mirror of
https://github.com/ml-explore/mlx.git
synced 2025-09-19 02:38:09 +08:00
docs update
This commit is contained in:

committed by
CircleCI Docs

parent
8f68182d95
commit
d8f7b8cda6
@@ -131,101 +131,143 @@ $(function() { codefold.init(0); });
|
||||
<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> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></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_2binary_8h.html#a6808bfb006cb5473da087a2758d0d867"> 40</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a6808bfb006cb5473da087a2758d0d867">binary_g_nd1</a>(</div>
|
||||
<div class="line"><a id="l00040" name="l00040"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#aa8c48b1b21d8f5a181f5443de2346589"> 40</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#aa8c48b1b21d8f5a181f5443de2346589">binary_sv2</a>(</div>
|
||||
<div class="line"><a id="l00041" name="l00041"></a><span class="lineno"> 41</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00042" name="l00042"></a><span class="lineno"> 42</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00043" name="l00043"></a><span class="lineno"> 43</span> device U* c,</div>
|
||||
<div class="line"><a id="l00044" name="l00044"></a><span class="lineno"> 44</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>& a_stride,</div>
|
||||
<div class="line"><a id="l00045" name="l00045"></a><span class="lineno"> 45</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>& b_stride,</div>
|
||||
<div class="line"><a id="l00046" name="l00046"></a><span class="lineno"> 46</span> uint index [[thread_position_in_grid]]) {</div>
|
||||
<div class="line"><a id="l00047" name="l00047"></a><span class="lineno"> 47</span> <span class="keyword">auto</span> a_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, a_stride);</div>
|
||||
<div class="line"><a id="l00048" name="l00048"></a><span class="lineno"> 48</span> <span class="keyword">auto</span> b_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, b_stride);</div>
|
||||
<div class="line"><a id="l00049" name="l00049"></a><span class="lineno"> 49</span> c[index] = Op()(a[a_idx], b[b_idx]);</div>
|
||||
<div class="line"><a id="l00050" name="l00050"></a><span class="lineno"> 50</span>}</div>
|
||||
<div class="line"><a id="l00044" name="l00044"></a><span class="lineno"> 44</span> uint2 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00045" name="l00045"></a><span class="lineno"> 45</span> uint2 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00046" name="l00046"></a><span class="lineno"> 46</span> <span class="keywordtype">size_t</span> offset = index.x + grid_dim.x * size_t(index.y);</div>
|
||||
<div class="line"><a id="l00047" name="l00047"></a><span class="lineno"> 47</span> c[offset] = Op()(a[0], b[offset]);</div>
|
||||
<div class="line"><a id="l00048" name="l00048"></a><span class="lineno"> 48</span>}</div>
|
||||
</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"> 52</span><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></div>
|
||||
<div class="foldopen" id="foldopen00053" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00053" name="l00053"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#a8cd5989852ec704c6fd132ae28f4fc14"> 53</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a8cd5989852ec704c6fd132ae28f4fc14">binary_g_nd2</a>(</div>
|
||||
<div class="line"><a id="l00054" name="l00054"></a><span class="lineno"> 54</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00055" name="l00055"></a><span class="lineno"> 55</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00056" name="l00056"></a><span class="lineno"> 56</span> device U* c,</div>
|
||||
<div class="line"><a id="l00057" name="l00057"></a><span class="lineno"> 57</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> a_strides[2],</div>
|
||||
<div class="line"><a id="l00058" name="l00058"></a><span class="lineno"> 58</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> b_strides[2],</div>
|
||||
<div class="line"><a id="l00059" name="l00059"></a><span class="lineno"> 59</span> uint2 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00060" name="l00060"></a><span class="lineno"> 60</span> uint2 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00061" name="l00061"></a><span class="lineno"> 61</span> <span class="keyword">auto</span> a_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, a_strides);</div>
|
||||
<div class="line"><a id="l00062" name="l00062"></a><span class="lineno"> 62</span> <span class="keyword">auto</span> b_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, b_strides);</div>
|
||||
<div class="line"><a id="l00063" name="l00063"></a><span class="lineno"> 63</span> <span class="keywordtype">size_t</span> out_idx = index.x + (size_t)grid_dim.x * index.y;</div>
|
||||
