docs update

This commit is contained in:
Awni Hannun
2024-08-10 09:24:35 -07:00
committed by CircleCI Docs
parent 6e9288a41c
commit f5dcb1c2b9
724 changed files with 14529 additions and 11046 deletions

View File

@@ -107,165 +107,191 @@ $(function() { codefold.init(0); });
<div class="line"><a id="l00018" name="l00018"></a><span class="lineno"> 18</span> </div>
<div class="line"><a id="l00019" name="l00019"></a><span class="lineno"> 19</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00020" data-start="{" data-end="}">
<div class="line"><a id="l00020" name="l00020"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77"> 20</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77">copy_g_nd1</a>(</div>
<div class="line"><a id="l00020" name="l00020"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a8023e9335cc5334847a8d315042be3a3"> 20</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a8023e9335cc5334847a8d315042be3a3">copy_s2</a>(</div>
<div class="line"><a id="l00021" name="l00021"></a><span class="lineno"> 21</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00022" name="l00022"></a><span class="lineno"> 22</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00023" name="l00023"></a><span class="lineno"> 23</span> constant <span class="keyword">const</span> int64_t&amp; src_stride [[buffer(3)]],</div>
<div class="line"><a id="l00024" name="l00024"></a><span class="lineno"> 24</span> uint index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00025" name="l00025"></a><span class="lineno"> 25</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, src_stride);</div>
<div class="line"><a id="l00026" name="l00026"></a><span class="lineno"> 26</span> dst[index] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00023" name="l00023"></a><span class="lineno"> 23</span> uint2 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00024" name="l00024"></a><span class="lineno"> 24</span> uint2 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00025" name="l00025"></a><span class="lineno"> 25</span> <span class="keywordtype">size_t</span> offset = index.x + grid_dim.x * size_t(index.y);</div>
<div class="line"><a id="l00026" name="l00026"></a><span class="lineno"> 26</span> dst[offset] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[0]);</div>
<div class="line"><a id="l00027" name="l00027"></a><span class="lineno"> 27</span>}</div>
</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> T, <span class="keyword">typename</span> U&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="metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c"> 30</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c">copy_g_nd2</a>(</div>
<div class="line"><a id="l00030" name="l00030"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#aee14a5326f53d9b30b0b38e27d180ef3"> 30</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#aee14a5326f53d9b30b0b38e27d180ef3">copy_v2</a>(</div>
<div class="line"><a id="l00031" name="l00031"></a><span class="lineno"> 31</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00032" name="l00032"></a><span class="lineno"> 32</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00033" name="l00033"></a><span class="lineno"> 33</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00034" name="l00034"></a><span class="lineno"> 34</span> uint2 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00035" name="l00035"></a><span class="lineno"> 35</span> uint2 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00036" name="l00036"></a><span class="lineno"> 36</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, src_strides);</div>
<div class="line"><a id="l00037" name="l00037"></a><span class="lineno"> 37</span> int64_t dst_idx = index.x + (int64_t)grid_dim.x * index.y;</div>
<div class="line"><a id="l00038" name="l00038"></a><span class="lineno"> 38</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00039" name="l00039"></a><span class="lineno"> 39</span>}</div>
<div class="line"><a id="l00033" name="l00033"></a><span class="lineno"> 33</span> uint2 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00034" name="l00034"></a><span class="lineno"> 34</span> uint2 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00035" name="l00035"></a><span class="lineno"> 35</span> <span class="keywordtype">size_t</span> offset = index.x + grid_dim.x * size_t(index.y);</div>
<div class="line"><a id="l00036" name="l00036"></a><span class="lineno"> 36</span> dst[offset] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[offset]);</div>
<div class="line"><a id="l00037" name="l00037"></a><span class="lineno"> 37</span>}</div>
</div>
<div class="line"><a id="l00040" name="l00040"></a><span class="lineno"> 40</span> </div>
<div class="line"><a id="l00041" name="l00041"></a><span class="lineno"> 41</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00042" data-start="{" data-end="}">
<div class="line"><a id="l00042" name="l00042"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff"> 42</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff">copy_g_nd3</a>(</div>
<div class="line"><a id="l00043" name="l00043"></a><span class="lineno"> 43</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00044" name="l00044"></a><span class="lineno"> 44</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00045" name="l00045"></a><span class="lineno"> 45</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00046" name="l00046"></a><span class="lineno"> 46</span> uint3 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00047" name="l00047"></a><span class="lineno"> 47</span> uint3 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00048" name="l00048"></a><span class="lineno"> 48</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, src_strides);</div>
