// 2. The thread then multiplies and adds to accumulate its local result for</span></div>
<div class="line"><a id="l00079" name="l00079"></a><span class="lineno"> 79</span> <span class="comment">// the block</span></div>
<div class="line"><a id="l00080" name="l00080"></a><span class="lineno"> 80</span> <span class="comment">// 3. At the end, each thread has accumulated results over all blocks across</span></div>
<div class="line"><a id="l00081" name="l00081"></a><span class="lineno"> 81</span> <span class="comment">// the rows. These are then summed up across the threadgroup</span></div>
<div class="line"><a id="l00082" name="l00082"></a><span class="lineno"> 82</span> <span class="comment">// 4. Each threadgroup writes its accumulated blockM outputs</span></div>
<div class="line"><a id="l00083" name="l00083"></a><span class="lineno"> 83</span> <span class="comment">//</span></div>
<div class="line"><a id="l00084" name="l00084"></a><span class="lineno"> 84</span> <span class="comment">// Edge case handling:</span></div>
<div class="line"><a id="l00085" name="l00085"></a><span class="lineno"> 85</span> <span class="comment">// - The threadgroup with the largest tid has blocks that exceed the matrix</span></div>
<div class="line"><a id="l00086" name="l00086"></a><span class="lineno"> 86</span> <span class="comment">// * The blocks that start outside the matrix are never read (thread results</span></div>
<div class="line"><a id="l00087" name="l00087"></a><span class="lineno"> 87</span> <span class="comment">// remain zero)</span></div>
<div class="line"><a id="l00088" name="l00088"></a><span class="lineno"> 88</span> <span class="comment">// * The last thread that partially overlaps with the matrix is shifted</span></div>
<div class="line"><a id="l00089" name="l00089"></a><span class="lineno"> 89</span> <span class="comment">// inwards such that the thread block fits exactly in the matrix</span></div>
<div class="line"><a id="l00090" name="l00090"></a><span class="lineno"> 90</span> </div>
<div class="line"><a id="l00091" name="l00091"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a9ef4d0e62094d7033069f5dda5efb236"> 91</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">short</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a9ef4d0e62094d7033069f5dda5efb236">tgp_mem_size</a> = BN &gt; 1 ? BN*(<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">blockM</a> + TM) : 0;</div>
<div class="line"><a id="l00092" name="l00092"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#ae8113fddf6fb637acfd12efd978b704c"> 92</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ae8113fddf6fb637acfd12efd978b704c">needs_tgp_reduction</a> = BN &gt; 1;</div>
<div class="line"><a id="l00093" name="l00093"></a><span class="lineno"> 93</span> </div>
<div class="line"><a id="l00094" name="l00094"></a><span class="lineno"> 94</span> <span class="keyword">static</span> METAL_FUNC <span class="keywordtype">void</span></div>
<div class="foldopen" id="foldopen00095" data-start="{" data-end="}">
<div class="line"><a id="l00095" name="l00095"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a6013e9c5b2f72fa1311dd038172df0ce"> 95</a></span> <a class="code hl_function" href="struct_g_e_m_v_kernel.html#a6013e9c5b2f72fa1311dd038172df0ce">load_unsafe</a>(<span class="keyword">const</span> device T* src, thread T dst[TN], <span class="keyword">const</span> <span class="keywordtype">int</span> src_offset = 0) {</div>
<div class="line"><a id="l00096" name="l00096"></a><span class="lineno"> 96</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00097" name="l00097"></a><span class="lineno"> 97</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00098" name="l00098"></a><span class="lineno"> 98</span> dst[tn] = src[src_offset + tn];</div>
<div class="line"><a id="l00099" name="l00099"></a><span class="lineno"> 99</span> }</div>
<div class="line"><a id="l00100" name="l00100"></a><span class="lineno"> 100</span> }</div>
</div>
<div class="line"><a id="l00101" name="l00101"></a><span class="lineno"> 101</span> </div>
<div class="foldopen" id="foldopen00102" data-start="{" data-end="}">
<div class="line"><a id="l00102" name="l00102"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#a04bb72da9a93d6d1eba468fa311bbba7"> 102</a></span> <span class="keyword">static</span> METAL_FUNC <span class="keywordtype">void</span> <a class="code hl_function" href="struct_g_e_m_v_kernel.html#a04bb72da9a93d6d1eba468fa311bbba7">load_safe</a>(</div>
<div class="line"><a id="l00103" name="l00103"></a><span class="lineno"> 103</span> <span class="keyword">const</span> device T* src,</div>
<div class="line"><a id="l00104" name="l00104"></a><span class="lineno"> 104</span> thread T dst[TN],</div>
<div class="line"><a id="l00105" name="l00105"></a><span class="lineno"> 105</span> <span class="keyword">const</span> <span class="keywordtype">int</span> src_offset = 0,</div>
<div class="line"><a id="l00106" name="l00106"></a><span class="lineno"> 106</span> <span class="keyword">const</span> <span class="keywordtype">int</span> src_size = TN) {</div>
<div class="line"><a id="l00107" name="l00107"></a><span class="lineno"> 107</span> <span class="keywordflow">if</span> (src_offset + TN &lt;= src_size) {</div>
<div class="line"><a id="l00108" name="l00108"></a><span class="lineno"> 108</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></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> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00110" name="l00110"></a><span class="lineno"> 110</span> dst[tn] = src[src_offset + tn];</div>
<div class="line"><a id="l00111" name="l00111"></a><span class="lineno"> 111</span> }</div>
<div class="line"><a id="l00112" name="l00112"></a><span class="lineno"> 112</span> } <span class="keywordflow">else</span> { <span class="comment">// Edgecase</span></div>
<div class="line"><a id="l00113" name="l00113"></a><span class="lineno"> 113</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00114" name="l00114"></a><span class="lineno"> 114</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00115" name="l00115"></a><span class="lineno"> 115</span> dst[tn] = src_offset + tn &lt; src_size ? src[src_offset + tn] : 0;</div>
<div class="line"><a id="l00116" name="l00116"></a><span class="lineno"> 116</span> }</div>
<div class="line"><a id="l00117" name="l00117"></a><span class="lineno"> 117</span> }</div>
<div class="line"><a id="l00118" name="l00118"></a><span class="lineno"> 118</span> }</div>
</div>
<div class="line"><a id="l00119" name="l00119"></a><span class="lineno"> 119</span> </div>
<div class="foldopen" id="foldopen00120" data-start="{" data-end="}">
<div class="line"><a id="l00120" name="l00120"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_kernel.html#ac4a7b5011a0ea938ab1949bb1767fc1a"> 120</a></span> <span class="keyword">static</span> METAL_FUNC <span class="keywordtype">void</span> <a class="code hl_function" href="struct_g_e_m_v_kernel.html#ac4a7b5011a0ea938ab1949bb1767fc1a">run</a>(</div>
<div class="line"><a id="l00121" name="l00121"></a><span class="lineno"> 121</span> <span class="keyword">const</span> device T* mat [[buffer(0)]],</div>
<div class="line"><a id="l00122" name="l00122"></a><span class="lineno"> 122</span> <span class="keyword">const</span> device T* in_vec [[buffer(1)]],</div>
<div class="line"><a id="l00123" name="l00123"></a><span class="lineno"> 123</span> device T* out_vec [[buffer(3)]],</div>
<div class="line"><a id="l00124" name="l00124"></a><span class="lineno"> 124</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; in_vec_size [[buffer(4)]],</div>
<div class="line"><a id="l00125" name="l00125"></a><span class="lineno"> 125</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; out_vec_size [[buffer(5)]],</div>
<div class="line"><a id="l00126" name="l00126"></a><span class="lineno"> 126</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; matrix_ld [[buffer(6)]],</div>
<div class="line"><a id="l00127" name="l00127"></a><span class="lineno"> 127</span> <span class="keyword">const</span> device out_mask_t* out_mask [[buffer(20)]],</div>
<div class="line"><a id="l00128" name="l00128"></a><span class="lineno"> 128</span> <span class="keyword">const</span> device op_mask_t* mat_mask [[buffer(21)]],</div>
<div class="line"><a id="l00129" name="l00129"></a><span class="lineno"> 129</span> <span class="keyword">const</span> device op_mask_t* vec_mask [[buffer(22)]],</div>
<div class="line"><a id="l00130" name="l00130"></a><span class="lineno"> 130</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* mask_strides [[buffer(23)]],</div>
<div class="line"><a id="l00131" name="l00131"></a><span class="lineno"> 131</span> threadgroup T* tgp_memory [[threadgroup(0)]],</div>
<div class="line"><a id="l00132" name="l00132"></a><span class="lineno"> 132</span> uint3 tid [[threadgroup_position_in_grid]],</div>
<div class="line"><a id="l00133" name="l00133"></a><span class="lineno"> 133</span> uint3 lid [[thread_position_in_threadgroup]],</div>
<div class="line"><a id="l00134" name="l00134"></a><span class="lineno"> 134</span> uint simd_gid [[simdgroup_index_in_threadgroup]],</div>
<div class="line"><a id="l00135" name="l00135"></a><span class="lineno"> 135</span> uint simd_lid [[thread_index_in_simdgroup]]) {</div>
<div class="line"><a id="l00136" name="l00136"></a><span class="lineno"> 136</span> <span class="comment">// Appease compiler</span></div>
<div class="line"><a id="l00137" name="l00137"></a><span class="lineno"> 137</span> (void)lid;</div>
<div class="line"><a id="l00138" name="l00138"></a><span class="lineno"> 138</span> </div>
<div class="line"><a id="l00139" name="l00139"></a><span class="lineno"> 139</span> <span class="comment">// Thread local accumulation results</span></div>
<div class="line"><a id="l00140" name="l00140"></a><span class="lineno"> 140</span> thread T result[TM] = {0};</div>
<div class="line"><a id="l00141" name="l00141"></a><span class="lineno"> 141</span> thread T inter[TN];</div>
<div class="line"><a id="l00142" name="l00142"></a><span class="lineno"> 142</span> thread T v_coeff[TN];</div>
<div class="line"><a id="l00143" name="l00143"></a><span class="lineno"> 143</span> </div>
<div class="line"><a id="l00144" name="l00144"></a><span class="lineno"> 144</span> <span class="keyword">const</span> <span class="keywordtype">int</span> thrM = SN != 32 ? simd_lid / SN : 0;</div>
<div class="line"><a id="l00145" name="l00145"></a><span class="lineno"> 145</span> <span class="keyword">const</span> <span class="keywordtype">int</span> thrN = SN != 32 ? simd_lid % SN : int(simd_lid);</div>
<div class="line"><a id="l00146" name="l00146"></a><span class="lineno"> 146</span> </div>
<div class="line"><a id="l00147" name="l00147"></a><span class="lineno"> 147</span> <span class="keyword">const</span> <span class="keywordtype">int</span> sgN = BN != 1 ? (simd_gid % BN) : 0;</div>
<div class="line"><a id="l00148" name="l00148"></a><span class="lineno"> 148</span> </div>
<div class="line"><a id="l00149" name="l00149"></a><span class="lineno"> 149</span> <span class="keyword">const</span> <span class="keywordtype">int</span> simdM = BN != 1 ? SM * (simd_gid / BN) : int(SM * simd_gid);</div>
<div class="line"><a id="l00150" name="l00150"></a><span class="lineno"> 150</span> <span class="keyword">const</span> <span class="keywordtype">int</span> simdN = BN != 1 ? SN * (simd_gid % BN) : 0;</div>
<div class="line"><a id="l00151" name="l00151"></a><span class="lineno"> 151</span> </div>
<div class="line"><a id="l00152" name="l00152"></a><span class="lineno"> 152</span> <span class="keywordtype">int</span> bm = (simdM + thrM) * TM;</div>
<div class="line"><a id="l00153" name="l00153"></a><span class="lineno"> 153</span> <span class="keywordtype">int</span> bn = (simdN + thrN) * TN;</div>
<div class="line"><a id="l00154" name="l00154"></a><span class="lineno"> 154</span> </div>
<div class="line"><a id="l00155" name="l00155"></a><span class="lineno"> 155</span> <span class="comment">// Block position</span></div>
<div class="line"><a id="l00156" name="l00156"></a><span class="lineno"> 156</span> <span class="keywordtype">int</span> out_row = tid.x * <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">blockM</a> + bm;</div>
<div class="line"><a id="l00157" name="l00157"></a><span class="lineno"> 157</span> </div>
<div class="line"><a id="l00158" name="l00158"></a><span class="lineno"> 158</span> <span class="comment">// Exit simdgroup if rows out of bound</span></div>
<div class="line"><a id="l00159" name="l00159"></a><span class="lineno"> 159</span> <span class="keywordflow">if</span> (out_row &gt;= out_vec_size)</div>
<div class="line"><a id="l00160" name="l00160"></a><span class="lineno"> 160</span> <span class="keywordflow">return</span>;</div>
<div class="line"><a id="l00161" name="l00161"></a><span class="lineno"> 161</span> </div>
<div class="line"><a id="l00162" name="l00162"></a><span class="lineno"> 162</span> <span class="comment">// Adjust tail simdgroup to ensure in bound reads</span></div>
<div class="line"><a id="l00163" name="l00163"></a><span class="lineno"> 163</span> out_row = out_row + TM &lt;= out_vec_size ? out_row : out_vec_size - TM;</div>
<div class="line"><a id="l00164" name="l00164"></a><span class="lineno"> 164</span> </div>
<div class="line"><a id="l00165" name="l00165"></a><span class="lineno"> 165</span> <span class="comment">// Prepare mask offsets</span></div>
<div class="line"><a id="l00166" name="l00166"></a><span class="lineno"> 166</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* out_mask_strides = mask_strides;</div>
<div class="line"><a id="l00167" name="l00167"></a><span class="lineno"> 167</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* mat_mask_strides =</div>
<div class="line"><a id="l00168" name="l00168"></a><span class="lineno"> 168</span> mask_strides + (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab8b64c94f4c8f6f09c0777415589b487">has_output_mask</a> ? 2 : 0);</div>
<div class="line"><a id="l00169" name="l00169"></a><span class="lineno"> 169</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* vec_mask_strides =</div>
<div class="line"><a id="l00170" name="l00170"></a><span class="lineno"> 170</span> mat_mask_strides + (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> ? 2 : 0);</div>
<div class="line"><a id="l00171" name="l00171"></a><span class="lineno"> 171</span> </div>
<div class="line"><a id="l00172" name="l00172"></a><span class="lineno"> 172</span> <span class="keyword">const</span> <span class="keywordtype">int</span> m_block_idx = <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b">blockN</a> &gt; <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">blockM</a> ? out_row / <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b">blockN</a> : int(tid.x);</div>