<div class="line"><a id="l00064" name="l00064"></a><span class="lineno"> 64</span> c[out_idx] = Op()(a[a_idx], b[b_idx]);</div>
|
||||
<div class="line"><a id="l00065" name="l00065"></a><span class="lineno"> 65</span>}</div>
|
||||
<div class="line"><a id="l00049" name="l00049"></a><span class="lineno"> 49</span> </div>
|
||||
<div class="line"><a id="l00050" name="l00050"></a><span class="lineno"> 50</span><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></div>
|
||||
<div class="foldopen" id="foldopen00051" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00051" name="l00051"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#a48bd82eb10f9c623ce7d28daec4fa512"> 51</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a48bd82eb10f9c623ce7d28daec4fa512">binary_vs2</a>(</div>
|
||||
<div class="line"><a id="l00052" name="l00052"></a><span class="lineno"> 52</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00053" name="l00053"></a><span class="lineno"> 53</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00054" name="l00054"></a><span class="lineno"> 54</span> device U* c,</div>
|
||||
<div class="line"><a id="l00055" name="l00055"></a><span class="lineno"> 55</span> uint2 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00056" name="l00056"></a><span class="lineno"> 56</span> uint2 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00057" name="l00057"></a><span class="lineno"> 57</span> <span class="keywordtype">size_t</span> offset = index.x + grid_dim.x * size_t(index.y);</div>
|
||||
<div class="line"><a id="l00058" name="l00058"></a><span class="lineno"> 58</span> c[offset] = Op()(a[offset], b[0]);</div>
|
||||
<div class="line"><a id="l00059" name="l00059"></a><span class="lineno"> 59</span>}</div>
|
||||
</div>
|
||||
<div class="line"><a id="l00066" name="l00066"></a><span class="lineno"> 66</span> </div>
|
||||
<div class="line"><a id="l00067" name="l00067"></a><span class="lineno"> 67</span><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></div>
|
||||
<div class="foldopen" id="foldopen00068" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00068" name="l00068"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#ac4979e60b993f7ffb602bcb91cd68bc9"> 68</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#ac4979e60b993f7ffb602bcb91cd68bc9">binary_g_nd3</a>(</div>
|
||||
<div class="line"><a id="l00069" name="l00069"></a><span class="lineno"> 69</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00070" name="l00070"></a><span class="lineno"> 70</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00071" name="l00071"></a><span class="lineno"> 71</span> device U* c,</div>
|
||||
<div class="line"><a id="l00072" name="l00072"></a><span class="lineno"> 72</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> a_strides[3],</div>
|
||||
<div class="line"><a id="l00073" name="l00073"></a><span class="lineno"> 73</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> b_strides[3],</div>
|
||||
<div class="line"><a id="l00074" name="l00074"></a><span class="lineno"> 74</span> uint3 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00075" name="l00075"></a><span class="lineno"> 75</span> uint3 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00076" name="l00076"></a><span class="lineno"> 76</span> <span class="keyword">auto</span> a_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, a_strides);</div>
|
||||
<div class="line"><a id="l00077" name="l00077"></a><span class="lineno"> 77</span> <span class="keyword">auto</span> b_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, b_strides);</div>
|
||||
<div class="line"><a id="l00078" name="l00078"></a><span class="lineno"> 78</span> <span class="keywordtype">size_t</span> out_idx =</div>
|
||||
<div class="line"><a id="l00079" name="l00079"></a><span class="lineno"> 79</span> index.x + (size_t)grid_dim.x * (index.y + (<span class="keywordtype">size_t</span>)grid_dim.y * index.z);</div>
|
||||
<div class="line"><a id="l00080" name="l00080"></a><span class="lineno"> 80</span> c[out_idx] = Op()(a[a_idx], b[b_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> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></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_2binary_8h.html#a19dbbf8fea68b64bdd25dc8d36865171"> 62</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a19dbbf8fea68b64bdd25dc8d36865171">binary_vv2</a>(</div>
|
||||
<div class="line"><a id="l00063" name="l00063"></a><span class="lineno"> 63</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00064" name="l00064"></a><span class="lineno"> 64</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00065" name="l00065"></a><span class="lineno"> 65</span> device U* c,</div>
|
||||