<div class="line"><a id="l00049" name="l00049"></a><span class="lineno"> 49</span> int64_t dst_idx =</div>
<div class="line"><a id="l00050" name="l00050"></a><span class="lineno"> 50</span> index.x + (int64_t)grid_dim.x * (index.y + (int64_t)grid_dim.y * index.z);</div>
<div class="line"><a id="l00051" name="l00051"></a><span class="lineno"> 51</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00052" name="l00052"></a><span class="lineno"> 52</span>}</div>
<div class="line"><a id="l00038" name="l00038"></a><span class="lineno"> 38</span> </div>
<div class="line"><a id="l00039" name="l00039"></a><span class="lineno"> 39</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00040" data-start="{" data-end="}">
<div class="line"><a id="l00040" name="l00040"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77"> 40</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77">copy_g_nd1</a>(</div>
<div class="line"><a id="l00041" name="l00041"></a><span class="lineno"> 41</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00042" name="l00042"></a><span class="lineno"> 42</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00043" name="l00043"></a><span class="lineno"> 43</span> constant <span class="keyword">const</span> int64_t&amp; src_stride [[buffer(3)]],</div>
<div class="line"><a id="l00044" name="l00044"></a><span class="lineno"> 44</span> uint index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00045" name="l00045"></a><span class="lineno"> 45</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, src_stride);</div>
<div class="line"><a id="l00046" name="l00046"></a><span class="lineno"> 46</span> dst[index] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00047" name="l00047"></a><span class="lineno"> 47</span>}</div>
</div>
<div class="line"><a id="l00053" name="l00053"></a><span class="lineno"> 53</span> </div>
<div class="line"><a id="l00054" name="l00054"></a><span class="lineno"> 54</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keywordtype">int</span> DIM&gt;</div>
<div class="foldopen" id="foldopen00055" data-start="{" data-end="}">
<div class="line"><a id="l00055" name="l00055"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a07336a7a024125b4f6d3e8556d04058f"> 55</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a07336a7a024125b4f6d3e8556d04058f">copy_g_nd</a>(</div>
<div class="line"><a id="l00056" name="l00056"></a><span class="lineno"> 56</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00057" name="l00057"></a><span class="lineno"> 57</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00058" name="l00058"></a><span class="lineno"> 58</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00059" name="l00059"></a><span class="lineno"> 59</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00060" name="l00060"></a><span class="lineno"> 60</span> uint3 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00061" name="l00061"></a><span class="lineno"> 61</span> uint3 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00062" name="l00062"></a><span class="lineno"> 62</span> <span class="keyword">auto</span> src_idx = elem_to_loc_nd&lt;DIM&gt;(index, src_shape, src_strides);</div>
<div class="line"><a id="l00063" name="l00063"></a><span class="lineno"> 63</span> int64_t dst_idx =</div>
<div class="line"><a id="l00064" name="l00064"></a><span class="lineno"> 64</span> index.x + (int64_t)grid_dim.x * (index.y + (int64_t)grid_dim.y * index.z);</div>
<div class="line"><a id="l00065" name="l00065"></a><span class="lineno"> 65</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00066" name="l00066"></a><span class="lineno"> 66</span>}</div>
<div class="line"><a id="l00048" name="l00048"></a><span class="lineno"> 48</span> </div>
<div class="line"><a id="l00049" name="l00049"></a><span class="lineno"> 49</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00050" data-start="{" data-end="}">
<div class="line"><a id="l00050" name="l00050"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c"> 50</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c">copy_g_nd2</a>(</div>
<div class="line"><a id="l00051" name="l00051"></a><span class="lineno"> 51</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00052" name="l00052"></a><span class="lineno"> 52</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00053" name="l00053"></a><span class="lineno"> 53</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00054" name="l00054"></a><span class="lineno"> 54</span> uint2 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00055" name="l00055"></a><span class="lineno"> 55</span> uint2 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00056" name="l00056"></a><span class="lineno"> 56</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, src_strides);</div>
<div class="line"><a id="l00057" name="l00057"></a><span class="lineno"> 57</span> int64_t dst_idx = index.x + (int64_t)grid_dim.x * index.y;</div>
<div class="line"><a id="l00058" name="l00058"></a><span class="lineno"> 58</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00059" name="l00059"></a><span class="lineno"> 59</span>}</div>
</div>
<div class="line"><a id="l00067" name="l00067"></a><span class="lineno"> 67</span> </div>
<div class="line"><a id="l00068" name="l00068"></a><span class="lineno"> 68</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00069" data-start="{" data-end="}">
<div class="line"><a id="l00069" name="l00069"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a1ea80f8edb842dd0ed8ffe553a2238b2"> 69</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a1ea80f8edb842dd0ed8ffe553a2238b2">copy_g</a>(</div>