<div class="line"><a id="l00173" name="l00173"></a><span class="lineno"> 173</span> </div>
<div class="line"><a id="l00174" name="l00174"></a><span class="lineno"> 174</span> <span class="keyword">const</span> <span class="keywordtype">int</span> out_mask_offset =</div>
<div class="line"><a id="l00175" name="l00175"></a><span class="lineno"> 175</span> !<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab8b64c94f4c8f6f09c0777415589b487">has_output_mask</a> ? 0 : m_block_idx * out_mask_strides[1];</div>
<div class="line"><a id="l00176" name="l00176"></a><span class="lineno"> 176</span> </div>
<div class="line"><a id="l00177" name="l00177"></a><span class="lineno"> 177</span> <span class="keywordtype">int</span> mat_mask_offset =</div>
<div class="line"><a id="l00178" name="l00178"></a><span class="lineno"> 178</span> !<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> ? 0 : m_block_idx * mat_mask_strides[1];</div>
<div class="line"><a id="l00179" name="l00179"></a><span class="lineno"> 179</span> <span class="keywordtype">int</span> vec_mask_offset = 0;</div>
<div class="line"><a id="l00180" name="l00180"></a><span class="lineno"> 180</span> <span class="keyword">const</span> <span class="keywordtype">int</span> mat_mask_step = !<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> ? 0 : mat_mask_strides[0];</div>
<div class="line"><a id="l00181" name="l00181"></a><span class="lineno"> 181</span> <span class="keyword">const</span> <span class="keywordtype">int</span> vec_mask_step = !<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> ? 0 : vec_mask_strides[1];</div>
<div class="line"><a id="l00182" name="l00182"></a><span class="lineno"> 182</span> </div>
<div class="line"><a id="l00183" name="l00183"></a><span class="lineno"> 183</span> T out_scale{1};</div>
<div class="line"><a id="l00184" name="l00184"></a><span class="lineno"> 184</span> </div>
<div class="line"><a id="l00185" name="l00185"></a><span class="lineno"> 185</span> <span class="comment">// Check output mask</span></div>
<div class="line"><a id="l00186" name="l00186"></a><span class="lineno"> 186</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab8b64c94f4c8f6f09c0777415589b487">has_output_mask</a>) {</div>
<div class="line"><a id="l00187" name="l00187"></a><span class="lineno"> 187</span> <span class="keyword">auto</span> mask_out = out_mask[out_mask_offset];</div>
<div class="line"><a id="l00188" name="l00188"></a><span class="lineno"> 188</span> </div>
<div class="line"><a id="l00189" name="l00189"></a><span class="lineno"> 189</span> <span class="comment">// Write zeros and return if mask is 0</span></div>
<div class="line"><a id="l00190" name="l00190"></a><span class="lineno"> 190</span> <span class="keywordflow">if</span> (!mask_out) {</div>
<div class="line"><a id="l00191" name="l00191"></a><span class="lineno"> 191</span> <span class="keywordflow">if</span> (simdN == 0 &amp;&amp; thrN == 0) {</div>
<div class="line"><a id="l00192" name="l00192"></a><span class="lineno"> 192</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00193" name="l00193"></a><span class="lineno"> 193</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00194" name="l00194"></a><span class="lineno"> 194</span> out_vec[out_row + tm] = T(0.);</div>
<div class="line"><a id="l00195" name="l00195"></a><span class="lineno"> 195</span> }</div>
<div class="line"><a id="l00196" name="l00196"></a><span class="lineno"> 196</span> }</div>
<div class="line"><a id="l00197" name="l00197"></a><span class="lineno"> 197</span> </div>
<div class="line"><a id="l00198" name="l00198"></a><span class="lineno"> 198</span> <span class="keywordflow">return</span>;</div>
<div class="line"><a id="l00199" name="l00199"></a><span class="lineno"> 199</span> }</div>
<div class="line"><a id="l00200" name="l00200"></a><span class="lineno"> 200</span> </div>
<div class="line"><a id="l00201" name="l00201"></a><span class="lineno"> 201</span> <span class="comment">// Store scalar if multiplicative mask</span></div>
<div class="line"><a id="l00202" name="l00202"></a><span class="lineno"> 202</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a0edbf2dd6a6563e7afa6dab6b670615c">has_mul_output_mask</a>) {</div>
<div class="line"><a id="l00203" name="l00203"></a><span class="lineno"> 203</span> out_scale = T(mask_out);</div>
<div class="line"><a id="l00204" name="l00204"></a><span class="lineno"> 204</span> }</div>
<div class="line"><a id="l00205" name="l00205"></a><span class="lineno"> 205</span> }</div>
<div class="line"><a id="l00206" name="l00206"></a><span class="lineno"> 206</span> </div>
<div class="line"><a id="l00207" name="l00207"></a><span class="lineno"> 207</span> <span class="comment">// Advance matrix</span></div>
<div class="line"><a id="l00208" name="l00208"></a><span class="lineno"> 208</span> mat += out_row * matrix_ld;</div>
<div class="line"><a id="l00209" name="l00209"></a><span class="lineno"> 209</span> </div>
<div class="line"><a id="l00210" name="l00210"></a><span class="lineno"> 210</span> <span class="comment">// Prepare for loop</span></div>
<div class="line"><a id="l00211" name="l00211"></a><span class="lineno"> 211</span> <span class="keyword">constexpr</span> <span class="keyword">const</span> uniform&lt;int&gt; loop_stride = make_uniform(<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b">blockN</a>);</div>
<div class="line"><a id="l00212" name="l00212"></a><span class="lineno"> 212</span> <span class="keyword">const</span> uniform&lt;int&gt; in_size = make_uniform(in_vec_size);</div>
<div class="line"><a id="l00213" name="l00213"></a><span class="lineno"> 213</span> <span class="keyword">const</span> uniform&lt;int&gt; n_iter = in_size / loop_stride;</div>
<div class="line"><a id="l00214" name="l00214"></a><span class="lineno"> 214</span> <span class="keyword">const</span> uniform&lt;int&gt; last_iter = loop_stride * n_iter;</div>
<div class="line"><a id="l00215" name="l00215"></a><span class="lineno"> 215</span> <span class="keyword">const</span> uniform&lt;int&gt; leftover = in_size - last_iter;</div>
<div class="line"><a id="l00216" name="l00216"></a><span class="lineno"> 216</span> </div>
<div class="line"><a id="l00217" name="l00217"></a><span class="lineno"> 217</span> <span class="comment">// Loop over in_vec in blocks of blockN</span></div>
<div class="line"><a id="l00218" name="l00218"></a><span class="lineno"> 218</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> i = 0; i &lt; n_iter; ++i) {</div>
<div class="line"><a id="l00219" name="l00219"></a><span class="lineno"> 219</span> <span class="keywordflow">if</span> (!<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> ||</div>
<div class="line"><a id="l00220" name="l00220"></a><span class="lineno"> 220</span> (<span class="keywordtype">bool</span>(mat_mask[mat_mask_offset]) &amp;&amp;</div>
<div class="line"><a id="l00221" name="l00221"></a><span class="lineno"> 221</span> <span class="keywordtype">bool</span>(vec_mask[vec_mask_offset]))) {</div>
<div class="line"><a id="l00222" name="l00222"></a><span class="lineno"> 222</span> T block_scale{1};</div>
<div class="line"><a id="l00223" name="l00223"></a><span class="lineno"> 223</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ad47223ee49b3cb7bf3746a2cec45f883">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00224" name="l00224"></a><span class="lineno"> 224</span> block_scale =</div>
<div class="line"><a id="l00225" name="l00225"></a><span class="lineno"> 225</span> T(mat_mask[mat_mask_offset]) * T(vec_mask[vec_mask_offset]);</div>
<div class="line"><a id="l00226" name="l00226"></a><span class="lineno"> 226</span> }</div>
<div class="line"><a id="l00227" name="l00227"></a><span class="lineno"> 227</span> </div>
<div class="line"><a id="l00228" name="l00228"></a><span class="lineno"> 228</span> <a class="code hl_function" href="struct_g_e_m_v_kernel.html#a6013e9c5b2f72fa1311dd038172df0ce">load_unsafe</a>(in_vec, v_coeff, bn);</div>
<div class="line"><a id="l00229" name="l00229"></a><span class="lineno"> 229</span> </div>
<div class="line"><a id="l00230" name="l00230"></a><span class="lineno"> 230</span> <span class="comment">// Apply scale</span></div>
<div class="line"><a id="l00231" name="l00231"></a><span class="lineno"> 231</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ad47223ee49b3cb7bf3746a2cec45f883">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00232" name="l00232"></a><span class="lineno"> 232</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00233" name="l00233"></a><span class="lineno"> 233</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00234" name="l00234"></a><span class="lineno"> 234</span> v_coeff[tn] *= block_scale;</div>
<div class="line"><a id="l00235" name="l00235"></a><span class="lineno"> 235</span> }</div>
<div class="line"><a id="l00236" name="l00236"></a><span class="lineno"> 236</span> }</div>
<div class="line"><a id="l00237" name="l00237"></a><span class="lineno"> 237</span> </div>
<div class="line"><a id="l00238" name="l00238"></a><span class="lineno"> 238</span> <span class="comment">// Per thread work loop</span></div>
<div class="line"><a id="l00239" name="l00239"></a><span class="lineno"> 239</span> <span class="keywordtype">int</span> mat_offset = 0;</div>
<div class="line"><a id="l00240" name="l00240"></a><span class="lineno"> 240</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00241" name="l00241"></a><span class="lineno"> 241</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00242" name="l00242"></a><span class="lineno"> 242</span> <span class="comment">// Load for the row</span></div>
<div class="line"><a id="l00243" name="l00243"></a><span class="lineno"> 243</span> <a class="code hl_function" href="struct_g_e_m_v_kernel.html#a6013e9c5b2f72fa1311dd038172df0ce">load_unsafe</a>(mat, inter, mat_offset + bn);</div>
<div class="line"><a id="l00244" name="l00244"></a><span class="lineno"> 244</span> </div>
<div class="line"><a id="l00245" name="l00245"></a><span class="lineno"> 245</span> <span class="comment">// Accumulate results</span></div>
<div class="line"><a id="l00246" name="l00246"></a><span class="lineno"> 246</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00247" name="l00247"></a><span class="lineno"> 247</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00248" name="l00248"></a><span class="lineno"> 248</span> result[tm] += inter[tn] * v_coeff[tn];</div>
<div class="line"><a id="l00249" name="l00249"></a><span class="lineno"> 249</span> }</div>
<div class="line"><a id="l00250" name="l00250"></a><span class="lineno"> 250</span> </div>
<div class="line"><a id="l00251" name="l00251"></a><span class="lineno"> 251</span> mat_offset += matrix_ld;</div>
<div class="line"><a id="l00252" name="l00252"></a><span class="lineno"> 252</span> }</div>
<div class="line"><a id="l00253" name="l00253"></a><span class="lineno"> 253</span> }</div>
<div class="line"><a id="l00254" name="l00254"></a><span class="lineno"> 254</span> </div>
<div class="line"><a id="l00255" name="l00255"></a><span class="lineno"> 255</span> bn += <a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b">blockN</a>;</div>
<div class="line"><a id="l00256" name="l00256"></a><span class="lineno"> 256</span> mat_mask_offset += mat_mask_step;</div>
<div class="line"><a id="l00257" name="l00257"></a><span class="lineno"> 257</span> vec_mask_offset += vec_mask_step;</div>
<div class="line"><a id="l00258" name="l00258"></a><span class="lineno"> 258</span> }</div>
<div class="line"><a id="l00259" name="l00259"></a><span class="lineno"> 259</span> </div>
<div class="line"><a id="l00260" name="l00260"></a><span class="lineno"> 260</span> <span class="keywordflow">if</span> (leftover &gt; 0 &amp;&amp;</div>
<div class="line"><a id="l00261" name="l00261"></a><span class="lineno"> 261</span> (!<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">has_operand_mask</a> ||</div>
<div class="line"><a id="l00262" name="l00262"></a><span class="lineno"> 262</span> (<span class="keywordtype">bool</span>(mat_mask[mat_mask_offset]) &amp;&amp;</div>
<div class="line"><a id="l00263" name="l00263"></a><span class="lineno"> 263</span> <span class="keywordtype">bool</span>(vec_mask[vec_mask_offset])))) {</div>
<div class="line"><a id="l00264" name="l00264"></a><span class="lineno"> 264</span> T block_scale{1};</div>
<div class="line"><a id="l00265" name="l00265"></a><span class="lineno"> 265</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ad47223ee49b3cb7bf3746a2cec45f883">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00266" name="l00266"></a><span class="lineno"> 266</span> block_scale =</div>
<div class="line"><a id="l00267" name="l00267"></a><span class="lineno"> 267</span> T(mat_mask[mat_mask_offset]) * T(vec_mask[vec_mask_offset]);</div>
<div class="line"><a id="l00268" name="l00268"></a><span class="lineno"> 268</span> }</div>
<div class="line"><a id="l00269" name="l00269"></a><span class="lineno"> 269</span> </div>
<div class="line"><a id="l00270" name="l00270"></a><span class="lineno"> 270</span> <a class="code hl_function" href="struct_g_e_m_v_kernel.html#a04bb72da9a93d6d1eba468fa311bbba7">load_safe</a>(in_vec, v_coeff, bn, in_size);</div>
<div class="line"><a id="l00271" name="l00271"></a><span class="lineno"> 271</span> </div>
<div class="line"><a id="l00272" name="l00272"></a><span class="lineno"> 272</span> <span class="comment">// Apply scale</span></div>
<div class="line"><a id="l00273" name="l00273"></a><span class="lineno"> 273</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ad47223ee49b3cb7bf3746a2cec45f883">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00274" name="l00274"></a><span class="lineno"> 274</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00275" name="l00275"></a><span class="lineno"> 275</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00276" name="l00276"></a><span class="lineno"> 276</span> v_coeff[tn] *= block_scale;</div>
<div class="line"><a id="l00277" name="l00277"></a><span class="lineno"> 277</span> }</div>
<div class="line"><a id="l00278" name="l00278"></a><span class="lineno"> 278</span> }</div>
<div class="line"><a id="l00279" name="l00279"></a><span class="lineno"> 279</span> </div>
<div class="line"><a id="l00280" name="l00280"></a><span class="lineno"> 280</span> <span class="comment">// Per thread work loop</span></div>
<div class="line"><a id="l00281" name="l00281"></a><span class="lineno"> 281</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00282" name="l00282"></a><span class="lineno"> 282</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00283" name="l00283"></a><span class="lineno"> 283</span> <span class="comment">// Load for the row</span></div>