<div class="line"><a id="l00066" name="l00066"></a><span class="lineno"> 66</span> uint2 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00067" name="l00067"></a><span class="lineno"> 67</span> uint2 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00068" name="l00068"></a><span class="lineno"> 68</span> <span class="keywordtype">size_t</span> offset = index.x + grid_dim.x * size_t(index.y);</div>
|
||||
<div class="line"><a id="l00069" name="l00069"></a><span class="lineno"> 69</span> c[offset] = Op()(a[offset], b[offset]);</div>
|
||||
<div class="line"><a id="l00070" name="l00070"></a><span class="lineno"> 70</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> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op, <span class="keywordtype">int</span> DIM></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_2binary_8h.html#a4f273fb6aad88b74bfc63f0f29499c03"> 84</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a4f273fb6aad88b74bfc63f0f29499c03">binary_g_nd</a>(</div>
|
||||
<div class="line"><a id="l00085" name="l00085"></a><span class="lineno"> 85</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00086" name="l00086"></a><span class="lineno"> 86</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00087" name="l00087"></a><span class="lineno"> 87</span> device U* c,</div>
|
||||
<div class="line"><a id="l00088" name="l00088"></a><span class="lineno"> 88</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span> shape[DIM],</div>
|
||||
<div class="line"><a id="l00089" name="l00089"></a><span class="lineno"> 89</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> a_strides[DIM],</div>
|
||||
<div class="line"><a id="l00090" name="l00090"></a><span class="lineno"> 90</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> b_strides[DIM],</div>
|
||||
<div class="line"><a id="l00091" name="l00091"></a><span class="lineno"> 91</span> uint3 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00092" name="l00092"></a><span class="lineno"> 92</span> uint3 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00093" name="l00093"></a><span class="lineno"> 93</span> <span class="keyword">auto</span> idx = elem_to_loc_2_nd<DIM>(index, shape, a_strides, b_strides);</div>
|
||||
<div class="line"><a id="l00094" name="l00094"></a><span class="lineno"> 94</span> <span class="keywordtype">size_t</span> out_idx =</div>
|
||||
<div class="line"><a id="l00095" name="l00095"></a><span class="lineno"> 95</span> index.x + (size_t)grid_dim.x * (index.y + (<span class="keywordtype">size_t</span>)grid_dim.y * index.z);</div>
|
||||
<div class="line"><a id="l00096" name="l00096"></a><span class="lineno"> 96</span> c[out_idx] = Op()(a[idx.x], b[idx.y]);</div>
|
||||
<div class="line"><a id="l00097" name="l00097"></a><span class="lineno"> 97</span>}</div>
|
||||
<div class="line"><a id="l00071" name="l00071"></a><span class="lineno"> 71</span> </div>
|
||||
<div class="line"><a id="l00072" name="l00072"></a><span class="lineno"> 72</span><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></div>
|
||||
<div class="foldopen" id="foldopen00073" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00073" name="l00073"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#a6808bfb006cb5473da087a2758d0d867"> 73</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a6808bfb006cb5473da087a2758d0d867">binary_g_nd1</a>(</div>
|
||||
<div class="line"><a id="l00074" name="l00074"></a><span class="lineno"> 74</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00075" name="l00075"></a><span class="lineno"> 75</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00076" name="l00076"></a><span class="lineno"> 76</span> device U* c,</div>
|
||||
<div class="line"><a id="l00077" name="l00077"></a><span class="lineno"> 77</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>& a_stride,</div>
|
||||
<div class="line"><a id="l00078" name="l00078"></a><span class="lineno"> 78</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>& b_stride,</div>
|
||||
<div class="line"><a id="l00079" name="l00079"></a><span class="lineno"> 79</span> uint index [[thread_position_in_grid]]) {</div>
|
||||
<div class="line"><a id="l00080" name="l00080"></a><span class="lineno"> 80</span> <span class="keyword">auto</span> a_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, a_stride);</div>
|
||||
<div class="line"><a id="l00081" name="l00081"></a><span class="lineno"> 81</span> <span class="keyword">auto</span> b_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a196a07022b812b241d4c06192c0fa83d">elem_to_loc_1</a>(index, b_stride);</div>
|
||||
<div class="line"><a id="l00082" name="l00082"></a><span class="lineno"> 82</span> c[index] = Op()(a[a_idx], b[b_idx]);</div>
|
||||
<div class="line"><a id="l00083" name="l00083"></a><span class="lineno"> 83</span>}</div>
|
||||
</div>
|
||||
<div class="line"><a id="l00098" name="l00098"></a><span class="lineno"> 98</span> </div>
|
||||
<div class="line"><a id="l00099" name="l00099"></a><span class="lineno"> 99</span><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></div>