<div class="line"><a id="l00070" name="l00070"></a><span class="lineno"> 70</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00071" name="l00071"></a><span class="lineno"> 71</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00072" name="l00072"></a><span class="lineno"> 72</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00073" name="l00073"></a><span class="lineno"> 73</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00074" name="l00074"></a><span class="lineno"> 74</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>&amp; ndim [[buffer(5)]],</div>
<div class="line"><a id="l00075" name="l00075"></a><span class="lineno"> 75</span> uint3 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00076" name="l00076"></a><span class="lineno"> 76</span> uint3 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00077" name="l00077"></a><span class="lineno"> 77</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2e49fa7ab8f6348543455c6c45d7e2a9">elem_to_loc</a>(index, src_shape, src_strides, ndim);</div>
<div class="line"><a id="l00078" name="l00078"></a><span class="lineno"> 78</span> int64_t dst_idx =</div>
<div class="line"><a id="l00079" name="l00079"></a><span class="lineno"> 79</span> index.x + (int64_t)grid_dim.x * (index.y + (int64_t)grid_dim.y * index.z);</div>
<div class="line"><a id="l00080" name="l00080"></a><span class="lineno"> 80</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00081" name="l00081"></a><span class="lineno"> 81</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">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00062" data-start="{" data-end="}">
<div class="line"><a id="l00062" name="l00062"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff"> 62</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff">copy_g_nd3</a>(</div>
<div class="line"><a id="l00063" name="l00063"></a><span class="lineno"> 63</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00064" name="l00064"></a><span class="lineno"> 64</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00065" name="l00065"></a><span class="lineno"> 65</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00066" name="l00066"></a><span class="lineno"> 66</span> uint3 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00067" name="l00067"></a><span class="lineno"> 67</span> uint3 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00068" name="l00068"></a><span class="lineno"> 68</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, src_strides);</div>
<div class="line"><a id="l00069" name="l00069"></a><span class="lineno"> 69</span> int64_t dst_idx =</div>
<div class="line"><a id="l00070" name="l00070"></a><span class="lineno"> 70</span> index.x + (int64_t)grid_dim.x * (index.y + (int64_t)grid_dim.y * index.z);</div>
<div class="line"><a id="l00071" name="l00071"></a><span class="lineno"> 71</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00072" name="l00072"></a><span class="lineno"> 72</span>}</div>
</div>
<div class="line"><a id="l00082" name="l00082"></a><span class="lineno"> 82</span> </div>
<div class="line"><a id="l00083" name="l00083"></a><span class="lineno"> 83</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00084" data-start="{" data-end="}">
<div class="line"><a id="l00084" name="l00084"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1"> 84</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1">copy_gg_nd1</a>(</div>
<div class="line"><a id="l00085" name="l00085"></a><span class="lineno"> 85</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00086" name="l00086"></a><span class="lineno"> 86</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00087" name="l00087"></a><span class="lineno"> 87</span> constant <span class="keyword">const</span> int64_t&amp; src_stride [[buffer(3)]],</div>
<div class="line"><a id="l00088" name="l00088"></a><span class="lineno"> 88</span> constant <span class="keyword">const</span> int64_t&amp; dst_stride [[buffer(4)]],</div>
<div class="line"><a id="l00089" name="l00089"></a><span class="lineno"> 89</span> uint index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00090" name="l00090"></a><span class="lineno"> 90</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, src_stride);</div>
<div class="line"><a id="l00091" name="l00091"></a><span class="lineno"> 91</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, dst_stride);</div>
<div class="line"><a id="l00092" name="l00092"></a><span class="lineno"> 92</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00093" name="l00093"></a><span class="lineno"> 93</span>}</div>
<div class="line"><a id="l00073" name="l00073"></a><span class="lineno"> 73</span> </div>
<div class="line"><a id="l00074" name="l00074"></a><span class="lineno"> 74</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keywordtype">int</span> DIM&gt;</div>
<div class="foldopen" id="foldopen00075" data-start="{" data-end="}">
<div class="line"><a id="l00075" name="l00075"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a07336a7a024125b4f6d3e8556d04058f"> 75</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a07336a7a024125b4f6d3e8556d04058f">copy_g_nd</a>(</div>
<div class="line"><a id="l00076" name="l00076"></a><span class="lineno"> 76</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00077" name="l00077"></a><span class="lineno"> 77</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00078" name="l00078"></a><span class="lineno"> 78</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00079" name="l00079"></a><span class="lineno"> 79</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00080" name="l00080"></a><span class="lineno"> 80</span> uint3 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00081" name="l00081"></a><span class="lineno"> 81</span> uint3 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00082" name="l00082"></a><span class="lineno"> 82</span> <span class="keyword">auto</span> src_idx = elem_to_loc_nd&lt;DIM&gt;(index, src_shape, src_strides);</div>