<div class="line"><a id="l00284" name="l00284"></a><span class="lineno"> 284</span> <a class="code hl_function" href="struct_g_e_m_v_kernel.html#a04bb72da9a93d6d1eba468fa311bbba7">load_safe</a>(&amp;mat[tm * matrix_ld], inter, bn, in_size);</div>
<div class="line"><a id="l00285" name="l00285"></a><span class="lineno"> 285</span> </div>
<div class="line"><a id="l00286" name="l00286"></a><span class="lineno"> 286</span> <span class="comment">// Accumulate results</span></div>
<div class="line"><a id="l00287" name="l00287"></a><span class="lineno"> 287</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00288" name="l00288"></a><span class="lineno"> 288</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00289" name="l00289"></a><span class="lineno"> 289</span> result[tm] += inter[tn] * v_coeff[tn];</div>
<div class="line"><a id="l00290" name="l00290"></a><span class="lineno"> 290</span> }</div>
<div class="line"><a id="l00291" name="l00291"></a><span class="lineno"> 291</span> }</div>
<div class="line"><a id="l00292" name="l00292"></a><span class="lineno"> 292</span> }</div>
<div class="line"><a id="l00293" name="l00293"></a><span class="lineno"> 293</span> </div>
<div class="line"><a id="l00294" name="l00294"></a><span class="lineno"> 294</span> <span class="comment">// Apply out scale</span></div>
<div class="line"><a id="l00295" name="l00295"></a><span class="lineno"> 295</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a0edbf2dd6a6563e7afa6dab6b670615c">has_mul_output_mask</a>) {</div>
<div class="line"><a id="l00296" name="l00296"></a><span class="lineno"> 296</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00297" name="l00297"></a><span class="lineno"> 297</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00298" name="l00298"></a><span class="lineno"> 298</span> result[tm] *= out_scale;</div>
<div class="line"><a id="l00299" name="l00299"></a><span class="lineno"> 299</span> }</div>
<div class="line"><a id="l00300" name="l00300"></a><span class="lineno"> 300</span> }</div>
<div class="line"><a id="l00301" name="l00301"></a><span class="lineno"> 301</span> </div>
<div class="line"><a id="l00302" name="l00302"></a><span class="lineno"> 302</span> <span class="comment">// Simdgroup accumulations</span></div>
<div class="line"><a id="l00303" name="l00303"></a><span class="lineno"> 303</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00304" name="l00304"></a><span class="lineno"> 304</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00305" name="l00305"></a><span class="lineno"> 305</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00306" name="l00306"></a><span class="lineno"> 306</span> <span class="keywordflow">for</span> (ushort sn = (SN / 2); sn &gt;= 1; sn &gt;&gt;= 1) {</div>
<div class="line"><a id="l00307" name="l00307"></a><span class="lineno"> 307</span> result[tm] += <a class="code hl_function" href="namespacemetal.html#af6e2dd7ae087aba6abac4f0350b7611c">simd_shuffle_down</a>(result[tm], sn);</div>
<div class="line"><a id="l00308" name="l00308"></a><span class="lineno"> 308</span> }</div>
<div class="line"><a id="l00309" name="l00309"></a><span class="lineno"> 309</span> }</div>
<div class="line"><a id="l00310" name="l00310"></a><span class="lineno"> 310</span> </div>
<div class="line"><a id="l00311" name="l00311"></a><span class="lineno"> 311</span> <span class="comment">// Threadgroup accumulation results</span></div>
<div class="line"><a id="l00312" name="l00312"></a><span class="lineno"> 312</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#ae8113fddf6fb637acfd12efd978b704c">needs_tgp_reduction</a>) {</div>
<div class="line"><a id="l00313" name="l00313"></a><span class="lineno"> 313</span> threadgroup T* tgp_results = tgp_memory + sgN * (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">blockM</a> + TM) + bm;</div>
<div class="line"><a id="l00314" name="l00314"></a><span class="lineno"> 314</span> <span class="keywordflow">if</span> (thrN == 0) {</div>
<div class="line"><a id="l00315" name="l00315"></a><span class="lineno"> 315</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00316" name="l00316"></a><span class="lineno"> 316</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00317" name="l00317"></a><span class="lineno"> 317</span> tgp_results[tm] = result[tm];</div>
<div class="line"><a id="l00318" name="l00318"></a><span class="lineno"> 318</span> }</div>
<div class="line"><a id="l00319" name="l00319"></a><span class="lineno"> 319</span> </div>
<div class="line"><a id="l00320" name="l00320"></a><span class="lineno"> 320</span> threadgroup_barrier(mem_flags::mem_none);</div>
<div class="line"><a id="l00321" name="l00321"></a><span class="lineno"> 321</span> </div>
<div class="line"><a id="l00322" name="l00322"></a><span class="lineno"> 322</span> <span class="keywordflow">if</span> (sgN == 0) {</div>
<div class="line"><a id="l00323" name="l00323"></a><span class="lineno"> 323</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00324" name="l00324"></a><span class="lineno"> 324</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> sgn = 1; sgn &lt; BN; sgn++) {</div>
<div class="line"><a id="l00325" name="l00325"></a><span class="lineno"> 325</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00326" name="l00326"></a><span class="lineno"> 326</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00327" name="l00327"></a><span class="lineno"> 327</span> result[tm] += tgp_results[sgn * (<a class="code hl_variable" href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">blockM</a> + TM) + tm];</div>
<div class="line"><a id="l00328" name="l00328"></a><span class="lineno"> 328</span> }</div>
<div class="line"><a id="l00329" name="l00329"></a><span class="lineno"> 329</span> }</div>
<div class="line"><a id="l00330" name="l00330"></a><span class="lineno"> 330</span> }</div>
<div class="line"><a id="l00331" name="l00331"></a><span class="lineno"> 331</span> }</div>
<div class="line"><a id="l00332" name="l00332"></a><span class="lineno"> 332</span> }</div>
<div class="line"><a id="l00333" name="l00333"></a><span class="lineno"> 333</span> </div>
<div class="line"><a id="l00334" name="l00334"></a><span class="lineno"> 334</span> <span class="comment">// Write outputs</span></div>
<div class="line"><a id="l00335" name="l00335"></a><span class="lineno"> 335</span> <span class="keywordflow">if</span> (simdN == 0 &amp;&amp; thrN == 0) {</div>
<div class="line"><a id="l00336" name="l00336"></a><span class="lineno"> 336</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00337" name="l00337"></a><span class="lineno"> 337</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00338" name="l00338"></a><span class="lineno"> 338</span> out_vec[out_row + tm] = result[tm];</div>
<div class="line"><a id="l00339" name="l00339"></a><span class="lineno"> 339</span> }</div>
<div class="line"><a id="l00340" name="l00340"></a><span class="lineno"> 340</span> }</div>
<div class="line"><a id="l00341" name="l00341"></a><span class="lineno"> 341</span> }</div>
</div>
<div class="line"><a id="l00342" name="l00342"></a><span class="lineno"> 342</span>};</div>
</div>
<div class="line"><a id="l00343" name="l00343"></a><span class="lineno"> 343</span> </div>
<div class="line"><a id="l00347" name="l00347"></a><span class="lineno"> 347</span> </div>
<div class="line"><a id="l00348" name="l00348"></a><span class="lineno"> 348</span><span class="keyword">template</span> &lt;</div>
<div class="line"><a id="l00349" name="l00349"></a><span class="lineno"> 349</span> <span class="keyword">typename</span> T,</div>
<div class="line"><a id="l00350" name="l00350"></a><span class="lineno"> 350</span> <span class="keyword">typename</span> out_mask_t,</div>
<div class="line"><a id="l00351" name="l00351"></a><span class="lineno"> 351</span> <span class="keyword">typename</span> op_mask_t,</div>
<div class="line"><a id="l00352" name="l00352"></a><span class="lineno"> 352</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BM, <span class="comment">/* Threadgroup rows (in simdgroups) */</span></div>
<div class="line"><a id="l00353" name="l00353"></a><span class="lineno"> 353</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BN, <span class="comment">/* Threadgroup cols (in simdgroups) */</span></div>
<div class="line"><a id="l00354" name="l00354"></a><span class="lineno"> 354</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SM, <span class="comment">/* Simdgroup rows (in threads) */</span></div>
<div class="line"><a id="l00355" name="l00355"></a><span class="lineno"> 355</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SN, <span class="comment">/* Simdgroup cols (in threads) */</span></div>
<div class="line"><a id="l00356" name="l00356"></a><span class="lineno"> 356</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TM, <span class="comment">/* Thread rows (in elements) */</span></div>
<div class="line"><a id="l00357" name="l00357"></a><span class="lineno"> 357</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TN&gt; <span class="comment">/* Thread cols (in elements) */</span></div>
<div class="foldopen" id="foldopen00358" data-start="{" data-end="};">
<div class="line"><a id="l00358" name="l00358"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html"> 358</a></span><span class="keyword">struct </span><a class="code hl_struct" href="struct_g_e_m_v_t_kernel.html">GEMVTKernel</a> {</div>
<div class="line"><a id="l00359" name="l00359"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a4a53e73a581aa8881b1f86ce653519e6"> 359</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a4a53e73a581aa8881b1f86ce653519e6">threadsM</a> = BM * SM;</div>
<div class="line"><a id="l00360" name="l00360"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#ade6f15a9744616de9dd71498ad7e758d"> 360</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#ade6f15a9744616de9dd71498ad7e758d">threadsN</a> = BN * SN;</div>
<div class="line"><a id="l00361" name="l00361"></a><span class="lineno"> 361</span> </div>
<div class="line"><a id="l00362" name="l00362"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a2ae8ce535d59cccf453381b4485a77f0"> 362</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a2ae8ce535d59cccf453381b4485a77f0">blockM</a> = <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a4a53e73a581aa8881b1f86ce653519e6">threadsM</a> * TM;</div>
<div class="line"><a id="l00363" name="l00363"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a"> 363</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">int</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a">blockN</a> = <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#ade6f15a9744616de9dd71498ad7e758d">threadsN</a> * TN;</div>
<div class="line"><a id="l00364" name="l00364"></a><span class="lineno"> 364</span> </div>
<div class="line"><a id="l00365" name="l00365"></a><span class="lineno"> 365</span> <span class="keyword">static_assert</span>(SM * SN == 32, <span class="stringliteral">&quot;simdgroup can only have 32 threads&quot;</span>);</div>
<div class="line"><a id="l00366" name="l00366"></a><span class="lineno"> 366</span> </div>
<div class="line"><a id="l00367" name="l00367"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869"> 367</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> = !metal::is_same_v&lt;op_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00368" name="l00368"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#aaefdf8f023da255bbb70a0c3e3408626"> 368</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#aaefdf8f023da255bbb70a0c3e3408626">has_output_mask</a> = !metal::is_same_v&lt;out_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00369" name="l00369"></a><span class="lineno"> 369</span> </div>
<div class="line"><a id="l00370" name="l00370"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a8db6f01f96a36b216acd801c34a96ef5"> 370</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8db6f01f96a36b216acd801c34a96ef5">has_mul_operand_mask</a> =</div>
<div class="line"><a id="l00371" name="l00371"></a><span class="lineno"> 371</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> &amp;&amp; !metal::is_same_v&lt;op_mask_t, bool&gt;;</div>
<div class="line"><a id="l00372" name="l00372"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a8eb06f6569e4042e24fee220b11fa10d"> 372</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8eb06f6569e4042e24fee220b11fa10d">has_mul_output_mask</a> =</div>
<div class="line"><a id="l00373" name="l00373"></a><span class="lineno"> 373</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#aaefdf8f023da255bbb70a0c3e3408626">has_output_mask</a> &amp;&amp; !metal::is_same_v&lt;out_mask_t, bool&gt;;</div>
<div class="line"><a id="l00374" name="l00374"></a><span class="lineno"> 374</span> </div>
<div class="line"><a id="l00375" name="l00375"></a><span class="lineno"> 375</span> <span class="comment">// - The matrix of size (M = in_vec_size, N = out_vec_size) is divided up</span></div>
<div class="line"><a id="l00376" name="l00376"></a><span class="lineno"> 376</span> <span class="comment">// into blocks of (blockM, blockN) divided among threadgroups</span></div>
<div class="line"><a id="l00377" name="l00377"></a><span class="lineno"> 377</span> <span class="comment">// - Every thread works on a block of (TM, TN)</span></div>
<div class="line"><a id="l00378" name="l00378"></a><span class="lineno"> 378</span> <span class="comment">// - We assume each threadgroup has (threadsN, threadsM, 1) threads</span></div>
<div class="line"><a id="l00379" name="l00379"></a><span class="lineno"> 379</span> <span class="comment">//</span></div>
<div class="line"><a id="l00380" name="l00380"></a><span class="lineno"> 380</span> <span class="comment">// 1. A thread loads TN elements each from mat along TM contiguous rows</span></div>
<div class="line"><a id="l00381" name="l00381"></a><span class="lineno"> 381</span> <span class="comment">// and the corresponding scalar from the vector</span></div>
<div class="line"><a id="l00382" name="l00382"></a><span class="lineno"> 382</span> <span class="comment">// 2. The thread then accumulates its local result for the block</span></div>
<div class="line"><a id="l00383" name="l00383"></a><span class="lineno"> 383</span> <span class="comment">// 3. At the end, each thread has accumulated results over all blocks across</span></div>
<div class="line"><a id="l00384" name="l00384"></a><span class="lineno"> 384</span> <span class="comment">// the rows. These are then summed up across the threadgroup</span></div>
<div class="line"><a id="l00385" name="l00385"></a><span class="lineno"> 385</span> <span class="comment">// 4. Each threadgroup writes its accumulated BN * TN outputs</span></div>
<div class="line"><a id="l00386" name="l00386"></a><span class="lineno"> 386</span> <span class="comment">//</span></div>