|
||||
<div class="foldopen" id="foldopen00100" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00100" name="l00100"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#a701e3cbc842992aa74f01b91a733774e"> 100</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a701e3cbc842992aa74f01b91a733774e">binary_g</a>(</div>
|
||||
<div class="line"><a id="l00101" name="l00101"></a><span class="lineno"> 101</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00102" name="l00102"></a><span class="lineno"> 102</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00103" name="l00103"></a><span class="lineno"> 103</span> device U* c,</div>
|
||||
<div class="line"><a id="l00104" name="l00104"></a><span class="lineno"> 104</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* shape,</div>
|
||||
<div class="line"><a id="l00105" name="l00105"></a><span class="lineno"> 105</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>* a_strides,</div>
|
||||
<div class="line"><a id="l00106" name="l00106"></a><span class="lineno"> 106</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>* b_strides,</div>
|
||||
<div class="line"><a id="l00107" name="l00107"></a><span class="lineno"> 107</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>& ndim,</div>
|
||||
<div class="line"><a id="l00108" name="l00108"></a><span class="lineno"> 108</span> uint3 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00109" name="l00109"></a><span class="lineno"> 109</span> uint3 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00110" name="l00110"></a><span class="lineno"> 110</span> <span class="keyword">auto</span> idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a4069a6398757e8158c14551539083181">elem_to_loc_2_nd</a>(index, shape, a_strides, b_strides, ndim);</div>
|
||||
<div class="line"><a id="l00111" name="l00111"></a><span class="lineno"> 111</span> <span class="keywordtype">size_t</span> out_idx = index.x + grid_dim.x * (index.y + grid_dim.y * index.z);</div>
|
||||
<div class="line"><a id="l00112" name="l00112"></a><span class="lineno"> 112</span> c[out_idx] = Op()(a[idx.x], b[idx.y]);</div>
|
||||
<div class="line"><a id="l00113" name="l00113"></a><span class="lineno"> 113</span>}</div>
|
||||
<div class="line"><a id="l00084" name="l00084"></a><span class="lineno"> 84</span> </div>
|
||||
<div class="line"><a id="l00085" name="l00085"></a><span class="lineno"> 85</span><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></div>
|
||||
<div class="foldopen" id="foldopen00086" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00086" name="l00086"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#a8cd5989852ec704c6fd132ae28f4fc14"> 86</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a8cd5989852ec704c6fd132ae28f4fc14">binary_g_nd2</a>(</div>
|
||||
<div class="line"><a id="l00087" name="l00087"></a><span class="lineno"> 87</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00088" name="l00088"></a><span class="lineno"> 88</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00089" name="l00089"></a><span class="lineno"> 89</span> device U* c,</div>
|
||||
<div class="line"><a id="l00090" name="l00090"></a><span class="lineno"> 90</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> a_strides[2],</div>
|
||||
<div class="line"><a id="l00091" name="l00091"></a><span class="lineno"> 91</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> b_strides[2],</div>
|
||||
<div class="line"><a id="l00092" name="l00092"></a><span class="lineno"> 92</span> uint2 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00093" name="l00093"></a><span class="lineno"> 93</span> uint2 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00094" name="l00094"></a><span class="lineno"> 94</span> <span class="keyword">auto</span> a_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, a_strides);</div>
|
||||
<div class="line"><a id="l00095" name="l00095"></a><span class="lineno"> 95</span> <span class="keyword">auto</span> b_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#ad6c45cacca97899cd362df49c06fea79">elem_to_loc_2</a>(index, b_strides);</div>
|
||||
<div class="line"><a id="l00096" name="l00096"></a><span class="lineno"> 96</span> <span class="keywordtype">size_t</span> out_idx = index.x + (size_t)grid_dim.x * index.y;</div>
|
||||
<div class="line"><a id="l00097" name="l00097"></a><span class="lineno"> 97</span> c[out_idx] = Op()(a[a_idx], b[b_idx]);</div>
|
||||
<div class="line"><a id="l00098" name="l00098"></a><span class="lineno"> 98</span>}</div>
|
||||
</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><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></div>
|
||||