<div class="line"><a id="l00083" name="l00083"></a><span class="lineno"> 83</span> int64_t dst_idx =</div>
<div class="line"><a id="l00084" name="l00084"></a><span class="lineno"> 84</span> index.x + (int64_t)grid_dim.x * (index.y + (int64_t)grid_dim.y * index.z);</div>
<div class="line"><a id="l00085" name="l00085"></a><span class="lineno"> 85</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00086" name="l00086"></a><span class="lineno"> 86</span>}</div>
</div>
<div class="line"><a id="l00094" name="l00094"></a><span class="lineno"> 94</span> </div>
<div class="line"><a id="l00095" name="l00095"></a><span class="lineno"> 95</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00096" data-start="{" data-end="}">
<div class="line"><a id="l00096" name="l00096"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950"> 96</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950">copy_gg_nd2</a>(</div>
<div class="line"><a id="l00097" name="l00097"></a><span class="lineno"> 97</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00098" name="l00098"></a><span class="lineno"> 98</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00099" name="l00099"></a><span class="lineno"> 99</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00100" name="l00100"></a><span class="lineno"> 100</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00101" name="l00101"></a><span class="lineno"> 101</span> uint2 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00102" name="l00102"></a><span class="lineno"> 102</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, src_strides);</div>
<div class="line"><a id="l00103" name="l00103"></a><span class="lineno"> 103</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, dst_strides);</div>
<div class="line"><a id="l00104" name="l00104"></a><span class="lineno"> 104</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00105" name="l00105"></a><span class="lineno"> 105</span>}</div>
<div class="line"><a id="l00087" name="l00087"></a><span class="lineno"> 87</span> </div>
<div class="line"><a id="l00088" name="l00088"></a><span class="lineno"> 88</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00089" data-start="{" data-end="}">
<div class="line"><a id="l00089" name="l00089"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a1ea80f8edb842dd0ed8ffe553a2238b2"> 89</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a1ea80f8edb842dd0ed8ffe553a2238b2">copy_g</a>(</div>
<div class="line"><a id="l00090" name="l00090"></a><span class="lineno"> 90</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00091" name="l00091"></a><span class="lineno"> 91</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00092" name="l00092"></a><span class="lineno"> 92</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00093" name="l00093"></a><span class="lineno"> 93</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00094" name="l00094"></a><span class="lineno"> 94</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>&amp; ndim [[buffer(5)]],</div>
<div class="line"><a id="l00095" name="l00095"></a><span class="lineno"> 95</span> uint3 index [[thread_position_in_grid]],</div>
<div class="line"><a id="l00096" name="l00096"></a><span class="lineno"> 96</span> uint3 grid_dim [[threads_per_grid]]) {</div>
<div class="line"><a id="l00097" name="l00097"></a><span class="lineno"> 97</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2e49fa7ab8f6348543455c6c45d7e2a9">elem_to_loc</a>(index, src_shape, src_strides, ndim);</div>
<div class="line"><a id="l00098" name="l00098"></a><span class="lineno"> 98</span> int64_t dst_idx =</div>
<div class="line"><a id="l00099" name="l00099"></a><span class="lineno"> 99</span> index.x + (int64_t)grid_dim.x * (index.y + (int64_t)grid_dim.y * index.z);</div>
<div class="line"><a id="l00100" name="l00100"></a><span class="lineno"> 100</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00101" name="l00101"></a><span class="lineno"> 101</span>}</div>
</div>
<div class="line"><a id="l00106" name="l00106"></a><span class="lineno"> 106</span> </div>
<div class="line"><a id="l00107" name="l00107"></a><span class="lineno"> 107</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00108" data-start="{" data-end="}">
<div class="line"><a id="l00108" name="l00108"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd"> 108</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd">copy_gg_nd3</a>(</div>
<div class="line"><a id="l00109" name="l00109"></a><span class="lineno"> 109</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00110" name="l00110"></a><span class="lineno"> 110</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00111" name="l00111"></a><span class="lineno"> 111</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00112" name="l00112"></a><span class="lineno"> 112</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00113" name="l00113"></a><span class="lineno"> 113</span> uint3 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00114" name="l00114"></a><span class="lineno"> 114</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, src_strides);</div>
<div class="line"><a id="l00115" name="l00115"></a><span class="lineno"> 115</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, dst_strides);</div>
<div class="line"><a id="l00116" name="l00116"></a><span class="lineno"> 116</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00117" name="l00117"></a><span class="lineno"> 117</span>}</div>