<div class="line"><a id="l00387" name="l00387"></a><span class="lineno"> 387</span> <span class="comment">// Edge case handling:</span></div>
<div class="line"><a id="l00388" name="l00388"></a><span class="lineno"> 388</span> <span class="comment">// - The threadgroup with the largest tid has blocks that exceed the matrix</span></div>
<div class="line"><a id="l00389" name="l00389"></a><span class="lineno"> 389</span> <span class="comment">// * The blocks that start outside the matrix are never read (thread results</span></div>
<div class="line"><a id="l00390" name="l00390"></a><span class="lineno"> 390</span> <span class="comment">// remain zero)</span></div>
<div class="line"><a id="l00391" name="l00391"></a><span class="lineno"> 391</span> <span class="comment">// * The last thread that partially overlaps with the matrix is shifted</span></div>
<div class="line"><a id="l00392" name="l00392"></a><span class="lineno"> 392</span> <span class="comment">// inwards such that the thread block fits exactly in the matrix</span></div>
<div class="line"><a id="l00393" name="l00393"></a><span class="lineno"> 393</span> </div>
<div class="line"><a id="l00394" name="l00394"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a48a09a21d7b822f380d040c752b785d7"> 394</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">short</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a48a09a21d7b822f380d040c752b785d7">tgp_mem_size</a> = BM &gt; 1 ? BM*(<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a">blockN</a> + TN) : 0;</div>
<div class="line"><a id="l00395" name="l00395"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a67be7ec69c3791f02e97ccdb00ae0e03"> 395</a></span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a> <span class="keywordtype">bool</span> <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a67be7ec69c3791f02e97ccdb00ae0e03">needs_tgp_reduction</a> = BM &gt; 1;</div>
<div class="line"><a id="l00396" name="l00396"></a><span class="lineno"> 396</span> </div>
<div class="foldopen" id="foldopen00397" data-start="{" data-end="}">
<div class="line"><a id="l00397" name="l00397"></a><span class="lineno"><a class="line" href="struct_g_e_m_v_t_kernel.html#a5d68656832de892f33db939005713927"> 397</a></span> <span class="keyword">static</span> METAL_FUNC <span class="keywordtype">void</span> <a class="code hl_function" href="struct_g_e_m_v_t_kernel.html#a5d68656832de892f33db939005713927">run</a>(</div>
<div class="line"><a id="l00398" name="l00398"></a><span class="lineno"> 398</span> <span class="keyword">const</span> device T* mat [[buffer(0)]],</div>
<div class="line"><a id="l00399" name="l00399"></a><span class="lineno"> 399</span> <span class="keyword">const</span> device T* in_vec [[buffer(1)]],</div>
<div class="line"><a id="l00400" name="l00400"></a><span class="lineno"> 400</span> device T* out_vec [[buffer(3)]],</div>
<div class="line"><a id="l00401" name="l00401"></a><span class="lineno"> 401</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; in_vec_size [[buffer(4)]],</div>
<div class="line"><a id="l00402" name="l00402"></a><span class="lineno"> 402</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; out_vec_size [[buffer(5)]],</div>
<div class="line"><a id="l00403" name="l00403"></a><span class="lineno"> 403</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; marix_ld [[buffer(6)]],</div>
<div class="line"><a id="l00404" name="l00404"></a><span class="lineno"> 404</span> <span class="keyword">const</span> device out_mask_t* out_mask [[buffer(20)]],</div>
<div class="line"><a id="l00405" name="l00405"></a><span class="lineno"> 405</span> <span class="keyword">const</span> device op_mask_t* mat_mask [[buffer(21)]],</div>
<div class="line"><a id="l00406" name="l00406"></a><span class="lineno"> 406</span> <span class="keyword">const</span> device op_mask_t* vec_mask [[buffer(22)]],</div>
<div class="line"><a id="l00407" name="l00407"></a><span class="lineno"> 407</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* mask_strides [[buffer(23)]],</div>
<div class="line"><a id="l00408" name="l00408"></a><span class="lineno"> 408</span> threadgroup T* tgp_memory [[threadgroup(0)]],</div>
<div class="line"><a id="l00409" name="l00409"></a><span class="lineno"> 409</span> uint3 tid [[threadgroup_position_in_grid]],</div>
<div class="line"><a id="l00410" name="l00410"></a><span class="lineno"> 410</span> uint3 lid [[thread_position_in_threadgroup]],</div>
<div class="line"><a id="l00411" name="l00411"></a><span class="lineno"> 411</span> uint simd_gid [[simdgroup_index_in_threadgroup]],</div>
<div class="line"><a id="l00412" name="l00412"></a><span class="lineno"> 412</span> uint simd_lid [[thread_index_in_simdgroup]]) {</div>
<div class="line"><a id="l00413" name="l00413"></a><span class="lineno"> 413</span> <span class="comment">// Appease compiler</span></div>
<div class="line"><a id="l00414" name="l00414"></a><span class="lineno"> 414</span> (void)lid;</div>
<div class="line"><a id="l00415" name="l00415"></a><span class="lineno"> 415</span> </div>
<div class="line"><a id="l00416" name="l00416"></a><span class="lineno"> 416</span> <span class="comment">// Thread local accumulation results</span></div>
<div class="line"><a id="l00417" name="l00417"></a><span class="lineno"> 417</span> T result[TN] = {0};</div>
<div class="line"><a id="l00418" name="l00418"></a><span class="lineno"> 418</span> T inter[TN];</div>
<div class="line"><a id="l00419" name="l00419"></a><span class="lineno"> 419</span> T v_coeff[TM];</div>
<div class="line"><a id="l00420" name="l00420"></a><span class="lineno"> 420</span> </div>
<div class="line"><a id="l00421" name="l00421"></a><span class="lineno"> 421</span> <span class="keyword">const</span> <span class="keywordtype">int</span> thrM = SN != 32 ? simd_lid / SN : 0;</div>
<div class="line"><a id="l00422" name="l00422"></a><span class="lineno"> 422</span> <span class="keyword">const</span> <span class="keywordtype">int</span> thrN = SN != 32 ? simd_lid % SN : int(simd_lid);</div>
<div class="line"><a id="l00423" name="l00423"></a><span class="lineno"> 423</span> </div>
<div class="line"><a id="l00424" name="l00424"></a><span class="lineno"> 424</span> <span class="keyword">const</span> <span class="keywordtype">int</span> sgM = BN != 1 ? (simd_gid / BN) : int(simd_gid);</div>
<div class="line"><a id="l00425" name="l00425"></a><span class="lineno"> 425</span> <span class="keyword">const</span> <span class="keywordtype">int</span> sgN = BN != 1 ? (simd_gid % BN) : 0;</div>
<div class="line"><a id="l00426" name="l00426"></a><span class="lineno"> 426</span> </div>
<div class="line"><a id="l00427" name="l00427"></a><span class="lineno"> 427</span> <span class="keyword">const</span> <span class="keywordtype">int</span> simdM = SM * sgM;</div>
<div class="line"><a id="l00428" name="l00428"></a><span class="lineno"> 428</span> <span class="keyword">const</span> <span class="keywordtype">int</span> simdN = SN * sgN;</div>
<div class="line"><a id="l00429" name="l00429"></a><span class="lineno"> 429</span> </div>
<div class="line"><a id="l00430" name="l00430"></a><span class="lineno"> 430</span> <span class="keywordtype">int</span> cm = (simdM + thrM);</div>
<div class="line"><a id="l00431" name="l00431"></a><span class="lineno"> 431</span> <span class="keywordtype">int</span> cn = (simdN + thrN);</div>
<div class="line"><a id="l00432" name="l00432"></a><span class="lineno"> 432</span> </div>
<div class="line"><a id="l00433" name="l00433"></a><span class="lineno"> 433</span> <span class="keywordtype">int</span> bm = cm * TM;</div>
<div class="line"><a id="l00434" name="l00434"></a><span class="lineno"> 434</span> <span class="keywordtype">int</span> bn = cn * TN;</div>
<div class="line"><a id="l00435" name="l00435"></a><span class="lineno"> 435</span> </div>
<div class="line"><a id="l00436" name="l00436"></a><span class="lineno"> 436</span> <span class="keywordtype">int</span> out_col = tid.x * <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a">blockN</a> + bn;</div>
<div class="line"><a id="l00437" name="l00437"></a><span class="lineno"> 437</span> </div>
<div class="line"><a id="l00438" name="l00438"></a><span class="lineno"> 438</span> <span class="comment">// Prepare mask offsets</span></div>
<div class="line"><a id="l00439" name="l00439"></a><span class="lineno"> 439</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* out_mask_strides = mask_strides;</div>
<div class="line"><a id="l00440" name="l00440"></a><span class="lineno"> 440</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* mat_mask_strides =</div>
<div class="line"><a id="l00441" name="l00441"></a><span class="lineno"> 441</span> out_mask_strides + (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#aaefdf8f023da255bbb70a0c3e3408626">has_output_mask</a> ? 2 : 0);</div>
<div class="line"><a id="l00442" name="l00442"></a><span class="lineno"> 442</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* vec_mask_strides =</div>
<div class="line"><a id="l00443" name="l00443"></a><span class="lineno"> 443</span> mat_mask_strides + (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> ? 2 : 0);</div>
<div class="line"><a id="l00444" name="l00444"></a><span class="lineno"> 444</span> </div>
<div class="line"><a id="l00445" name="l00445"></a><span class="lineno"> 445</span> <span class="keyword">const</span> <span class="keywordtype">int</span> n_block_idx = <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a2ae8ce535d59cccf453381b4485a77f0">blockM</a> &gt; <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a">blockN</a> ? out_col / <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a2ae8ce535d59cccf453381b4485a77f0">blockM</a> : int(tid.x);</div>
<div class="line"><a id="l00446" name="l00446"></a><span class="lineno"> 446</span> </div>
<div class="line"><a id="l00447" name="l00447"></a><span class="lineno"> 447</span> <span class="keyword">const</span> <span class="keywordtype">int</span> out_mask_offset =</div>
<div class="line"><a id="l00448" name="l00448"></a><span class="lineno"> 448</span> !<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#aaefdf8f023da255bbb70a0c3e3408626">has_output_mask</a> ? 0 : n_block_idx; <span class="comment">// * out_mask_strides[0];</span></div>
<div class="line"><a id="l00449" name="l00449"></a><span class="lineno"> 449</span> </div>
<div class="line"><a id="l00450" name="l00450"></a><span class="lineno"> 450</span> <span class="keywordtype">int</span> mat_mask_offset =</div>
<div class="line"><a id="l00451" name="l00451"></a><span class="lineno"> 451</span> !<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> ? 0 : n_block_idx * mat_mask_strides[0];</div>
<div class="line"><a id="l00452" name="l00452"></a><span class="lineno"> 452</span> <span class="keywordtype">int</span> vec_mask_offset = 0;</div>
<div class="line"><a id="l00453" name="l00453"></a><span class="lineno"> 453</span> <span class="keyword">const</span> <span class="keywordtype">int</span> mat_mask_step = !<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> ? 0 : mat_mask_strides[1];</div>
<div class="line"><a id="l00454" name="l00454"></a><span class="lineno"> 454</span> <span class="keyword">const</span> <span class="keywordtype">int</span> vec_mask_step = !<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> ? 0 : vec_mask_strides[0];</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> T out_scale{1};</div>
<div class="line"><a id="l00457" name="l00457"></a><span class="lineno"> 457</span> </div>
<div class="line"><a id="l00458" name="l00458"></a><span class="lineno"> 458</span> <span class="comment">// Check output mask</span></div>
<div class="line"><a id="l00459" name="l00459"></a><span class="lineno"> 459</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#aaefdf8f023da255bbb70a0c3e3408626">has_output_mask</a>) {</div>
<div class="line"><a id="l00460" name="l00460"></a><span class="lineno"> 460</span> <span class="keyword">auto</span> mask_out = out_mask[out_mask_offset];</div>
<div class="line"><a id="l00461" name="l00461"></a><span class="lineno"> 461</span> </div>
<div class="line"><a id="l00462" name="l00462"></a><span class="lineno"> 462</span> <span class="comment">// Write zeros and return if mask is 0</span></div>
<div class="line"><a id="l00463" name="l00463"></a><span class="lineno"> 463</span> <span class="keywordflow">if</span> (!mask_out) {</div>
<div class="line"><a id="l00464" name="l00464"></a><span class="lineno"> 464</span> <span class="keywordflow">if</span> (cm == 0 &amp;&amp; out_col &lt; out_vec_size) {</div>
<div class="line"><a id="l00465" name="l00465"></a><span class="lineno"> 465</span> <span class="keywordflow">if</span> (out_col + TN &lt;= out_vec_size) {</div>
<div class="line"><a id="l00466" name="l00466"></a><span class="lineno"> 466</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00467" name="l00467"></a><span class="lineno"> 467</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00468" name="l00468"></a><span class="lineno"> 468</span> out_vec[out_col + tn] = T(0.);</div>
<div class="line"><a id="l00469" name="l00469"></a><span class="lineno"> 469</span> }</div>
<div class="line"><a id="l00470" name="l00470"></a><span class="lineno"> 470</span> } <span class="keywordflow">else</span> {</div>
<div class="line"><a id="l00471" name="l00471"></a><span class="lineno"> 471</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN &amp;&amp; (out_col + tn) &lt; out_vec_size; tn++) {</div>
<div class="line"><a id="l00472" name="l00472"></a><span class="lineno"> 472</span> out_vec[out_col + tn] = T(0.);</div>
<div class="line"><a id="l00473" name="l00473"></a><span class="lineno"> 473</span> }</div>
<div class="line"><a id="l00474" name="l00474"></a><span class="lineno"> 474</span> }</div>
<div class="line"><a id="l00475" name="l00475"></a><span class="lineno"> 475</span> }</div>
<div class="line"><a id="l00476" name="l00476"></a><span class="lineno"> 476</span> </div>
<div class="line"><a id="l00477" name="l00477"></a><span class="lineno"> 477</span> <span class="keywordflow">return</span>;</div>
<div class="line"><a id="l00478" name="l00478"></a><span class="lineno"> 478</span> }</div>
<div class="line"><a id="l00479" name="l00479"></a><span class="lineno"> 479</span> </div>
<div class="line"><a id="l00480" name="l00480"></a><span class="lineno"> 480</span> <span class="comment">// Store scalar if multiplicative mask</span></div>
<div class="line"><a id="l00481" name="l00481"></a><span class="lineno"> 481</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8eb06f6569e4042e24fee220b11fa10d">has_mul_output_mask</a>) {</div>
<div class="line"><a id="l00482" name="l00482"></a><span class="lineno"> 482</span> out_scale = T(mask_out);</div>
<div class="line"><a id="l00483" name="l00483"></a><span class="lineno"> 483</span> }</div>