<div class="foldopen" id="foldopen00101" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00101" name="l00101"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#ac4979e60b993f7ffb602bcb91cd68bc9"> 101</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#ac4979e60b993f7ffb602bcb91cd68bc9">binary_g_nd3</a>(</div>
|
||||
<div class="line"><a id="l00102" name="l00102"></a><span class="lineno"> 102</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00103" name="l00103"></a><span class="lineno"> 103</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00104" name="l00104"></a><span class="lineno"> 104</span> device U* c,</div>
|
||||
<div class="line"><a id="l00105" name="l00105"></a><span class="lineno"> 105</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> a_strides[3],</div>
|
||||
<div class="line"><a id="l00106" name="l00106"></a><span class="lineno"> 106</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> b_strides[3],</div>
|
||||
<div class="line"><a id="l00107" name="l00107"></a><span class="lineno"> 107</span> uint3 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00108" name="l00108"></a><span class="lineno"> 108</span> uint3 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00109" name="l00109"></a><span class="lineno"> 109</span> <span class="keyword">auto</span> a_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, a_strides);</div>
|
||||
<div class="line"><a id="l00110" name="l00110"></a><span class="lineno"> 110</span> <span class="keyword">auto</span> b_idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a2c34ed54714c69e6e1b44344f9e6e330">elem_to_loc_3</a>(index, b_strides);</div>
|
||||
<div class="line"><a id="l00111" name="l00111"></a><span class="lineno"> 111</span> <span class="keywordtype">size_t</span> out_idx =</div>
|
||||
<div class="line"><a id="l00112" name="l00112"></a><span class="lineno"> 112</span> index.x + (size_t)grid_dim.x * (index.y + (<span class="keywordtype">size_t</span>)grid_dim.y * index.z);</div>
|
||||
<div class="line"><a id="l00113" name="l00113"></a><span class="lineno"> 113</span> c[out_idx] = Op()(a[a_idx], b[b_idx]);</div>
|
||||
<div class="line"><a id="l00114" name="l00114"></a><span class="lineno"> 114</span>}</div>
|
||||
</div>
|
||||
<div class="line"><a id="l00115" name="l00115"></a><span class="lineno"> 115</span> </div>
|
||||
<div class="line"><a id="l00116" name="l00116"></a><span class="lineno"> 116</span><span class="keyword">template</span> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op, <span class="keywordtype">int</span> DIM></div>
|
||||
<div class="foldopen" id="foldopen00117" data-start="{" data-end="}">
|
||||
<div class="line"><a id="l00117" name="l00117"></a><span class="lineno"><a class="line" href="metal_2kernels_2binary_8h.html#a4f273fb6aad88b74bfc63f0f29499c03"> 117</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a4f273fb6aad88b74bfc63f0f29499c03">binary_g_nd</a>(</div>
|
||||
<div class="line"><a id="l00118" name="l00118"></a><span class="lineno"> 118</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00119" name="l00119"></a><span class="lineno"> 119</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00120" name="l00120"></a><span class="lineno"> 120</span> device U* c,</div>
|
||||
<div class="line"><a id="l00121" name="l00121"></a><span class="lineno"> 121</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span> shape[DIM],</div>
|
||||
<div class="line"><a id="l00122" name="l00122"></a><span class="lineno"> 122</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> a_strides[DIM],</div>
|
||||
<div class="line"><a id="l00123" name="l00123"></a><span class="lineno"> 123</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span> b_strides[DIM],</div>
|
||||
<div class="line"><a id="l00124" name="l00124"></a><span class="lineno"> 124</span> uint3 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00125" name="l00125"></a><span class="lineno"> 125</span> uint3 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00126" name="l00126"></a><span class="lineno"> 126</span> <span class="keyword">auto</span> idx = elem_to_loc_2_nd<DIM>(index, shape, a_strides, b_strides);</div>
|
||||
<div class="line"><a id="l00127" name="l00127"></a><span class="lineno"> 127</span> <span class="keywordtype">size_t</span> out_idx =</div>
|
||||
<div class="line"><a id="l00128" name="l00128"></a><span class="lineno"> 128</span> index.x + (size_t)grid_dim.x * (index.y + (<span class="keywordtype">size_t</span>)grid_dim.y * index.z);</div>
|
||||
<div class="line"><a id="l00129" name="l00129"></a><span class="lineno"> 129</span> c[out_idx] = Op()(a[idx.x], b[idx.y]);</div>
|
||||
<div class="line"><a id="l00130" name="l00130"></a><span class="lineno"> 130</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> <<span class="keyword">typename</span> T, <span class="keyword">typename</span> U, <span class="keyword">typename</span> Op></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_2binary_8h.html#a701e3cbc842992aa74f01b91a733774e"> 133</a></span>[[kernel]] <span class="keywordtype">void</span> <a class="code hl_function" href="metal_2kernels_2binary_8h.html#a701e3cbc842992aa74f01b91a733774e">binary_g</a>(</div>