<div class="line"><a id="l00102" name="l00102"></a><span class="lineno"> 102</span> </div>
<div class="line"><a id="l00103" name="l00103"></a><span class="lineno"> 103</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00104" data-start="{" data-end="}">
<div class="line"><a id="l00104" name="l00104"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1"> 104</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1">copy_gg_nd1</a>(</div>
<div class="line"><a id="l00105" name="l00105"></a><span class="lineno"> 105</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00106" name="l00106"></a><span class="lineno"> 106</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00107" name="l00107"></a><span class="lineno"> 107</span> constant <span class="keyword">const</span> int64_t&amp; src_stride [[buffer(3)]],</div>
<div class="line"><a id="l00108" name="l00108"></a><span class="lineno"> 108</span> constant <span class="keyword">const</span> int64_t&amp; dst_stride [[buffer(4)]],</div>
<div class="line"><a id="l00109" name="l00109"></a><span class="lineno"> 109</span> uint index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00110" name="l00110"></a><span class="lineno"> 110</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, src_stride);</div>
<div class="line"><a id="l00111" name="l00111"></a><span class="lineno"> 111</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, dst_stride);</div>
<div class="line"><a id="l00112" name="l00112"></a><span class="lineno"> 112</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00113" name="l00113"></a><span class="lineno"> 113</span>}</div>
</div>
<div class="line"><a id="l00118" name="l00118"></a><span class="lineno"> 118</span> </div>
<div class="line"><a id="l00119" name="l00119"></a><span class="lineno"> 119</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keywordtype">int</span> DIM&gt;</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="metal_2kernels_2copy_8h.html#a3197d34e9c68e7059419ae0635667059"> 120</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a3197d34e9c68e7059419ae0635667059">copy_gg_nd</a>(</div>
<div class="line"><a id="l00121" name="l00121"></a><span class="lineno"> 121</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00122" name="l00122"></a><span class="lineno"> 122</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00123" name="l00123"></a><span class="lineno"> 123</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00124" name="l00124"></a><span class="lineno"> 124</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00125" name="l00125"></a><span class="lineno"> 125</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00126" name="l00126"></a><span class="lineno"> 126</span> uint3 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00127" name="l00127"></a><span class="lineno"> 127</span> <span class="keyword">auto</span> src_idx = elem_to_loc_nd&lt;DIM&gt;(index, src_shape, src_strides);</div>
<div class="line"><a id="l00128" name="l00128"></a><span class="lineno"> 128</span> <span class="keyword">auto</span> dst_idx = elem_to_loc_nd&lt;DIM&gt;(index, src_shape, dst_strides);</div>
<div class="line"><a id="l00129" name="l00129"></a><span class="lineno"> 129</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00130" name="l00130"></a><span class="lineno"> 130</span>}</div>
<div class="line"><a id="l00114" name="l00114"></a><span class="lineno"> 114</span> </div>
<div class="line"><a id="l00115" name="l00115"></a><span class="lineno"> 115</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00116" data-start="{" data-end="}">
<div class="line"><a id="l00116" name="l00116"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950"> 116</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950">copy_gg_nd2</a>(</div>
<div class="line"><a id="l00117" name="l00117"></a><span class="lineno"> 117</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00118" name="l00118"></a><span class="lineno"> 118</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00119" name="l00119"></a><span class="lineno"> 119</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00120" name="l00120"></a><span class="lineno"> 120</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00121" name="l00121"></a><span class="lineno"> 121</span> uint2 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00122" name="l00122"></a><span class="lineno"> 122</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, src_strides);</div>
<div class="line"><a id="l00123" name="l00123"></a><span class="lineno"> 123</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, dst_strides);</div>
<div class="line"><a id="l00124" name="l00124"></a><span class="lineno"> 124</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00125" name="l00125"></a><span class="lineno"> 125</span>}</div>
</div>
<div class="line"><a id="l00131" name="l00131"></a><span class="lineno"> 131</span> </div>
<div class="line"><a id="l00132" name="l00132"></a><span class="lineno"> 132</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00133" data-start="{" data-end="}">
<div class="line"><a id="l00133" name="l00133"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a9fd1d98f0dcf976089754cd47f55784d"> 133</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a9fd1d98f0dcf976089754cd47f55784d">copy_gg</a>(</div>