<div class="line"><a id="l00484" name="l00484"></a><span class="lineno"> 484</span> }</div>
<div class="line"><a id="l00485" name="l00485"></a><span class="lineno"> 485</span> </div>
<div class="line"><a id="l00486" name="l00486"></a><span class="lineno"> 486</span> <span class="comment">// Prepare for loop</span></div>
<div class="line"><a id="l00487" name="l00487"></a><span class="lineno"> 487</span> <span class="keyword">constexpr</span> <span class="keyword">const</span> uniform&lt;int&gt; loop_stride = make_uniform(<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a2ae8ce535d59cccf453381b4485a77f0">blockM</a>);</div>
<div class="line"><a id="l00488" name="l00488"></a><span class="lineno"> 488</span> <span class="keyword">const</span> uniform&lt;int&gt; in_size = make_uniform(in_vec_size);</div>
<div class="line"><a id="l00489" name="l00489"></a><span class="lineno"> 489</span> <span class="keyword">const</span> uniform&lt;int&gt; n_iter = in_size / loop_stride;</div>
<div class="line"><a id="l00490" name="l00490"></a><span class="lineno"> 490</span> <span class="keyword">const</span> uniform&lt;int&gt; last_iter = loop_stride * n_iter;</div>
<div class="line"><a id="l00491" name="l00491"></a><span class="lineno"> 491</span> <span class="keyword">const</span> uniform&lt;int&gt; leftover = in_size - last_iter;</div>
<div class="line"><a id="l00492" name="l00492"></a><span class="lineno"> 492</span> </div>
<div class="line"><a id="l00493" name="l00493"></a><span class="lineno"> 493</span> <span class="comment">// Edgecase handling</span></div>
<div class="line"><a id="l00494" name="l00494"></a><span class="lineno"> 494</span> <span class="keywordflow">if</span> (out_col &lt; out_vec_size) {</div>
<div class="line"><a id="l00495" name="l00495"></a><span class="lineno"> 495</span> out_col = (out_col + TN) &lt;= out_vec_size ? out_col : out_vec_size - TN;</div>
<div class="line"><a id="l00496" name="l00496"></a><span class="lineno"> 496</span> </div>
<div class="line"><a id="l00497" name="l00497"></a><span class="lineno"> 497</span> <span class="comment">// Per thread accumulation main loop</span></div>
<div class="line"><a id="l00498" name="l00498"></a><span class="lineno"> 498</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> i = 0; i &lt; n_iter; ++i) {</div>
<div class="line"><a id="l00499" name="l00499"></a><span class="lineno"> 499</span> <span class="comment">// Adding a threadgroup_barrier improves performance slightly</span></div>
<div class="line"><a id="l00500" name="l00500"></a><span class="lineno"> 500</span> <span class="comment">// This is possibly it may help exploit cache better</span></div>
<div class="line"><a id="l00501" name="l00501"></a><span class="lineno"> 501</span> threadgroup_barrier(mem_flags::mem_none);</div>
<div class="line"><a id="l00502" name="l00502"></a><span class="lineno"> 502</span> </div>
<div class="line"><a id="l00503" name="l00503"></a><span class="lineno"> 503</span> <span class="keywordflow">if</span> (!<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> ||</div>
<div class="line"><a id="l00504" name="l00504"></a><span class="lineno"> 504</span> (<span class="keywordtype">bool</span>(mat_mask[mat_mask_offset]) &amp;&amp;</div>
<div class="line"><a id="l00505" name="l00505"></a><span class="lineno"> 505</span> <span class="keywordtype">bool</span>(vec_mask[vec_mask_offset]))) {</div>
<div class="line"><a id="l00506" name="l00506"></a><span class="lineno"> 506</span> T block_scale{1};</div>
<div class="line"><a id="l00507" name="l00507"></a><span class="lineno"> 507</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8db6f01f96a36b216acd801c34a96ef5">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00508" name="l00508"></a><span class="lineno"> 508</span> block_scale =</div>
<div class="line"><a id="l00509" name="l00509"></a><span class="lineno"> 509</span> T(mat_mask[mat_mask_offset]) * T(vec_mask[vec_mask_offset]);</div>
<div class="line"><a id="l00510" name="l00510"></a><span class="lineno"> 510</span> }</div>
<div class="line"><a id="l00511" name="l00511"></a><span class="lineno"> 511</span> </div>
<div class="line"><a id="l00512" name="l00512"></a><span class="lineno"> 512</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00513" name="l00513"></a><span class="lineno"> 513</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00514" name="l00514"></a><span class="lineno"> 514</span> v_coeff[tm] = in_vec[bm + tm];</div>
<div class="line"><a id="l00515" name="l00515"></a><span class="lineno"> 515</span> }</div>
<div class="line"><a id="l00516" name="l00516"></a><span class="lineno"> 516</span> </div>
<div class="line"><a id="l00517" name="l00517"></a><span class="lineno"> 517</span> <span class="comment">// Apply scale</span></div>
<div class="line"><a id="l00518" name="l00518"></a><span class="lineno"> 518</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8db6f01f96a36b216acd801c34a96ef5">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00519" name="l00519"></a><span class="lineno"> 519</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00520" name="l00520"></a><span class="lineno"> 520</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00521" name="l00521"></a><span class="lineno"> 521</span> v_coeff[tm] *= block_scale;</div>
<div class="line"><a id="l00522" name="l00522"></a><span class="lineno"> 522</span> }</div>
<div class="line"><a id="l00523" name="l00523"></a><span class="lineno"> 523</span> }</div>
<div class="line"><a id="l00524" name="l00524"></a><span class="lineno"> 524</span> </div>
<div class="line"><a id="l00525" name="l00525"></a><span class="lineno"> 525</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00526" name="l00526"></a><span class="lineno"> 526</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM; tm++) {</div>
<div class="line"><a id="l00527" name="l00527"></a><span class="lineno"> 527</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00528" name="l00528"></a><span class="lineno"> 528</span> inter[tn] = mat[(bm + tm) * marix_ld + out_col + tn];</div>
<div class="line"><a id="l00529" name="l00529"></a><span class="lineno"> 529</span> }</div>
<div class="line"><a id="l00530" name="l00530"></a><span class="lineno"> 530</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00531" name="l00531"></a><span class="lineno"> 531</span> result[tn] += v_coeff[tm] * inter[tn];</div>
<div class="line"><a id="l00532" name="l00532"></a><span class="lineno"> 532</span> }</div>
<div class="line"><a id="l00533" name="l00533"></a><span class="lineno"> 533</span> }</div>
<div class="line"><a id="l00534" name="l00534"></a><span class="lineno"> 534</span> }</div>
<div class="line"><a id="l00535" name="l00535"></a><span class="lineno"> 535</span> </div>
<div class="line"><a id="l00536" name="l00536"></a><span class="lineno"> 536</span> bm += <a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a2ae8ce535d59cccf453381b4485a77f0">blockM</a>;</div>
<div class="line"><a id="l00537" name="l00537"></a><span class="lineno"> 537</span> mat_mask_offset += mat_mask_step;</div>
<div class="line"><a id="l00538" name="l00538"></a><span class="lineno"> 538</span> vec_mask_offset += vec_mask_step;</div>
<div class="line"><a id="l00539" name="l00539"></a><span class="lineno"> 539</span> }</div>
<div class="line"><a id="l00540" name="l00540"></a><span class="lineno"> 540</span> </div>
<div class="line"><a id="l00541" name="l00541"></a><span class="lineno"> 541</span> <span class="keywordflow">if</span> (leftover &gt; 0 &amp;&amp;</div>
<div class="line"><a id="l00542" name="l00542"></a><span class="lineno"> 542</span> (!<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">has_operand_mask</a> ||</div>
<div class="line"><a id="l00543" name="l00543"></a><span class="lineno"> 543</span> (<span class="keywordtype">bool</span>(mat_mask[mat_mask_offset]) &amp;&amp;</div>
<div class="line"><a id="l00544" name="l00544"></a><span class="lineno"> 544</span> <span class="keywordtype">bool</span>(vec_mask[vec_mask_offset])))) {</div>
<div class="line"><a id="l00545" name="l00545"></a><span class="lineno"> 545</span> T block_scale{1};</div>
<div class="line"><a id="l00546" name="l00546"></a><span class="lineno"> 546</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8db6f01f96a36b216acd801c34a96ef5">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00547" name="l00547"></a><span class="lineno"> 547</span> block_scale =</div>
<div class="line"><a id="l00548" name="l00548"></a><span class="lineno"> 548</span> T(mat_mask[mat_mask_offset]) * T(vec_mask[vec_mask_offset]);</div>
<div class="line"><a id="l00549" name="l00549"></a><span class="lineno"> 549</span> }</div>
<div class="line"><a id="l00550" name="l00550"></a><span class="lineno"> 550</span> </div>
<div class="line"><a id="l00551" name="l00551"></a><span class="lineno"> 551</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tm = 0; tm &lt; TM &amp;&amp; bm + tm &lt; in_vec_size; tm++) {</div>
<div class="line"><a id="l00552" name="l00552"></a><span class="lineno"> 552</span> v_coeff[tm] = in_vec[bm + tm];</div>
<div class="line"><a id="l00553" name="l00553"></a><span class="lineno"> 553</span> </div>
<div class="line"><a id="l00554" name="l00554"></a><span class="lineno"> 554</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8db6f01f96a36b216acd801c34a96ef5">has_mul_operand_mask</a>) {</div>
<div class="line"><a id="l00555" name="l00555"></a><span class="lineno"> 555</span> v_coeff[tm] *= block_scale;</div>
<div class="line"><a id="l00556" name="l00556"></a><span class="lineno"> 556</span> }</div>
<div class="line"><a id="l00557" name="l00557"></a><span class="lineno"> 557</span> </div>
<div class="line"><a id="l00558" name="l00558"></a><span class="lineno"> 558</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00559" name="l00559"></a><span class="lineno"> 559</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00560" name="l00560"></a><span class="lineno"> 560</span> inter[tn] = mat[(bm + tm) * marix_ld + out_col + tn];</div>
<div class="line"><a id="l00561" name="l00561"></a><span class="lineno"> 561</span> }</div>
<div class="line"><a id="l00562" name="l00562"></a><span class="lineno"> 562</span> </div>
<div class="line"><a id="l00563" name="l00563"></a><span class="lineno"> 563</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00564" name="l00564"></a><span class="lineno"> 564</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00565" name="l00565"></a><span class="lineno"> 565</span> result[tn] += v_coeff[tm] * inter[tn];</div>
<div class="line"><a id="l00566" name="l00566"></a><span class="lineno"> 566</span> }</div>
<div class="line"><a id="l00567" name="l00567"></a><span class="lineno"> 567</span> }</div>
<div class="line"><a id="l00568" name="l00568"></a><span class="lineno"> 568</span> }</div>
<div class="line"><a id="l00569" name="l00569"></a><span class="lineno"> 569</span> }</div>
<div class="line"><a id="l00570" name="l00570"></a><span class="lineno"> 570</span> </div>
<div class="line"><a id="l00571" name="l00571"></a><span class="lineno"> 571</span> <span class="comment">// Apply out scale</span></div>
<div class="line"><a id="l00572" name="l00572"></a><span class="lineno"> 572</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a8eb06f6569e4042e24fee220b11fa10d">has_mul_output_mask</a>) {</div>
<div class="line"><a id="l00573" name="l00573"></a><span class="lineno"> 573</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00574" name="l00574"></a><span class="lineno"> 574</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00575" name="l00575"></a><span class="lineno"> 575</span> result[tn] *= out_scale;</div>
<div class="line"><a id="l00576" name="l00576"></a><span class="lineno"> 576</span> }</div>
<div class="line"><a id="l00577" name="l00577"></a><span class="lineno"> 577</span> }</div>
<div class="line"><a id="l00578" name="l00578"></a><span class="lineno"> 578</span> </div>
<div class="line"><a id="l00579" name="l00579"></a><span class="lineno"> 579</span> <span class="comment">// Simdgroup accumulations</span></div>
<div class="line"><a id="l00580" name="l00580"></a><span class="lineno"> 580</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00581" name="l00581"></a><span class="lineno"> 581</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00582" name="l00582"></a><span class="lineno"> 582</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00583" name="l00583"></a><span class="lineno"> 583</span> <span class="keywordflow">for</span> (ushort sm = (SM / 2); sm &gt;= 1; sm &gt;&gt;= 1) {</div>
<div class="line"><a id="l00584" name="l00584"></a><span class="lineno"> 584</span> result[tn] += <a class="code hl_function" href="namespacemetal.html#af6e2dd7ae087aba6abac4f0350b7611c">simd_shuffle_down</a>(result[tn], SN * sm);</div>
<div class="line"><a id="l00585" name="l00585"></a><span class="lineno"> 585</span> }</div>
<div class="line"><a id="l00586" name="l00586"></a><span class="lineno"> 586</span> }</div>
<div class="line"><a id="l00587" name="l00587"></a><span class="lineno"> 587</span> </div>
<div class="line"><a id="l00588" name="l00588"></a><span class="lineno"> 588</span> <span class="comment">// Threadgroup accumulation results</span></div>
<div class="line"><a id="l00589" name="l00589"></a><span class="lineno"> 589</span> <span class="keywordflow">if</span> (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a67be7ec69c3791f02e97ccdb00ae0e03">needs_tgp_reduction</a>) {</div>
<div class="line"><a id="l00590" name="l00590"></a><span class="lineno"> 590</span> threadgroup T* tgp_results = tgp_memory + sgM * (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a">blockN</a> + TN) + bn;</div>
<div class="line"><a id="l00591" name="l00591"></a><span class="lineno"> 591</span> <span class="keywordflow">if</span> (thrM == 0) {</div>
<div class="line"><a id="l00592" name="l00592"></a><span class="lineno"> 592</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00593" name="l00593"></a><span class="lineno"> 593</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00594" name="l00594"></a><span class="lineno"> 594</span> tgp_results[tn] = result[tn];</div>
<div class="line"><a id="l00595" name="l00595"></a><span class="lineno"> 595</span> }</div>
<div class="line"><a id="l00596" name="l00596"></a><span class="lineno"> 596</span> </div>
<div class="line"><a id="l00597" name="l00597"></a><span class="lineno"> 597</span> threadgroup_barrier(mem_flags::mem_none);</div>
<div class="line"><a id="l00598" name="l00598"></a><span class="lineno"> 598</span> </div>
<div class="line"><a id="l00599" name="l00599"></a><span class="lineno"> 599</span> <span class="keywordflow">if</span> (sgM == 0) {</div>