|
||||
<div class="line"><a id="l00134" name="l00134"></a><span class="lineno"> 134</span> device <span class="keyword">const</span> T* a,</div>
|
||||
<div class="line"><a id="l00135" name="l00135"></a><span class="lineno"> 135</span> device <span class="keyword">const</span> T* b,</div>
|
||||
<div class="line"><a id="l00136" name="l00136"></a><span class="lineno"> 136</span> device U* c,</div>
|
||||
<div class="line"><a id="l00137" name="l00137"></a><span class="lineno"> 137</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>* shape,</div>
|
||||
<div class="line"><a id="l00138" name="l00138"></a><span class="lineno"> 138</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>* a_strides,</div>
|
||||
<div class="line"><a id="l00139" name="l00139"></a><span class="lineno"> 139</span> constant <span class="keyword">const</span> <span class="keywordtype">size_t</span>* b_strides,</div>
|
||||
<div class="line"><a id="l00140" name="l00140"></a><span class="lineno"> 140</span> constant <span class="keyword">const</span> <span class="keywordtype">int</span>& ndim,</div>
|
||||
<div class="line"><a id="l00141" name="l00141"></a><span class="lineno"> 141</span> uint3 index [[thread_position_in_grid]],</div>
|
||||
<div class="line"><a id="l00142" name="l00142"></a><span class="lineno"> 142</span> uint3 grid_dim [[threads_per_grid]]) {</div>
|
||||
<div class="line"><a id="l00143" name="l00143"></a><span class="lineno"> 143</span> <span class="keyword">auto</span> idx = <a class="code hl_function" href="backend_2metal_2kernels_2utils_8h.html#a4069a6398757e8158c14551539083181">elem_to_loc_2_nd</a>(index, shape, a_strides, b_strides, ndim);</div>
|
||||
<div class="line"><a id="l00144" name="l00144"></a><span class="lineno"> 144</span> <span class="keywordtype">size_t</span> out_idx = index.x + grid_dim.x * (index.y + grid_dim.y * index.z);</div>
|
||||
<div class="line"><a id="l00145" name="l00145"></a><span class="lineno"> 145</span> c[out_idx] = Op()(a[idx.x], b[idx.y]);</div>
|
||||
<div class="line"><a id="l00146" name="l00146"></a><span class="lineno"> 146</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 &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_a4069a6398757e8158c14551539083181"><div class="ttname"><a href="backend_2metal_2kernels_2utils_8h.html#a4069a6398757e8158c14551539083181">elem_to_loc_2_nd</a></div><div class="ttdeci">METAL_FUNC uint2 elem_to_loc_2_nd(uint3 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:200</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_2binary_8h_html_a19dbbf8fea68b64bdd25dc8d36865171"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a19dbbf8fea68b64bdd25dc8d36865171">binary_vv2</a></div><div class="ttdeci">void binary_vv2(device const T *a, device const T *b, device U *c, uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:62</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a242b8b29a852c255467e50628c6dccf5"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a242b8b29a852c255467e50628c6dccf5">binary_ss</a></div><div class="ttdeci">void binary_ss(device const T *a, device const T *b, device U *c, uint index)</div><div class="ttdef"><b>Definition</b> binary.h:4</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a4116c35f2e4632366d1611d5a95ba141"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a4116c35f2e4632366d1611d5a95ba141">binary_sv</a></div><div class="ttdeci">void binary_sv(device const T *a, device const T *b, device U *c, uint index)</div><div class="ttdef"><b>Definition</b> binary.h:13</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a4f273fb6aad88b74bfc63f0f29499c03"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a4f273fb6aad88b74bfc63f0f29499c03">binary_g_nd</a></div><div class="ttdeci">void binary_g_nd(device const T *a, device const T *b, device U *c, constant const int shape[DIM], constant const size_t a_strides[DIM], constant const size_t b_strides[DIM], uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:84</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a48bd82eb10f9c623ce7d28daec4fa512"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a48bd82eb10f9c623ce7d28daec4fa512">binary_vs2</a></div><div class="ttdeci">void binary_vs2(device const T *a, device const T *b, device U *c, uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:51</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a4f273fb6aad88b74bfc63f0f29499c03"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a4f273fb6aad88b74bfc63f0f29499c03">binary_g_nd</a></div><div class="ttdeci">void binary_g_nd(device const T *a, device const T *b, device U *c, constant const int shape[DIM], constant const size_t a_strides[DIM], constant const size_t b_strides[DIM], uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:117</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a649851d133358dd5832a73b1061b3313"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a649851d133358dd5832a73b1061b3313">binary_vs</a></div><div class="ttdeci">void binary_vs(device const T *a, device const T *b, device U *c, uint index)</div><div class="ttdef"><b>Definition</b> binary.h:22</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a6808bfb006cb5473da087a2758d0d867"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a6808bfb006cb5473da087a2758d0d867">binary_g_nd1</a></div><div class="ttdeci">void binary_g_nd1(device const T *a, device const T *b, device U *c, constant const size_t &a_stride, constant const size_t &b_stride, uint index)</div><div class="ttdef"><b>Definition</b> binary.h:40</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a701e3cbc842992aa74f01b91a733774e"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a701e3cbc842992aa74f01b91a733774e">binary_g</a></div><div class="ttdeci">void binary_g(device const T *a, device const T *b, device U *c, constant const int *shape, constant const size_t *a_strides, constant const size_t *b_strides, constant const int &ndim, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:100</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a8cd5989852ec704c6fd132ae28f4fc14"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a8cd5989852ec704c6fd132ae28f4fc14">binary_g_nd2</a></div><div class="ttdeci">void binary_g_nd2(device const T *a, device const T *b, device U *c, constant const size_t a_strides[2], constant const size_t b_strides[2], uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:53</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_ac4979e60b993f7ffb602bcb91cd68bc9"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#ac4979e60b993f7ffb602bcb91cd68bc9">binary_g_nd3</a></div><div class="ttdeci">void binary_g_nd3(device const T *a, device const T *b, device U *c, constant const size_t a_strides[3], constant const size_t b_strides[3], uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:68</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a6808bfb006cb5473da087a2758d0d867"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a6808bfb006cb5473da087a2758d0d867">binary_g_nd1</a></div><div class="ttdeci">void binary_g_nd1(device const T *a, device const T *b, device U *c, constant const size_t &a_stride, constant const size_t &b_stride, uint index)</div><div class="ttdef"><b>Definition</b> binary.h:73</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a701e3cbc842992aa74f01b91a733774e"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a701e3cbc842992aa74f01b91a733774e">binary_g</a></div><div class="ttdeci">void binary_g(device const T *a, device const T *b, device U *c, constant const int *shape, constant const size_t *a_strides, constant const size_t *b_strides, constant const int &ndim, uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:133</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_a8cd5989852ec704c6fd132ae28f4fc14"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#a8cd5989852ec704c6fd132ae28f4fc14">binary_g_nd2</a></div><div class="ttdeci">void binary_g_nd2(device const T *a, device const T *b, device U *c, constant const size_t a_strides[2], constant const size_t b_strides[2], uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:86</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_aa8c48b1b21d8f5a181f5443de2346589"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#aa8c48b1b21d8f5a181f5443de2346589">binary_sv2</a></div><div class="ttdeci">void binary_sv2(device const T *a, device const T *b, device U *c, uint2 index, uint2 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:40</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_ac4979e60b993f7ffb602bcb91cd68bc9"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#ac4979e60b993f7ffb602bcb91cd68bc9">binary_g_nd3</a></div><div class="ttdeci">void binary_g_nd3(device const T *a, device const T *b, device U *c, constant const size_t a_strides[3], constant const size_t b_strides[3], uint3 index, uint3 grid_dim)</div><div class="ttdef"><b>Definition</b> binary.h:101</div></div>
|
||||
<div class="ttc" id="ametal_2kernels_2binary_8h_html_add6a9aeee3cb0ba909574f27fa9ecd5b"><div class="ttname"><a href="metal_2kernels_2binary_8h.html#add6a9aeee3cb0ba909574f27fa9ecd5b">binary_vv</a></div><div class="ttdeci">void binary_vv(device const T *a, device const T *b, device U *c, uint index)</div><div class="ttdef"><b>Definition</b> binary.h:31</div></div>
|
||||
</div><!-- fragment --></div><!-- contents -->
|
||||
<!-- start footer part -->
|
||||
|
Reference in New Issue
Block a user