<div class="line"><a id="l00134" name="l00134"></a><span class="lineno"> 134</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00135" name="l00135"></a><span class="lineno"> 135</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00136" name="l00136"></a><span class="lineno"> 136</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00137" name="l00137"></a><span class="lineno"> 137</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00138" name="l00138"></a><span class="lineno"> 138</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00139" name="l00139"></a><span class="lineno"> 139</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>&amp; ndim [[buffer(5)]],</div>
<div class="line"><a id="l00140" name="l00140"></a><span class="lineno"> 140</span> uint3 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00141" name="l00141"></a><span class="lineno"> 141</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2e49fa7ab8f6348543455c6c45d7e2a9">elem_to_loc</a>(index, src_shape, src_strides, ndim);</div>
<div class="line"><a id="l00142" name="l00142"></a><span class="lineno"> 142</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2e49fa7ab8f6348543455c6c45d7e2a9">elem_to_loc</a>(index, src_shape, dst_strides, ndim);</div>
<div class="line"><a id="l00143" name="l00143"></a><span class="lineno"> 143</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00144" name="l00144"></a><span class="lineno"> 144</span>}</div>
<div class="line"><a id="l00126" name="l00126"></a><span class="lineno"> 126</span> </div>
<div class="line"><a id="l00127" name="l00127"></a><span class="lineno"> 127</span><span class="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00128" data-start="{" data-end="}">
<div class="line"><a id="l00128" name="l00128"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd"> 128</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd">copy_gg_nd3</a>(</div>
<div class="line"><a id="l00129" name="l00129"></a><span class="lineno"> 129</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00130" name="l00130"></a><span class="lineno"> 130</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00131" name="l00131"></a><span class="lineno"> 131</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00132" name="l00132"></a><span class="lineno"> 132</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00133" name="l00133"></a><span class="lineno"> 133</span> uint3 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00134" name="l00134"></a><span class="lineno"> 134</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, src_strides);</div>
<div class="line"><a id="l00135" name="l00135"></a><span class="lineno"> 135</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, dst_strides);</div>
<div class="line"><a id="l00136" name="l00136"></a><span class="lineno"> 136</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00137" name="l00137"></a><span class="lineno"> 137</span>}</div>
</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="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keywordtype">int</span> DIM&gt;</div>
<div class="foldopen" id="foldopen00140" data-start="{" data-end="}">
<div class="line"><a id="l00140" name="l00140"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a3197d34e9c68e7059419ae0635667059"> 140</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a3197d34e9c68e7059419ae0635667059">copy_gg_nd</a>(</div>
<div class="line"><a id="l00141" name="l00141"></a><span class="lineno"> 141</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00142" name="l00142"></a><span class="lineno"> 142</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00143" name="l00143"></a><span class="lineno"> 143</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00144" name="l00144"></a><span class="lineno"> 144</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00145" name="l00145"></a><span class="lineno"> 145</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00146" name="l00146"></a><span class="lineno"> 146</span> uint3 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00147" name="l00147"></a><span class="lineno"> 147</span> <span class="keyword">auto</span> src_idx = elem_to_loc_nd&lt;DIM&gt;(index, src_shape, src_strides);</div>
<div class="line"><a id="l00148" name="l00148"></a><span class="lineno"> 148</span> <span class="keyword">auto</span> dst_idx = elem_to_loc_nd&lt;DIM&gt;(index, src_shape, dst_strides);</div>
<div class="line"><a id="l00149" name="l00149"></a><span class="lineno"> 149</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00150" name="l00150"></a><span class="lineno"> 150</span>}</div>
</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="keyword">template</span> &lt;<span class="keyword">typename</span> T, <span class="keyword">typename</span> U&gt;</div>
<div class="foldopen" id="foldopen00153" data-start="{" data-end="}">
<div class="line"><a id="l00153" name="l00153"></a><span class="lineno"><a class="line" href="metal_2kernels_2copy_8h.html#a9fd1d98f0dcf976089754cd47f55784d"> 153</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2copy_8h.html#a9fd1d98f0dcf976089754cd47f55784d">copy_gg</a>(</div>
<div class="line"><a id="l00154" name="l00154"></a><span class="lineno"> 154</span> device <span class="keyword">const</span> T* src [[buffer(0)]],</div>
<div class="line"><a id="l00155" name="l00155"></a><span class="lineno"> 155</span> device U* dst [[buffer(1)]],</div>
<div class="line"><a id="l00156" name="l00156"></a><span class="lineno"> 156</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* src_shape [[buffer(2)]],</div>
<div class="line"><a id="l00157" name="l00157"></a><span class="lineno"> 157</span> constant <span class="keyword">const</span> int64_t* src_strides [[buffer(3)]],</div>
<div class="line"><a id="l00158" name="l00158"></a><span class="lineno"> 158</span> constant <span class="keyword">const</span> int64_t* dst_strides [[buffer(4)]],</div>
<div class="line"><a id="l00159" name="l00159"></a><span class="lineno"> 159</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>&amp; ndim [[buffer(5)]],</div>
<div class="line"><a id="l00160" name="l00160"></a><span class="lineno"> 160</span> uint3 index [[thread_position_in_grid]]) {</div>
<div class="line"><a id="l00161" name="l00161"></a><span class="lineno"> 161</span> <span class="keyword">auto</span> src_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2e49fa7ab8f6348543455c6c45d7e2a9">elem_to_loc</a>(index, src_shape, src_strides, ndim);</div>
<div class="line"><a id="l00162" name="l00162"></a><span class="lineno"> 162</span> <span class="keyword">auto</span> dst_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2e49fa7ab8f6348543455c6c45d7e2a9">elem_to_loc</a>(index, src_shape, dst_strides, ndim);</div>
<div class="line"><a id="l00163" name="l00163"></a><span class="lineno"> 163</span> dst[dst_idx] = <span class="keyword">static_cast&lt;</span>U<span class="keyword">&gt;</span>(src[src_idx]);</div>
<div class="line"><a id="l00164" name="l00164"></a><span class="lineno"> 164</span>}</div>
</div>
<div class="ttc" id="abackend_2metal_2kernels_2utils_8h_html_a196a07022b812b241d4c06192c0fa83d"><div class="ttname"><a href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a></div><div class="ttdeci">METAL_FUNC stride_t elem_to_loc_1(uint elem, constant const stride_t &amp;stride)</div><div class="ttdef"><b>Definition</b> utils.h:123</div></div>
<div class="ttc" id="abackend_2metal_2kernels_2utils_8h_html_a2c34ed54714c69e6e1b44344f9e6e330"><div class="ttname"><a href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a></div><div class="ttdeci">METAL_FUNC stride_t elem_to_loc_3(uint3 elem, constant const stride_t strides[3])</div><div class="ttdef"><b>Definition</b> utils.h:135</div></div>
<div class="ttc" id="abackend_2metal_2kernels_2utils_8h_html_a2e49fa7ab8f6348543455c6c45d7e2a9"><div class="ttname"><a href="backend_2metal_2kernels_2utils_8h.html#a2e49fa7ab8f6348543455c6c45d7e2a9">elem_to_loc</a></div><div class="ttdeci">METAL_FUNC stride_t elem_to_loc(uint elem, device const int *shape, device const stride_t *strides, int ndim)</div><div class="ttdef"><b>Definition</b> utils.h:77</div></div>
<div class="ttc" id="abackend_2metal_2kernels_2utils_8h_html_ad6c45cacca97899cd362df49c06fea79"><div class="ttname"><a href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a></div><div class="ttdeci">METAL_FUNC stride_t elem_to_loc_2(uint2 elem, constant const stride_t strides[2])</div><div class="ttdef"><b>Definition</b> utils.h:129</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a07336a7a024125b4f6d3e8556d04058f"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a07336a7a024125b4f6d3e8556d04058f">copy_g_nd</a></div><div class="ttdeci">void copy_g_nd(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:55</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a1ea80f8edb842dd0ed8ffe553a2238b2"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a1ea80f8edb842dd0ed8ffe553a2238b2">copy_g</a></div><div class="ttdeci">void copy_g(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, constant const int &amp;ndim, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:69</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a3197d34e9c68e7059419ae0635667059"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a3197d34e9c68e7059419ae0635667059">copy_gg_nd</a></div><div class="ttdeci">void copy_gg_nd(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, constant const int64_t *dst_strides, uint3 index)</div><div class="ttdef"><b>Definition</b> copy.h:120</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a3278d9c999718bee3ccbe2922f501bf1"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1">copy_gg_nd1</a></div><div class="ttdeci">void copy_gg_nd1(device const T *src, device U *dst, constant const int64_t &amp;src_stride, constant const int64_t &amp;dst_stride, uint index)</div><div class="ttdef"><b>Definition</b> copy.h:84</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a3e2d3cc7f34f56170409b6735f51a950"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950">copy_gg_nd2</a></div><div class="ttdeci">void copy_gg_nd2(device const T *src, device U *dst, constant const int64_t *src_strides, constant const int64_t *dst_strides, uint2 index)</div><div class="ttdef"><b>Definition</b> copy.h:96</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a59f43b5bffed936d7559ceb06a10aabd"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd">copy_gg_nd3</a></div><div class="ttdeci">void copy_gg_nd3(device const T *src, device U *dst, constant const int64_t *src_strides, constant const int64_t *dst_strides, uint3 index)</div><div class="ttdef"><b>Definition</b> copy.h:108</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a821f8f3f3891159a295c66fc25aed1ff"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff">copy_g_nd3</a></div><div class="ttdeci">void copy_g_nd3(device const T *src, device U *dst, constant const int64_t *src_strides, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:42</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a9fd1d98f0dcf976089754cd47f55784d"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a9fd1d98f0dcf976089754cd47f55784d">copy_gg</a></div><div class="ttdeci">void copy_gg(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, constant const int64_t *dst_strides, constant const int &amp;ndim, uint3 index)</div><div class="ttdef"><b>Definition</b> copy.h:133</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_aba4530a7db6a61ca36f50e4f5e58fb77"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77">copy_g_nd1</a></div><div class="ttdeci">void copy_g_nd1(device const T *src, device U *dst, constant const int64_t &amp;src_stride, uint index)</div><div class="ttdef"><b>Definition</b> copy.h:20</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a07336a7a024125b4f6d3e8556d04058f"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a07336a7a024125b4f6d3e8556d04058f">copy_g_nd</a></div><div class="ttdeci">void copy_g_nd(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:75</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a1ea80f8edb842dd0ed8ffe553a2238b2"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a1ea80f8edb842dd0ed8ffe553a2238b2">copy_g</a></div><div class="ttdeci">void copy_g(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, constant const int &amp;ndim, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:89</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a3197d34e9c68e7059419ae0635667059"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a3197d34e9c68e7059419ae0635667059">copy_gg_nd</a></div><div class="ttdeci">void copy_gg_nd(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, constant const int64_t *dst_strides, uint3 index)</div><div class="ttdef"><b>Definition</b> copy.h:140</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a3278d9c999718bee3ccbe2922f501bf1"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a3278d9c999718bee3ccbe2922f501bf1">copy_gg_nd1</a></div><div class="ttdeci">void copy_gg_nd1(device const T *src, device U *dst, constant const int64_t &amp;src_stride, constant const int64_t &amp;dst_stride, uint index)</div><div class="ttdef"><b>Definition</b> copy.h:104</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a3e2d3cc7f34f56170409b6735f51a950"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a3e2d3cc7f34f56170409b6735f51a950">copy_gg_nd2</a></div><div class="ttdeci">void copy_gg_nd2(device const T *src, device U *dst, constant const int64_t *src_strides, constant const int64_t *dst_strides, uint2 index)</div><div class="ttdef"><b>Definition</b> copy.h:116</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a59f43b5bffed936d7559ceb06a10aabd"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a59f43b5bffed936d7559ceb06a10aabd">copy_gg_nd3</a></div><div class="ttdeci">void copy_gg_nd3(device const T *src, device U *dst, constant const int64_t *src_strides, constant const int64_t *dst_strides, uint3 index)</div><div class="ttdef"><b>Definition</b> copy.h:128</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a8023e9335cc5334847a8d315042be3a3"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a8023e9335cc5334847a8d315042be3a3">copy_s2</a></div><div class="ttdeci">void copy_s2(device const T *src, device U *dst, uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:20</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a821f8f3f3891159a295c66fc25aed1ff"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a821f8f3f3891159a295c66fc25aed1ff">copy_g_nd3</a></div><div class="ttdeci">void copy_g_nd3(device const T *src, device U *dst, constant const int64_t *src_strides, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:62</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_a9fd1d98f0dcf976089754cd47f55784d"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#a9fd1d98f0dcf976089754cd47f55784d">copy_gg</a></div><div class="ttdeci">void copy_gg(device const T *src, device U *dst, constant const int *src_shape, constant const int64_t *src_strides, constant const int64_t *dst_strides, constant const int &amp;ndim, uint3 index)</div><div class="ttdef"><b>Definition</b> copy.h:153</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_aba4530a7db6a61ca36f50e4f5e58fb77"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#aba4530a7db6a61ca36f50e4f5e58fb77">copy_g_nd1</a></div><div class="ttdeci">void copy_g_nd1(device const T *src, device U *dst, constant const int64_t &amp;src_stride, uint index)</div><div class="ttdef"><b>Definition</b> copy.h:40</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_ae26a13e0c8e6c15f7b10078e65970659"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#ae26a13e0c8e6c15f7b10078e65970659">copy_v</a></div><div class="ttdeci">void copy_v(device const T *src, device U *dst, uint index)</div><div class="ttdef"><b>Definition</b> copy.h:12</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_aee678c7c31119f3e609685589f37490c"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c">copy_g_nd2</a></div><div class="ttdeci">void copy_g_nd2(device const T *src, device U *dst, constant const int64_t *src_strides, uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:30</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_aee14a5326f53d9b30b0b38e27d180ef3"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#aee14a5326f53d9b30b0b38e27d180ef3">copy_v2</a></div><div class="ttdeci">void copy_v2(device const T *src, device U *dst, uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:30</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_aee678c7c31119f3e609685589f37490c"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#aee678c7c31119f3e609685589f37490c">copy_g_nd2</a></div><div class="ttdeci">void copy_g_nd2(device const T *src, device U *dst, constant const int64_t *src_strides, uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> copy.h:50</div></div>
<div class="ttc" id="ametal_2kernels_2copy_8h_html_aef09f9b9475345b1bba121d037d222ea"><div class="ttname"><a href="metal_2kernels_2copy_8h.html#aef09f9b9475345b1bba121d037d222ea">copy_s</a></div><div class="ttdeci">void copy_s(device const T *src, device U *dst, uint index)</div><div class="ttdef"><b>Definition</b> copy.h:4</div></div>
</div><!-- fragment --></div><!-- contents -->
<!-- start footer part -->