<div class="line"><a id="l00600" name="l00600"></a><span class="lineno"> 600</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00601" name="l00601"></a><span class="lineno"> 601</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> sgm = 1; sgm &lt; BM; sgm++) {</div>
<div class="line"><a id="l00602" name="l00602"></a><span class="lineno"> 602</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00603" name="l00603"></a><span class="lineno"> 603</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> tn = 0; tn &lt; TN; tn++) {</div>
<div class="line"><a id="l00604" name="l00604"></a><span class="lineno"> 604</span> result[tn] += tgp_results[sgm * (<a class="code hl_variable" href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a">blockN</a> + TN) + tn];</div>
<div class="line"><a id="l00605" name="l00605"></a><span class="lineno"> 605</span> }</div>
<div class="line"><a id="l00606" name="l00606"></a><span class="lineno"> 606</span> }</div>
<div class="line"><a id="l00607" name="l00607"></a><span class="lineno"> 607</span> }</div>
<div class="line"><a id="l00608" name="l00608"></a><span class="lineno"> 608</span> }</div>
<div class="line"><a id="l00609" name="l00609"></a><span class="lineno"> 609</span> }</div>
<div class="line"><a id="l00610" name="l00610"></a><span class="lineno"> 610</span> </div>
<div class="line"><a id="l00611" name="l00611"></a><span class="lineno"> 611</span> <span class="comment">// Threadgroup accumulation and writing out results</span></div>
<div class="line"><a id="l00612" name="l00612"></a><span class="lineno"> 612</span> <span class="keywordflow">if</span> (cm == 0 &amp;&amp; out_col &lt; out_vec_size) {</div>
<div class="line"><a id="l00613" name="l00613"></a><span class="lineno"> 613</span> <a class="code hl_define" href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div>
<div class="line"><a id="l00614" name="l00614"></a><span class="lineno"> 614</span> <span class="keywordflow">for</span> (<span class="keywordtype">int</span> j = 0; j &lt; TN; j++) {</div>
<div class="line"><a id="l00615" name="l00615"></a><span class="lineno"> 615</span> out_vec[out_col + j] = result[j];</div>
<div class="line"><a id="l00616" name="l00616"></a><span class="lineno"> 616</span> }</div>
<div class="line"><a id="l00617" name="l00617"></a><span class="lineno"> 617</span> }</div>
<div class="line"><a id="l00618" name="l00618"></a><span class="lineno"> 618</span> }</div>
</div>
<div class="line"><a id="l00619" name="l00619"></a><span class="lineno"> 619</span>};</div>
</div>
<div class="line"><a id="l00620" name="l00620"></a><span class="lineno"> 620</span> </div>
<div class="line"><a id="l00624" name="l00624"></a><span class="lineno"> 624</span> </div>
<div class="line"><a id="l00625" name="l00625"></a><span class="lineno"> 625</span><span class="keyword">template</span> &lt;</div>
<div class="line"><a id="l00626" name="l00626"></a><span class="lineno"> 626</span> <span class="keyword">typename</span> T,</div>
<div class="line"><a id="l00627" name="l00627"></a><span class="lineno"> 627</span> <span class="keyword">typename</span> out_mask_t,</div>
<div class="line"><a id="l00628" name="l00628"></a><span class="lineno"> 628</span> <span class="keyword">typename</span> op_mask_t,</div>
<div class="line"><a id="l00629" name="l00629"></a><span class="lineno"> 629</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BM, <span class="comment">/* Threadgroup rows (in simdgroups) */</span></div>
<div class="line"><a id="l00630" name="l00630"></a><span class="lineno"> 630</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BN, <span class="comment">/* Threadgroup cols (in simdgroups) */</span></div>
<div class="line"><a id="l00631" name="l00631"></a><span class="lineno"> 631</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SM, <span class="comment">/* Simdgroup rows (in threads) */</span></div>
<div class="line"><a id="l00632" name="l00632"></a><span class="lineno"> 632</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SN, <span class="comment">/* Simdgroup cols (in threads) */</span></div>
<div class="line"><a id="l00633" name="l00633"></a><span class="lineno"> 633</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TM, <span class="comment">/* Thread rows (in elements) */</span></div>
<div class="line"><a id="l00634" name="l00634"></a><span class="lineno"> 634</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TN, <span class="comment">/* Thread cols (in elements) */</span></div>
<div class="line"><a id="l00635" name="l00635"></a><span class="lineno"> 635</span> <span class="keyword">const</span> <span class="keywordtype">bool</span> kDoNCBatch&gt; <span class="comment">/* Batch ndim &gt; 1 */</span></div>
<div class="foldopen" id="foldopen00636" data-start="{" data-end="}">
<div class="line"><a id="l00636" name="l00636"></a><span class="lineno"><a class="line" href="kernels_2gemv__masked_8h.html#ab3070d14cdecb1dd7dc220a551da6b7b"> 636</a></span>[[kernel, max_total_threads_per_threadgroup(BM* BN * 32)]] <span class="keywordtype">void</span> <a class="code hl_function" href="kernels_2gemv__masked_8h.html#ab3070d14cdecb1dd7dc220a551da6b7b">gemv_masked</a>(</div>
<div class="line"><a id="l00637" name="l00637"></a><span class="lineno"> 637</span> <span class="keyword">const</span> device T* mat [[buffer(0)]],</div>
<div class="line"><a id="l00638" name="l00638"></a><span class="lineno"> 638</span> <span class="keyword">const</span> device T* in_vec [[buffer(1)]],</div>
<div class="line"><a id="l00639" name="l00639"></a><span class="lineno"> 639</span> device T* out_vec [[buffer(3)]],</div>
<div class="line"><a id="l00640" name="l00640"></a><span class="lineno"> 640</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; in_vec_size [[buffer(4)]],</div>
<div class="line"><a id="l00641" name="l00641"></a><span class="lineno"> 641</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; out_vec_size [[buffer(5)]],</div>
<div class="line"><a id="l00642" name="l00642"></a><span class="lineno"> 642</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; marix_ld [[buffer(6)]],</div>
<div class="line"><a id="l00643" name="l00643"></a><span class="lineno"> 643</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; batch_ndim [[buffer(9)]],</div>
<div class="line"><a id="l00644" name="l00644"></a><span class="lineno"> 644</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* batch_shape [[buffer(10)]],</div>
<div class="line"><a id="l00645" name="l00645"></a><span class="lineno"> 645</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* vector_batch_stride [[buffer(11)]],</div>
<div class="line"><a id="l00646" name="l00646"></a><span class="lineno"> 646</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* matrix_batch_stride [[buffer(12)]],</div>
<div class="line"><a id="l00647" name="l00647"></a><span class="lineno"> 647</span> <span class="keyword">const</span> device out_mask_t* out_mask [[buffer(20)]],</div>
<div class="line"><a id="l00648" name="l00648"></a><span class="lineno"> 648</span> <span class="keyword">const</span> device op_mask_t* mat_mask [[buffer(21)]],</div>
<div class="line"><a id="l00649" name="l00649"></a><span class="lineno"> 649</span> <span class="keyword">const</span> device op_mask_t* vec_mask [[buffer(22)]],</div>
<div class="line"><a id="l00650" name="l00650"></a><span class="lineno"> 650</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* mask_strides [[buffer(23)]],</div>
<div class="line"><a id="l00651" name="l00651"></a><span class="lineno"> 651</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* mask_batch_strides [[buffer(24)]],</div>
<div class="line"><a id="l00652" name="l00652"></a><span class="lineno"> 652</span> uint3 tid [[threadgroup_position_in_grid]],</div>
<div class="line"><a id="l00653" name="l00653"></a><span class="lineno"> 653</span> uint3 lid [[thread_position_in_threadgroup]],</div>
<div class="line"><a id="l00654" name="l00654"></a><span class="lineno"> 654</span> uint simd_gid [[simdgroup_index_in_threadgroup]],</div>
<div class="line"><a id="l00655" name="l00655"></a><span class="lineno"> 655</span> uint simd_lid [[thread_index_in_simdgroup]]) {</div>
<div class="line"><a id="l00656" name="l00656"></a><span class="lineno"> 656</span> <span class="keyword">using </span>gemv_kernel =</div>
<div class="line"><a id="l00657" name="l00657"></a><span class="lineno"> 657</span> <a class="code hl_struct" href="struct_g_e_m_v_kernel.html">GEMVKernel&lt;T, out_mask_t, op_mask_t, BM, BN, SM, SN, TM, TN&gt;</a>;</div>
<div class="line"><a id="l00658" name="l00658"></a><span class="lineno"> 658</span> threadgroup T tgp_memory</div>
<div class="line"><a id="l00659" name="l00659"></a><span class="lineno"> 659</span> [gemv_kernel::tgp_mem_size == 0 ? 1 : gemv_kernel::tgp_mem_size];</div>
<div class="line"><a id="l00660" name="l00660"></a><span class="lineno"> 660</span> </div>
<div class="line"><a id="l00661" name="l00661"></a><span class="lineno"> 661</span> <span class="keyword">constexpr</span> <span class="keywordtype">bool</span> has_operand_mask = !metal::is_same_v&lt;op_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00662" name="l00662"></a><span class="lineno"> 662</span> <span class="keyword">constexpr</span> <span class="keywordtype">bool</span> has_output_mask = !metal::is_same_v&lt;out_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00663" name="l00663"></a><span class="lineno"> 663</span> </div>
<div class="line"><a id="l00664" name="l00664"></a><span class="lineno"> 664</span> <span class="comment">// Update batch offsets</span></div>
<div class="line"><a id="l00665" name="l00665"></a><span class="lineno"> 665</span> <span class="keywordflow">if</span> (kDoNCBatch) {</div>
<div class="line"><a id="l00666" name="l00666"></a><span class="lineno"> 666</span> in_vec += <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.z, batch_shape, vector_batch_stride, batch_ndim);</div>
<div class="line"><a id="l00667" name="l00667"></a><span class="lineno"> 667</span> mat += <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.z, batch_shape, matrix_batch_stride, batch_ndim);</div>
<div class="line"><a id="l00668" name="l00668"></a><span class="lineno"> 668</span> </div>
<div class="line"><a id="l00669" name="l00669"></a><span class="lineno"> 669</span> <span class="keywordflow">if</span> (has_output_mask) {</div>
<div class="line"><a id="l00670" name="l00670"></a><span class="lineno"> 670</span> out_mask +=</div>
<div class="line"><a id="l00671" name="l00671"></a><span class="lineno"> 671</span> <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.z, batch_shape, mask_batch_strides, batch_ndim);</div>
<div class="line"><a id="l00672" name="l00672"></a><span class="lineno"> 672</span> mask_batch_strides += batch_ndim;</div>
<div class="line"><a id="l00673" name="l00673"></a><span class="lineno"> 673</span> }</div>
<div class="line"><a id="l00674" name="l00674"></a><span class="lineno"> 674</span> </div>
<div class="line"><a id="l00675" name="l00675"></a><span class="lineno"> 675</span> <span class="keywordflow">if</span> (has_operand_mask) {</div>
<div class="line"><a id="l00676" name="l00676"></a><span class="lineno"> 676</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* mask_strides_mat = mask_batch_strides;</div>
<div class="line"><a id="l00677" name="l00677"></a><span class="lineno"> 677</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* mask_strides_vec = mask_strides_mat + batch_ndim;</div>
<div class="line"><a id="l00678" name="l00678"></a><span class="lineno"> 678</span> </div>
<div class="line"><a id="l00679" name="l00679"></a><span class="lineno"> 679</span> ulong2 batch_offsets = <a class="code hl_function" href="backend_2metal_2kernels_2steel_2utils_8h.html#aaf4974425147d6f26d031691e321637f">elem_to_loc_broadcast</a>(</div>
<div class="line"><a id="l00680" name="l00680"></a><span class="lineno"> 680</span> tid.z, batch_shape, mask_strides_mat, mask_strides_vec, batch_ndim);</div>
<div class="line"><a id="l00681" name="l00681"></a><span class="lineno"> 681</span> </div>
<div class="line"><a id="l00682" name="l00682"></a><span class="lineno"> 682</span> mat_mask += batch_offsets.x;</div>
<div class="line"><a id="l00683" name="l00683"></a><span class="lineno"> 683</span> vec_mask += batch_offsets.y;</div>
<div class="line"><a id="l00684" name="l00684"></a><span class="lineno"> 684</span> }</div>
<div class="line"><a id="l00685" name="l00685"></a><span class="lineno"> 685</span> </div>
<div class="line"><a id="l00686" name="l00686"></a><span class="lineno"> 686</span> } <span class="keywordflow">else</span> {</div>
<div class="line"><a id="l00687" name="l00687"></a><span class="lineno"> 687</span> in_vec += tid.z * vector_batch_stride[0];</div>
<div class="line"><a id="l00688" name="l00688"></a><span class="lineno"> 688</span> mat += tid.z * matrix_batch_stride[0];</div>
<div class="line"><a id="l00689" name="l00689"></a><span class="lineno"> 689</span> </div>
<div class="line"><a id="l00690" name="l00690"></a><span class="lineno"> 690</span> <span class="keywordflow">if</span> (has_output_mask) {</div>
<div class="line"><a id="l00691" name="l00691"></a><span class="lineno"> 691</span> out_mask += tid.z * mask_batch_strides[0];</div>
<div class="line"><a id="l00692" name="l00692"></a><span class="lineno"> 692</span> mask_batch_strides += batch_ndim;</div>
<div class="line"><a id="l00693" name="l00693"></a><span class="lineno"> 693</span> }</div>
<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> <span class="keywordflow">if</span> (has_operand_mask) {</div>
<div class="line"><a id="l00696" name="l00696"></a><span class="lineno"> 696</span> mat_mask += tid.z * mask_batch_strides[0];</div>
<div class="line"><a id="l00697" name="l00697"></a><span class="lineno"> 697</span> vec_mask += tid.z * mask_batch_strides[batch_ndim];</div>
<div class="line"><a id="l00698" name="l00698"></a><span class="lineno"> 698</span> }</div>
<div class="line"><a id="l00699" name="l00699"></a><span class="lineno"> 699</span> }</div>
<div class="line"><a id="l00700" name="l00700"></a><span class="lineno"> 700</span> </div>
<div class="line"><a id="l00701" name="l00701"></a><span class="lineno"> 701</span> out_vec += tid.z * out_vec_size;</div>
<div class="line"><a id="l00702" name="l00702"></a><span class="lineno"> 702</span> </div>
<div class="line"><a id="l00703" name="l00703"></a><span class="lineno"> 703</span> gemv_kernel::run(</div>
<div class="line"><a id="l00704" name="l00704"></a><span class="lineno"> 704</span> mat,</div>
<div class="line"><a id="l00705" name="l00705"></a><span class="lineno"> 705</span> in_vec,</div>
<div class="line"><a id="l00706" name="l00706"></a><span class="lineno"> 706</span> out_vec,</div>
<div class="line"><a id="l00707" name="l00707"></a><span class="lineno"> 707</span> in_vec_size,</div>
<div class="line"><a id="l00708" name="l00708"></a><span class="lineno"> 708</span> out_vec_size,</div>
<div class="line"><a id="l00709" name="l00709"></a><span class="lineno"> 709</span> marix_ld,</div>
<div class="line"><a id="l00710" name="l00710"></a><span class="lineno"> 710</span> out_mask,</div>
<div class="line"><a id="l00711" name="l00711"></a><span class="lineno"> 711</span> mat_mask,</div>
<div class="line"><a id="l00712" name="l00712"></a><span class="lineno"> 712</span> vec_mask,</div>
<div class="line"><a id="l00713" name="l00713"></a><span class="lineno"> 713</span> mask_strides,</div>
<div class="line"><a id="l00714" name="l00714"></a><span class="lineno"> 714</span> gemv_kernel::tgp_mem_size == 0 ? <span class="keyword">nullptr</span> : tgp_memory,</div>
<div class="line"><a id="l00715" name="l00715"></a><span class="lineno"> 715</span> tid,</div>
<div class="line"><a id="l00716" name="l00716"></a><span class="lineno"> 716</span> lid,</div>
<div class="line"><a id="l00717" name="l00717"></a><span class="lineno"> 717</span> simd_gid,</div>
<div class="line"><a id="l00718" name="l00718"></a><span class="lineno"> 718</span> simd_lid);</div>
<div class="line"><a id="l00719" name="l00719"></a><span class="lineno"> 719</span>}</div>
</div>
<div class="line"><a id="l00720" name="l00720"></a><span class="lineno"> 720</span> </div>
<div class="line"><a id="l00724" name="l00724"></a><span class="lineno"> 724</span> </div>
<div class="line"><a id="l00725" name="l00725"></a><span class="lineno"> 725</span><span class="keyword">template</span> &lt;</div>
<div class="line"><a id="l00726" name="l00726"></a><span class="lineno"> 726</span> <span class="keyword">typename</span> T,</div>
<div class="line"><a id="l00727" name="l00727"></a><span class="lineno"> 727</span> <span class="keyword">typename</span> out_mask_t,</div>
<div class="line"><a id="l00728" name="l00728"></a><span class="lineno"> 728</span> <span class="keyword">typename</span> op_mask_t,</div>
<div class="line"><a id="l00729" name="l00729"></a><span class="lineno"> 729</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BM, <span class="comment">/* Threadgroup rows (in simdgroups) */</span></div>
<div class="line"><a id="l00730" name="l00730"></a><span class="lineno"> 730</span> <span class="keyword">const</span> <span class="keywordtype">int</span> BN, <span class="comment">/* Threadgroup cols (in simdgroups) */</span></div>
<div class="line"><a id="l00731" name="l00731"></a><span class="lineno"> 731</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SM, <span class="comment">/* Simdgroup rows (in threads) */</span></div>
<div class="line"><a id="l00732" name="l00732"></a><span class="lineno"> 732</span> <span class="keyword">const</span> <span class="keywordtype">int</span> SN, <span class="comment">/* Simdgroup cols (in threads) */</span></div>
<div class="line"><a id="l00733" name="l00733"></a><span class="lineno"> 733</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TM, <span class="comment">/* Thread rows (in elements) */</span></div>
<div class="line"><a id="l00734" name="l00734"></a><span class="lineno"> 734</span> <span class="keyword">const</span> <span class="keywordtype">int</span> TN, <span class="comment">/* Thread cols (in elements) */</span></div>
<div class="line"><a id="l00735" name="l00735"></a><span class="lineno"> 735</span> <span class="keyword">const</span> <span class="keywordtype">bool</span> kDoNCBatch&gt; <span class="comment">/* Batch ndim &gt; 1 */</span></div>
<div class="foldopen" id="foldopen00736" data-start="{" data-end="}">
<div class="line"><a id="l00736" name="l00736"></a><span class="lineno"><a class="line" href="kernels_2gemv__masked_8h.html#a0c8d353fc453e448b2d0ed9a19431b63"> 736</a></span>[[kernel, max_total_threads_per_threadgroup(BM* BN * 32)]] <span class="keywordtype">void</span> <a class="code hl_function" href="kernels_2gemv__masked_8h.html#a0c8d353fc453e448b2d0ed9a19431b63">gemv_t_masked</a>(</div>
<div class="line"><a id="l00737" name="l00737"></a><span class="lineno"> 737</span> <span class="keyword">const</span> device T* mat [[buffer(0)]],</div>
<div class="line"><a id="l00738" name="l00738"></a><span class="lineno"> 738</span> <span class="keyword">const</span> device T* in_vec [[buffer(1)]],</div>
<div class="line"><a id="l00739" name="l00739"></a><span class="lineno"> 739</span> device T* out_vec [[buffer(3)]],</div>
<div class="line"><a id="l00740" name="l00740"></a><span class="lineno"> 740</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; in_vec_size [[buffer(4)]],</div>
<div class="line"><a id="l00741" name="l00741"></a><span class="lineno"> 741</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; out_vec_size [[buffer(5)]],</div>
<div class="line"><a id="l00742" name="l00742"></a><span class="lineno"> 742</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; marix_ld [[buffer(6)]],</div>
<div class="line"><a id="l00743" name="l00743"></a><span class="lineno"> 743</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>&amp; batch_ndim [[buffer(9)]],</div>
<div class="line"><a id="l00744" name="l00744"></a><span class="lineno"> 744</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* batch_shape [[buffer(10)]],</div>
<div class="line"><a id="l00745" name="l00745"></a><span class="lineno"> 745</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* vector_batch_stride [[buffer(11)]],</div>
<div class="line"><a id="l00746" name="l00746"></a><span class="lineno"> 746</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* matrix_batch_stride [[buffer(12)]],</div>
<div class="line"><a id="l00747" name="l00747"></a><span class="lineno"> 747</span> <span class="keyword">const</span> device out_mask_t* out_mask [[buffer(20)]],</div>
<div class="line"><a id="l00748" name="l00748"></a><span class="lineno"> 748</span> <span class="keyword">const</span> device op_mask_t* mat_mask [[buffer(21)]],</div>
<div class="line"><a id="l00749" name="l00749"></a><span class="lineno"> 749</span> <span class="keyword">const</span> device op_mask_t* vec_mask [[buffer(22)]],</div>
<div class="line"><a id="l00750" name="l00750"></a><span class="lineno"> 750</span> <span class="keyword">const</span> constant <span class="keywordtype">int</span>* mask_strides [[buffer(23)]],</div>
<div class="line"><a id="l00751" name="l00751"></a><span class="lineno"> 751</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* mask_batch_strides [[buffer(24)]],</div>
<div class="line"><a id="l00752" name="l00752"></a><span class="lineno"> 752</span> uint3 tid [[threadgroup_position_in_grid]],</div>
<div class="line"><a id="l00753" name="l00753"></a><span class="lineno"> 753</span> uint3 lid [[thread_position_in_threadgroup]],</div>
<div class="line"><a id="l00754" name="l00754"></a><span class="lineno"> 754</span> uint simd_gid [[simdgroup_index_in_threadgroup]],</div>
<div class="line"><a id="l00755" name="l00755"></a><span class="lineno"> 755</span> uint simd_lid [[thread_index_in_simdgroup]]) {</div>
<div class="line"><a id="l00756" name="l00756"></a><span class="lineno"> 756</span> <span class="keyword">using </span>gemv_kernel =</div>
<div class="line"><a id="l00757" name="l00757"></a><span class="lineno"> 757</span> <a class="code hl_struct" href="struct_g_e_m_v_t_kernel.html">GEMVTKernel&lt;T, out_mask_t, op_mask_t, BM, BN, SM, SN, TM, TN&gt;</a>;</div>
<div class="line"><a id="l00758" name="l00758"></a><span class="lineno"> 758</span> threadgroup T tgp_memory</div>
<div class="line"><a id="l00759" name="l00759"></a><span class="lineno"> 759</span> [gemv_kernel::tgp_mem_size == 0 ? 1 : gemv_kernel::tgp_mem_size];</div>
<div class="line"><a id="l00760" name="l00760"></a><span class="lineno"> 760</span> </div>
<div class="line"><a id="l00761" name="l00761"></a><span class="lineno"> 761</span> <span class="keyword">constexpr</span> <span class="keywordtype">bool</span> has_operand_mask = !metal::is_same_v&lt;op_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00762" name="l00762"></a><span class="lineno"> 762</span> <span class="keyword">constexpr</span> <span class="keywordtype">bool</span> has_output_mask = !metal::is_same_v&lt;out_mask_t, nomask_t&gt;;</div>
<div class="line"><a id="l00763" name="l00763"></a><span class="lineno"> 763</span> </div>
<div class="line"><a id="l00764" name="l00764"></a><span class="lineno"> 764</span> <span class="comment">// Update batch offsets</span></div>
<div class="line"><a id="l00765" name="l00765"></a><span class="lineno"> 765</span> <span class="keywordflow">if</span> (kDoNCBatch) {</div>
<div class="line"><a id="l00766" name="l00766"></a><span class="lineno"> 766</span> in_vec += <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.z, batch_shape, vector_batch_stride, batch_ndim);</div>
<div class="line"><a id="l00767" name="l00767"></a><span class="lineno"> 767</span> mat += <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.z, batch_shape, matrix_batch_stride, batch_ndim);</div>
<div class="line"><a id="l00768" name="l00768"></a><span class="lineno"> 768</span> </div>
<div class="line"><a id="l00769" name="l00769"></a><span class="lineno"> 769</span> <span class="keywordflow">if</span> (has_output_mask) {</div>
<div class="line"><a id="l00770" name="l00770"></a><span class="lineno"> 770</span> out_mask +=</div>
<div class="line"><a id="l00771" name="l00771"></a><span class="lineno"> 771</span> <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a22eaa505dbc7dd2a63a895f2e16712f5">elem_to_loc</a>(tid.z, batch_shape, mask_batch_strides, batch_ndim);</div>
<div class="line"><a id="l00772" name="l00772"></a><span class="lineno"> 772</span> mask_batch_strides += batch_ndim;</div>
<div class="line"><a id="l00773" name="l00773"></a><span class="lineno"> 773</span> }</div>
<div class="line"><a id="l00774" name="l00774"></a><span class="lineno"> 774</span> </div>
<div class="line"><a id="l00775" name="l00775"></a><span class="lineno"> 775</span> <span class="keywordflow">if</span> (has_operand_mask) {</div>
<div class="line"><a id="l00776" name="l00776"></a><span class="lineno"> 776</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* mask_strides_mat = mask_batch_strides;</div>
<div class="line"><a id="l00777" name="l00777"></a><span class="lineno"> 777</span> <span class="keyword">const</span> constant <span class="keywordtype">size_t</span>* mask_strides_vec = mask_strides_mat + batch_ndim;</div>
<div class="line"><a id="l00778" name="l00778"></a><span class="lineno"> 778</span> </div>
<div class="line"><a id="l00779" name="l00779"></a><span class="lineno"> 779</span> ulong2 batch_offsets = <a class="code hl_function" href="backend_2metal_2kernels_2steel_2utils_8h.html#aaf4974425147d6f26d031691e321637f">elem_to_loc_broadcast</a>(</div>
<div class="line"><a id="l00780" name="l00780"></a><span class="lineno"> 780</span> tid.z, batch_shape, mask_strides_mat, mask_strides_vec, batch_ndim);</div>
<div class="line"><a id="l00781" name="l00781"></a><span class="lineno"> 781</span> </div>
<div class="line"><a id="l00782" name="l00782"></a><span class="lineno"> 782</span> mat_mask += batch_offsets.x;</div>
<div class="line"><a id="l00783" name="l00783"></a><span class="lineno"> 783</span> vec_mask += batch_offsets.y;</div>
<div class="line"><a id="l00784" name="l00784"></a><span class="lineno"> 784</span> }</div>
<div class="line"><a id="l00785" name="l00785"></a><span class="lineno"> 785</span> </div>
<div class="line"><a id="l00786" name="l00786"></a><span class="lineno"> 786</span> } <span class="keywordflow">else</span> {</div>
<div class="line"><a id="l00787" name="l00787"></a><span class="lineno"> 787</span> in_vec += tid.z * vector_batch_stride[0];</div>
<div class="line"><a id="l00788" name="l00788"></a><span class="lineno"> 788</span> mat += tid.z * matrix_batch_stride[0];</div>
<div class="line"><a id="l00789" name="l00789"></a><span class="lineno"> 789</span> </div>
<div class="line"><a id="l00790" name="l00790"></a><span class="lineno"> 790</span> <span class="keywordflow">if</span> (has_output_mask) {</div>
<div class="line"><a id="l00791" name="l00791"></a><span class="lineno"> 791</span> out_mask += tid.z * mask_batch_strides[0];</div>
<div class="line"><a id="l00792" name="l00792"></a><span class="lineno"> 792</span> mask_batch_strides += batch_ndim;</div>
<div class="line"><a id="l00793" name="l00793"></a><span class="lineno"> 793</span> }</div>
<div class="line"><a id="l00794" name="l00794"></a><span class="lineno"> 794</span> </div>
<div class="line"><a id="l00795" name="l00795"></a><span class="lineno"> 795</span> <span class="keywordflow">if</span> (has_operand_mask) {</div>
<div class="line"><a id="l00796" name="l00796"></a><span class="lineno"> 796</span> mat_mask += tid.z * mask_batch_strides[0];</div>
<div class="line"><a id="l00797" name="l00797"></a><span class="lineno"> 797</span> vec_mask += tid.z * mask_batch_strides[batch_ndim];</div>
<div class="line"><a id="l00798" name="l00798"></a><span class="lineno"> 798</span> }</div>
<div class="line"><a id="l00799" name="l00799"></a><span class="lineno"> 799</span> }</div>
<div class="line"><a id="l00800" name="l00800"></a><span class="lineno"> 800</span> </div>
<div class="line"><a id="l00801" name="l00801"></a><span class="lineno"> 801</span> out_vec += tid.z * out_vec_size;</div>
<div class="line"><a id="l00802" name="l00802"></a><span class="lineno"> 802</span> </div>
<div class="line"><a id="l00803" name="l00803"></a><span class="lineno"> 803</span> gemv_kernel::run(</div>
<div class="line"><a id="l00804" name="l00804"></a><span class="lineno"> 804</span> mat,</div>
<div class="line"><a id="l00805" name="l00805"></a><span class="lineno"> 805</span> in_vec,</div>
<div class="line"><a id="l00806" name="l00806"></a><span class="lineno"> 806</span> out_vec,</div>
<div class="line"><a id="l00807" name="l00807"></a><span class="lineno"> 807</span> in_vec_size,</div>
<div class="line"><a id="l00808" name="l00808"></a><span class="lineno"> 808</span> out_vec_size,</div>
<div class="line"><a id="l00809" name="l00809"></a><span class="lineno"> 809</span> marix_ld,</div>
<div class="line"><a id="l00810" name="l00810"></a><span class="lineno"> 810</span> out_mask,</div>
<div class="line"><a id="l00811" name="l00811"></a><span class="lineno"> 811</span> mat_mask,</div>
<div class="line"><a id="l00812" name="l00812"></a><span class="lineno"> 812</span> vec_mask,</div>
<div class="line"><a id="l00813" name="l00813"></a><span class="lineno"> 813</span> mask_strides,</div>
<div class="line"><a id="l00814" name="l00814"></a><span class="lineno"> 814</span> gemv_kernel::tgp_mem_size == 0 ? <span class="keyword">nullptr</span> : tgp_memory,</div>
<div class="line"><a id="l00815" name="l00815"></a><span class="lineno"> 815</span> tid,</div>
<div class="line"><a id="l00816" name="l00816"></a><span class="lineno"> 816</span> lid,</div>
<div class="line"><a id="l00817" name="l00817"></a><span class="lineno"> 817</span> simd_gid,</div>
<div class="line"><a id="l00818" name="l00818"></a><span class="lineno"> 818</span> simd_lid);</div>
<div class="line"><a id="l00819" name="l00819"></a><span class="lineno"> 819</span>}</div>
</div>
<div class="ttc" id="abackend_2metal_2kernels_2steel_2utils_8h_html"><div class="ttname"><a href="backend_2metal_2kernels_2steel_2utils_8h.html">utils.h</a></div></div>
<div class="ttc" id="abackend_2metal_2kernels_2steel_2utils_8h_html_aaf4974425147d6f26d031691e321637f"><div class="ttname"><a href="backend_2metal_2kernels_2steel_2utils_8h.html#aaf4974425147d6f26d031691e321637f">elem_to_loc_broadcast</a></div><div class="ttdeci">METAL_FUNC ulong2 elem_to_loc_broadcast(uint elem, constant const int *shape, constant const size_t *a_strides, constant const size_t *b_strides, int ndim)</div><div class="ttdef"><b>Definition</b> utils.h:7</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="akernels_2gemv__masked_8h_html_a0386011c52d03e60885a31e6fbd903dd"><div class="ttname"><a href="kernels_2gemv__masked_8h.html#a0386011c52d03e60885a31e6fbd903dd">MLX_MTL_CONST</a></div><div class="ttdeci">#define MLX_MTL_CONST</div><div class="ttdef"><b>Definition</b> gemv_masked.h:7</div></div>
<div class="ttc" id="akernels_2gemv__masked_8h_html_a069b682d7d21827461544817d722bfd3"><div class="ttname"><a href="kernels_2gemv__masked_8h.html#a069b682d7d21827461544817d722bfd3">MLX_MTL_PRAGMA_UNROLL</a></div><div class="ttdeci">#define MLX_MTL_PRAGMA_UNROLL</div><div class="ttdef"><b>Definition</b> gemv_masked.h:8</div></div>
<div class="ttc" id="akernels_2gemv__masked_8h_html_a0c8d353fc453e448b2d0ed9a19431b63"><div class="ttname"><a href="kernels_2gemv__masked_8h.html#a0c8d353fc453e448b2d0ed9a19431b63">gemv_t_masked</a></div><div class="ttdeci">void gemv_t_masked(const device T *mat, const device T *in_vec, device T *out_vec, const constant int &amp;in_vec_size, const constant int &amp;out_vec_size, const constant int &amp;marix_ld, const constant int &amp;batch_ndim, const constant int *batch_shape, const constant size_t *vector_batch_stride, const constant size_t *matrix_batch_stride, const device out_mask_t *out_mask, const device op_mask_t *mat_mask, const device op_mask_t *vec_mask, const constant int *mask_strides, const constant size_t *mask_batch_strides, uint3 tid, uint3 lid, uint simd_gid, uint simd_lid)</div><div class="ttdoc">Vector matrix multiplication.</div><div class="ttdef"><b>Definition</b> gemv_masked.h:736</div></div>
<div class="ttc" id="akernels_2gemv__masked_8h_html_ab3070d14cdecb1dd7dc220a551da6b7b"><div class="ttname"><a href="kernels_2gemv__masked_8h.html#ab3070d14cdecb1dd7dc220a551da6b7b">gemv_masked</a></div><div class="ttdeci">void gemv_masked(const device T *mat, const device T *in_vec, device T *out_vec, const constant int &amp;in_vec_size, const constant int &amp;out_vec_size, const constant int &amp;marix_ld, const constant int &amp;batch_ndim, const constant int *batch_shape, const constant size_t *vector_batch_stride, const constant size_t *matrix_batch_stride, const device out_mask_t *out_mask, const device op_mask_t *mat_mask, const device op_mask_t *vec_mask, const constant int *mask_strides, const constant size_t *mask_batch_strides, uint3 tid, uint3 lid, uint simd_gid, uint simd_lid)</div><div class="ttdoc">Matrix vector multiplication.</div><div class="ttdef"><b>Definition</b> gemv_masked.h:636</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_af6e2dd7ae087aba6abac4f0350b7611c"><div class="ttname"><a href="namespacemetal.html#af6e2dd7ae087aba6abac4f0350b7611c">metal::simd_shuffle_down</a></div><div class="ttdeci">METAL_FUNC bfloat16_t simd_shuffle_down(bfloat16_t data, ushort delta)</div><div class="ttdef"><b>Definition</b> bf16_math.h:377</div></div>
<div class="ttc" id="astruct___no_mask_html"><div class="ttname"><a href="struct___no_mask.html">_NoMask</a></div><div class="ttdef"><b>Definition</b> gemv_masked.h:10</div></div>
<div class="ttc" id="astruct___no_mask_html_a0c4a4557d5c97ceafe3a2c4e521cdf7e"><div class="ttname"><a href="struct___no_mask.html#a0c4a4557d5c97ceafe3a2c4e521cdf7e">_NoMask::x</a></div><div class="ttdeci">char x</div><div class="ttdef"><b>Definition</b> gemv_masked.h:11</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html"><div class="ttname"><a href="struct_g_e_m_v_kernel.html">GEMVKernel</a></div><div class="ttdef"><b>Definition</b> gemv_masked.h:48</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a04bb72da9a93d6d1eba468fa311bbba7"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a04bb72da9a93d6d1eba468fa311bbba7">GEMVKernel::load_safe</a></div><div class="ttdeci">static METAL_FUNC void load_safe(const device T *src, thread T dst[TN], const int src_offset=0, const int src_size=TN)</div><div class="ttdef"><b>Definition</b> gemv_masked.h:102</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a0edbf2dd6a6563e7afa6dab6b670615c"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a0edbf2dd6a6563e7afa6dab6b670615c">GEMVKernel::has_mul_output_mask</a></div><div class="ttdeci">static constant constexpr const bool has_mul_output_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:68</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a1dd943fcbf5e7be435fc36bed589a641"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a1dd943fcbf5e7be435fc36bed589a641">GEMVKernel::threadsM</a></div><div class="ttdeci">static constant constexpr const int threadsM</div><div class="ttdef"><b>Definition</b> gemv_masked.h:49</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a2fef17f9c9aa0bdf530ad3554fb0988b"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a2fef17f9c9aa0bdf530ad3554fb0988b">GEMVKernel::blockN</a></div><div class="ttdeci">static constant constexpr const int blockN</div><div class="ttdef"><b>Definition</b> gemv_masked.h:53</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a47bfab7d21dd18760d3e0937ad36b19d"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a47bfab7d21dd18760d3e0937ad36b19d">GEMVKernel::threadsN</a></div><div class="ttdeci">static constant constexpr const int threadsN</div><div class="ttdef"><b>Definition</b> gemv_masked.h:50</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a6013e9c5b2f72fa1311dd038172df0ce"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a6013e9c5b2f72fa1311dd038172df0ce">GEMVKernel::load_unsafe</a></div><div class="ttdeci">static METAL_FUNC void load_unsafe(const device T *src, thread T dst[TN], const int src_offset=0)</div><div class="ttdef"><b>Definition</b> gemv_masked.h:95</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a7281520100658811076400060663903c"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a7281520100658811076400060663903c">GEMVKernel::blockM</a></div><div class="ttdeci">static constant constexpr const int blockM</div><div class="ttdef"><b>Definition</b> gemv_masked.h:52</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_a9ef4d0e62094d7033069f5dda5efb236"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#a9ef4d0e62094d7033069f5dda5efb236">GEMVKernel::tgp_mem_size</a></div><div class="ttdeci">static constant constexpr const short tgp_mem_size</div><div class="ttdef"><b>Definition</b> gemv_masked.h:91</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_ab00784dff1512a7b0919fcb4cfa5d50e"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#ab00784dff1512a7b0919fcb4cfa5d50e">GEMVKernel::has_operand_mask</a></div><div class="ttdeci">static constant constexpr const bool has_operand_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:63</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_ab8b64c94f4c8f6f09c0777415589b487"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#ab8b64c94f4c8f6f09c0777415589b487">GEMVKernel::has_output_mask</a></div><div class="ttdeci">static constant constexpr const bool has_output_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:64</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_ac4a7b5011a0ea938ab1949bb1767fc1a"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#ac4a7b5011a0ea938ab1949bb1767fc1a">GEMVKernel::run</a></div><div class="ttdeci">static METAL_FUNC void run(const device T *mat, const device T *in_vec, device T *out_vec, const constant int &amp;in_vec_size, const constant int &amp;out_vec_size, const constant int &amp;matrix_ld, const device out_mask_t *out_mask, const device op_mask_t *mat_mask, const device op_mask_t *vec_mask, const constant int *mask_strides, threadgroup T *tgp_memory, uint3 tid, uint3 lid, uint simd_gid, uint simd_lid)</div><div class="ttdef"><b>Definition</b> gemv_masked.h:120</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_ad47223ee49b3cb7bf3746a2cec45f883"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#ad47223ee49b3cb7bf3746a2cec45f883">GEMVKernel::has_mul_operand_mask</a></div><div class="ttdeci">static constant constexpr const bool has_mul_operand_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:66</div></div>
<div class="ttc" id="astruct_g_e_m_v_kernel_html_ae8113fddf6fb637acfd12efd978b704c"><div class="ttname"><a href="struct_g_e_m_v_kernel.html#ae8113fddf6fb637acfd12efd978b704c">GEMVKernel::needs_tgp_reduction</a></div><div class="ttdeci">static constant constexpr const bool needs_tgp_reduction</div><div class="ttdef"><b>Definition</b> gemv_masked.h:92</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html">GEMVTKernel</a></div><div class="ttdoc">Vector matrix multiplication.</div><div class="ttdef"><b>Definition</b> gemv_masked.h:358</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a2ae8ce535d59cccf453381b4485a77f0"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a2ae8ce535d59cccf453381b4485a77f0">GEMVTKernel::blockM</a></div><div class="ttdeci">static constant constexpr const int blockM</div><div class="ttdef"><b>Definition</b> gemv_masked.h:362</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a48a09a21d7b822f380d040c752b785d7"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a48a09a21d7b822f380d040c752b785d7">GEMVTKernel::tgp_mem_size</a></div><div class="ttdeci">static constant constexpr const short tgp_mem_size</div><div class="ttdef"><b>Definition</b> gemv_masked.h:394</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a4a53e73a581aa8881b1f86ce653519e6"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a4a53e73a581aa8881b1f86ce653519e6">GEMVTKernel::threadsM</a></div><div class="ttdeci">static constant constexpr const int threadsM</div><div class="ttdef"><b>Definition</b> gemv_masked.h:359</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a5d68656832de892f33db939005713927"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a5d68656832de892f33db939005713927">GEMVTKernel::run</a></div><div class="ttdeci">static METAL_FUNC void run(const device T *mat, const device T *in_vec, device T *out_vec, const constant int &amp;in_vec_size, const constant int &amp;out_vec_size, const constant int &amp;marix_ld, const device out_mask_t *out_mask, const device op_mask_t *mat_mask, const device op_mask_t *vec_mask, const constant int *mask_strides, threadgroup T *tgp_memory, uint3 tid, uint3 lid, uint simd_gid, uint simd_lid)</div><div class="ttdef"><b>Definition</b> gemv_masked.h:397</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a60be87666006ba0bf88bc8e6902da42a"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a60be87666006ba0bf88bc8e6902da42a">GEMVTKernel::blockN</a></div><div class="ttdeci">static constant constexpr const int blockN</div><div class="ttdef"><b>Definition</b> gemv_masked.h:363</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a6729d6e63e76a1e9c7c8e78d9aac4869"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a6729d6e63e76a1e9c7c8e78d9aac4869">GEMVTKernel::has_operand_mask</a></div><div class="ttdeci">static constant constexpr const bool has_operand_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:367</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a67be7ec69c3791f02e97ccdb00ae0e03"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a67be7ec69c3791f02e97ccdb00ae0e03">GEMVTKernel::needs_tgp_reduction</a></div><div class="ttdeci">static constant constexpr const bool needs_tgp_reduction</div><div class="ttdef"><b>Definition</b> gemv_masked.h:395</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a8db6f01f96a36b216acd801c34a96ef5"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a8db6f01f96a36b216acd801c34a96ef5">GEMVTKernel::has_mul_operand_mask</a></div><div class="ttdeci">static constant constexpr const bool has_mul_operand_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:370</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_a8eb06f6569e4042e24fee220b11fa10d"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#a8eb06f6569e4042e24fee220b11fa10d">GEMVTKernel::has_mul_output_mask</a></div><div class="ttdeci">static constant constexpr const bool has_mul_output_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:372</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_aaefdf8f023da255bbb70a0c3e3408626"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#aaefdf8f023da255bbb70a0c3e3408626">GEMVTKernel::has_output_mask</a></div><div class="ttdeci">static constant constexpr const bool has_output_mask</div><div class="ttdef"><b>Definition</b> gemv_masked.h:368</div></div>
<div class="ttc" id="astruct_g_e_m_v_t_kernel_html_ade6f15a9744616de9dd71498ad7e758d"><div class="ttname"><a href="struct_g_e_m_v_t_kernel.html#ade6f15a9744616de9dd71498ad7e758d">GEMVTKernel::threadsN</a></div><div class="ttdeci">static constant constexpr const int threadsN</div><div class="ttdef"><b>Definition</b> gemv_masked.h:360</div></div>
<div class="ttc" id="astruct_scale_op_html"><div class="ttname"><a href="struct_scale_op.html">ScaleOp</a></div><div class="ttdef"><b>Definition</b> gemv_masked.h:30</div></div>
<div class="ttc" id="astruct_scale_op_html_a02043fac21c68fb8d6863a01f45ede4b"><div class="ttname"><a href="struct_scale_op.html#a02043fac21c68fb8d6863a01f45ede4b">ScaleOp::scale</a></div><div class="ttdeci">OutT scale</div><div class="ttdef"><b>Definition</b> gemv_masked.h:31</div></div>
<div class="ttc" id="astruct_scale_op_html_a69f82bc925843a4e1c14dfe8ad2f3218"><div class="ttname"><a href="struct_scale_op.html#a69f82bc925843a4e1c14dfe8ad2f3218">ScaleOp::apply</a></div><div class="ttdeci">METAL_FUNC OutT apply(InT x) const</div><div class="ttdef"><b>Definition</b> gemv_masked.h:33</div></div>
</div><!-- fragment --></div><!-- contents -->
<!-- start footer part -->
<hr class="footer"/><address class="footer"><small>
Generated by&#160;<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 -->
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink