{"id":697,"date":"2021-12-07T00:32:21","date_gmt":"2021-12-06T16:32:21","guid":{"rendered":"https:\/\/blog.cauchyschwarz.com\/?p=697"},"modified":"2021-12-07T00:36:17","modified_gmt":"2021-12-06T16:36:17","slug":"neon%e5%8a%a0%e9%80%9f","status":"publish","type":"post","link":"https:\/\/blog.cauchyschwarz.com\/?p=697","title":{"rendered":"NEON\u52a0\u901f"},"content":{"rendered":"\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_82_2 ez-toc-wrap-right counter-hierarchy ez-toc-counter ez-toc-grey ez-toc-container-direction\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">Table of Contents<\/p>\n<label for=\"ez-toc-cssicon-toggle-item-69e5f97da975b\" class=\"ez-toc-cssicon-toggle-label\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #999;color:#999\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #999;color:#999\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/label><input type=\"checkbox\"  id=\"ez-toc-cssicon-toggle-item-69e5f97da975b\" checked aria-label=\"Toggle\" \/><nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-1'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#NEON%E6%8A%80%E6%9C%AF\" >NEON\u6280\u672f<\/a><ul class='ez-toc-list-level-2' ><li class='ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#NEON%E6%8A%80%E6%9C%AF%E7%AE%80%E4%BB%8B\" >NEON\u6280\u672f\u7b80\u4ecb<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-1'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E5%9C%A8CC%E4%B8%AD%E4%BD%BF%E7%94%A8NEON%E6%8A%80%E6%9C%AF\" >\u5728C\/C++\u4e2d\u4f7f\u7528NEON\u6280\u672f<\/a><ul class='ez-toc-list-level-2' ><li class='ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E9%A6%96%E5%85%88%E7%A1%AE%E5%AE%9A%E8%83%BD%E5%90%A6%E4%BD%BF%E7%94%A8NEON%E6%8A%80%E6%9C%AF\" >\u9996\u5148\u786e\u5b9a\u80fd\u5426\u4f7f\u7528NEON\u6280\u672f<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E5%9C%A8CC%E4%B8%AD%E4%BD%BF%E7%94%A8NEON%E6%8A%80%E6%9C%AF-2\" >\u5728C\/C++\u4e2d\u4f7f\u7528NEON\u6280\u672f<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#NEON%E6%8A%80%E6%9C%AF%E5%8E%9F%E7%90%86\" >NEON\u6280\u672f\u539f\u7406<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E4%BD%BF%E7%94%A8NEON_Intrinsics\" >\u4f7f\u7528NEON Intrinsics<\/a><ul class='ez-toc-list-level-4' ><li class='ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E5%8C%85%E5%90%AB%E5%BF%85%E8%A6%81%E7%9A%84%E5%A4%B4%E6%96%87%E4%BB%B6\" >\u5305\u542b\u5fc5\u8981\u7684\u5934\u6587\u4ef6<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E5%B0%86%E6%95%B0%E6%8D%AE%E6%94%BE%E5%85%A5D%E6%88%96%E8%80%85Q%E5%AF%84%E5%AD%98%E5%99%A8\" >\u5c06\u6570\u636e\u653e\u5165D\u6216\u8005Q\u5bc4\u5b58\u5668<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E8%BF%9B%E8%A1%8C%E6%89%80%E9%9C%80%E8%A6%81%E7%9A%84%E8%BF%90%E7%AE%97\" >\u8fdb\u884c\u6240\u9700\u8981\u7684\u8fd0\u7b97<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-4'><a class=\"ez-toc-link ez-toc-heading-11\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E4%BB%8ED%E6%88%96%E8%80%85Q%E5%AF%84%E5%AD%98%E5%99%A8%E4%B8%AD%E5%8F%96%E5%87%BA%E8%BF%90%E7%AE%97%E7%BB%93%E6%9E%9C\" >\u4eceD\u6216\u8005Q\u5bc4\u5b58\u5668\u4e2d\u53d6\u51fa\u8fd0\u7b97\u7ed3\u679c<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-12\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E7%BC%96%E8%AF%91%E5%99%A8%E9%80%89%E9%A1%B9%E8%AE%BE%E7%BD%AE\" >\u7f16\u8bd1\u5668\u9009\u9879\u8bbe\u7f6e<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-13\" href=\"https:\/\/blog.cauchyschwarz.com\/?p=697\/#%E4%B8%80%E4%B8%AA%E7%9F%A9%E9%98%B5%E4%B9%98%E6%B3%95%E7%9A%84%E4%BE%8B%E5%AD%90\" >\u4e00\u4e2a\u77e9\u9635\u4e58\u6cd5\u7684\u4f8b\u5b50<\/a><\/li><\/ul><\/li><\/ul><\/li><\/ul><\/nav><\/div>\n<h1 class=\"wp-block-heading\" id=\"neon%E6%8A%80%E6%9C%AF\"><span class=\"ez-toc-section\" id=\"NEON%E6%8A%80%E6%9C%AF\"><\/span>NEON\u6280\u672f<span class=\"ez-toc-section-end\"><\/span><\/h1>\n\n\n\n<p>\u5173\u4e8eARM\u5904\u7406\u5668NEON\u6280\u672f\u7684\u4ecb\u7ecd\u53ef\u4ee5\u53c2\u8003\u7ef4\u57fa\u767e\u79d1\u3001\u767e\u5ea6\u767e\u79d1\u8bcd\u6761\u3001ARM\u5b98\u7f51\uff0c\u5173\u4e8eNEON\u6280\u672f\u7684\u4f7f\u7528\u53ef\u4ee5\u53c2\u8003\u7f51\u4e0a\u7684\u5404\u79cd\u535a\u5ba2\uff0c\u4f46\u662f\u4f3c\u4e4e\u6765\u81ea\u4e8eARM\u5b98\u7f51\u7684\u8fd9\u4efd<a href=\"http:\/\/infocenter.arm.com\/help\/topic\/com.arm.doc.den0018a\/\">NEON Programmer&#8217;s Guide<\/a>(\u6216\u8005\u4ece<a href=\"https:\/\/file+.vscode-resource.vscode-webview.net\/blog\/assets\/public\/manual\/DEN0018A_neon_programmers_guide.pdf\">\u672c\u7f51\u7ad9\u76f4\u63a5\u4e0b\u8f7d<\/a>)\u6700\u4e3a\u5168\u9762\u53ef\u9760\u3002<\/p>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"neon%E6%8A%80%E6%9C%AF%E7%AE%80%E4%BB%8B\"><span class=\"ez-toc-section\" id=\"NEON%E6%8A%80%E6%9C%AF%E7%AE%80%E4%BB%8B\"><\/span>NEON\u6280\u672f\u7b80\u4ecb<span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<p>\u5904\u7406\u5668\u5904\u7406\u6570\u636e\u7684\u6307\u4ee4\u5927\u81f4\u6709\u4e09\u79cd\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\"><li>Single Instruction Single Data \u4e00\u4e2a\u6307\u4ee4\u7684\u6e90\u64cd\u4f5c\u6570\u548c\u76ee\u7684\u64cd\u4f5c\u6570\u90fd\u662f\u552f\u4e00\u7684\u3002\u8fd9\u91cc\u7684\u552f\u4e00\u662f\u6307\u6e90\u64cd\u4f5c\u6570\u548c\u76ee\u7684\u64cd\u4f5c\u6570\u90fd\u662f\u6807\u91cf(scalar)\/\u5355\u4e2a\u7684\u6570\u636e\u3002<br>\u6bd4\u5982<code>add r0, r5<\/code>\u8fd9\u79cd\u6307\u4ee4\u6240\u505a\u7684\u5c31\u4ec5\u4ec5\u662f\u628a<em>r0<\/em>,<em>r5<\/em>\u8fd9\u4e24\u4e2a\u5bc4\u5b58\u5668\u91cc\u7684\u4e1c\u897f\u52a0\u8d77\u6765\u3002<\/li><li>Single Instruction Multiple Data(vector mode) \u4e00\u4e2a\u6307\u4ee4\u7684\u6e90\u64cd\u4f5c\u6570\u548c\u76ee\u7684\u64cd\u4f5c\u6570\u662f\u5411\u91cf(\u6216\u8005\u8bf4\u6570\u7ec4)\uff0c\u8be5\u6307\u4ee4\u4f1a\u4f9d\u6b21\/\u987a\u5e8f\u4e3a\u6570\u7ec4\u7684\u6bcf\u4e2a\u5143\u7d20\u6267\u884c\u76f8\u5e94\u7684\u64cd\u4f5c\u3002<br>\u6bd4\u5982<code>VADD.F32 S24, S8, S16<\/code>\u8fd9\u6761\u6307\u4ee4\u5b9e\u9645\u4e0a\u4f1a\u4f9d\u6b21\u6267\u884c\u4e0b\u9762\u56db\u4e2a\u64cd\u4f5c\uff1a<code>S24 = S8 + S16 S25 = S9 + S17 S26 = S10 + S18 S27 = S11 + S20<\/code><\/li><li>Single Instruction Multiple Data(packed data) \u4e00\u4e2a\u6307\u4ee4\u7684\u6e90\u64cd\u4f5c\u6570\u548c\u76ee\u7684\u64cd\u4f5c\u6570\u662f\u5411\u91cf\uff0c\u8be5\u6307\u4ee4\u4f1a\u540c\u65f6\u4e3a\u6570\u7ec4\u7684\u6bcf\u4e2a\u5143\u7d20\u6267\u884c\u76f8\u5e94\u64cd\u4f5c\u3002\u7531\u4e8e\u8fd9\u4e2a<strong>\u540c\u65f6<\/strong>\uff0c\u56e0\u6b64\u8fd9\u79cd\u6307\u4ee4\u5c31\u6bd4\u4e0a\u9762\u7b2c\u4e8c\u79cd\u987a\u5e8f\u6267\u884c\u7684\u5411\u91cf\u6a21\u5f0f\u7684SIMD\u66f4\u6709\u6548\u7387\u3002\u81f3\u4e8e\u6240\u8c13\u7684<strong>packed data<\/strong>\u7684\u542b\u4e49\u662f\u8fd9\u79cd\u6307\u4ee4\u7684\u6e90\u64cd\u4f5c\u6570\u548c\u76ee\u7684\u64cd\u4f5c\u6570\u5fc5\u987b\u653e\u5728\u7279\u6b8a\u7684\u5bc4\u5b58\u5668(\u6240\u8c13\u7684<em>D<\/em>\u5bc4\u5b58\u5668\u548c<em>Q<\/em>\u5bc4\u5b58\u5668)\u91cc\u624d\u80fd\u8fdb\u884c\u8fd9\u79cd\u8fd0\u7b97\u3002ARM\u79f0\u8fd9\u79cd\u6280\u672f\u4e3a<em>Advanced SIMD technology<\/em>\u6216\u8005<em>NEON technology<\/em>\u3002<\/li><\/ol>\n\n\n\n<h1 class=\"wp-block-heading\" id=\"%E5%9C%A8cc%E4%B8%AD%E4%BD%BF%E7%94%A8neon%E6%8A%80%E6%9C%AF\"><span class=\"ez-toc-section\" id=\"%E5%9C%A8CC%E4%B8%AD%E4%BD%BF%E7%94%A8NEON%E6%8A%80%E6%9C%AF\"><\/span>\u5728C\/C++\u4e2d\u4f7f\u7528NEON\u6280\u672f<span class=\"ez-toc-section-end\"><\/span><\/h1>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"%E9%A6%96%E5%85%88%E7%A1%AE%E5%AE%9A%E8%83%BD%E5%90%A6%E4%BD%BF%E7%94%A8neon%E6%8A%80%E6%9C%AF\"><span class=\"ez-toc-section\" id=\"%E9%A6%96%E5%85%88%E7%A1%AE%E5%AE%9A%E8%83%BD%E5%90%A6%E4%BD%BF%E7%94%A8NEON%E6%8A%80%E6%9C%AF\"><\/span>\u9996\u5148\u786e\u5b9a\u80fd\u5426\u4f7f\u7528NEON\u6280\u672f<span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<ul class=\"wp-block-list\"><li>ARMv7\u5f00\u59cb\u7684\u5904\u7406\u5668\u624d\u6709NEON\u6280\u672f\uff0c\u53ef\u4ee5\u901a\u8fc7<code>\/proc\/cpuinfo<\/code>\u68c0\u67e5\u4e00\u4e0b\u5904\u7406\u5668\u662f\u5426\u5e26<em>neon<\/em>\u6269\u5c55\u5c31\u884c\uff1b<\/li><li>\u901a\u8fc7\u5728\u4ee3\u7801\u4e2d\u68c0\u67e5\u7f16\u8bd1\u94fe\u662f\u5426\u5177\u6709<code>__ARM_NEON__<\/code>\u8fd9\u4e2a\u9884\u5b9a\u4e49\u5b8f\u786e\u8ba4\u662f\u5426\u53ef\u4ee5\u4f7f\u7528NEON\u6280\u672f\uff0c\u5373\u4ee3\u7801\u5e94\u8be5\u662f\u8fd9\u6837\u7684\uff1a<code>#ifdef __ARM_NEON__ <em>\/\/\u4f7f\u7528\u4e86NEON\u6280\u672f\u7684\u4ee3\u7801<\/em> #else <em>\/\/\u6ca1\u6709\u4f7f\u7528NEON\u6280\u672f\u7684\u4ee3\u7801<\/em> #endif<\/code><\/li><\/ul>\n\n\n\n<h2 class=\"wp-block-heading\" id=\"e59ca8cce4b8ade4bdbfe794a8neone68a80e69caf-1\"><span class=\"ez-toc-section\" id=\"%E5%9C%A8CC%E4%B8%AD%E4%BD%BF%E7%94%A8NEON%E6%8A%80%E6%9C%AF-2\"><\/span>\u5728C\/C++\u4e2d\u4f7f\u7528NEON\u6280\u672f<span class=\"ez-toc-section-end\"><\/span><\/h2>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"neon%E6%8A%80%E6%9C%AF%E5%8E%9F%E7%90%86\"><span class=\"ez-toc-section\" id=\"NEON%E6%8A%80%E6%9C%AF%E5%8E%9F%E7%90%86\"><\/span>NEON\u6280\u672f\u539f\u7406<span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>\u9996\u5148\u53d9\u8ff0\u4e00\u4e0b\u6211\u6240\u7406\u89e3\u7684NEON\u6280\u672f\uff0c\u5728ARM\u5904\u7406\u5668\u5b58\u5728\u7740\u6240\u8c13\u7684<em>D<\/em>\u5bc4\u5b58\u5668\u548c<em>Q<\/em>\u5bc4\u5b58\u5668\u3002<em>D<\/em>\u5bc4\u5b58\u5668\u957f\u5ea6\u4e3a64\u4f4d\uff0c\u53ef\u4ee5\u5b582\u4e2a32\u4f4d\u957f\u5ea6\u7684\u6570\u636e\u7c7b\u578b(\u5982int_32\/uint_32)\u6216\u80054\u4e2a16\u4f4d\u7684\u6570\u636e\u7c7b\u578b\u6216\u80058\u4e2a8\u4f4d\u7684\u6570\u636e\u7c7b\u578b(\u603b\u4e4b\u5c31\u662f\u4e00\u6837\u7684\u6570\u636e\u7c7b\u578b\u51d1\u9f5064\u4f4d)\uff0c\u8fd9\u5bc4\u5b58\u5668\u91cc\u7684\u6bcf\u4e2a\u6570\u636e\u7c7b\u578b\u53eb\u505a\u4e00\u4e2a<em>lane<\/em>(\u6216\u8005\u8bf4\u5bf9\u5e94\u4e00\u4e2a<em>lane<\/em>\u5427)\u3002<em>Q<\/em>\u5bc4\u5b58\u5668\u4e5f\u662f\u7c7b\u4f3c\u7684\uff0c\u53ea\u662f<em>Q<\/em>\u5bc4\u5b58\u5668\u957f\u5ea6\u66f4\u957f\uff0c\u6709128\u4f4d\u800c\u5df2\u3002<em>D<\/em>\u5bc4\u5b58\u5668\u548c<em>Q<\/em>\u5bc4\u5b58\u5668\u91cc\u7684\u6570\u636e\u5c31\u53ef\u4ee5\u7528\u6765\u6267\u884c\u4e00\u5f00\u59cb\u8bb2\u7684SIMD(packed data)\u8fd9\u79cd\u9ad8\u6548\u7387\u6307\u4ee4\u3002\u6bd4\u5982\u4e24\u4e2a<em>D<\/em>\u5bc4\u5b58\u5668D1,D2\u90fd\u5305\u542b8\u4e2auint_8\u7684\u6570\u636e\uff0c\u5728\u5c06\u8fd9\u4e24\u4e2a\u5bc4\u5b58\u5668\u6267\u884c\u76f8\u52a0\u64cd\u4f5c\u65f6\uff0c\u5904\u7406\u5668\u4f1a\u628a\u4e24\u4e2a\u5904\u7406\u5668\u5bf9\u5e94\u4f4d\u7f6e\u4e0a\u7684uint_8\u76f8\u52a0\uff0c\u7136\u540e\u5b58\u5230\u76ee\u7684\u5bc4\u5b58\u5668\u6bd4\u5982D3\u91cc\u3002\u6bd4\u5982D1\u7684\u7b2c\u4e00\u4e2auint_8\u548cD2\u7684\u7b2c\u4e00\u4e2auint_8\u76f8\u52a0\uff0c\u7ed3\u679c\u5b58\u5230D3\u7684\u7b2c\u4e00\u4e2auint_8\u91cc\uff0cD1\u7684\u7b2c\u4e8c\u4e2auint_8\u548cD2\u7684\u7b2c\u4e8c\u4e2auint_8\u76f8\u52a0\uff0c\u7ed3\u679c\u5b58\u5230D3\u7684\u7b2c\u4e8c\u4e2auint_8\u91cc\uff0c\u4ee5\u6b64\u7c7b\u63a8\u3002<\/p>\n\n\n\n<p>\u6240\u4ee5\uff0c\u53ea\u8981\u6211\u4eec\u80fd\u591f\u628a\u6211\u4eec\u8981\u4f5c\u8fd0\u7b97\u7684\u6570\u636e\u7ed9\u653e\u5165<em>D<\/em>\u6216\u8005<em>Q<\/em>\u5bc4\u5b58\u5668\u91cc\uff0c\u6267\u884c\u76f8\u5e94\u7684\u8fd0\u7b97\uff0c\u518d\u628a\u6570\u636e\u4ece<em>D<\/em>\u6216\u8005<em>Q<\/em>\u5bc4\u5b58\u5668\u91cc\u53d6\u51fa\u6765\uff0c\u5c31\u8fbe\u5230\u4e86\u4f7f\u7528NEON\u6280\u672f\u52a0\u901f(\u52a0\u901f\u4e0d\u52a0\u901f\u8fd8\u5f97\u770b\u5b9e\u9645\u6548\u679c)\u8fd0\u7b97\u7684\u76ee\u7684\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"%E4%BD%BF%E7%94%A8neon-intrinsics\"><span class=\"ez-toc-section\" id=\"%E4%BD%BF%E7%94%A8NEON_Intrinsics\"><\/span>\u4f7f\u7528NEON Intrinsics<span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>\u6700\u76f4\u63a5\u7684\u4f7f\u7528NEON\u6280\u672f\u7684\u65b9\u5f0f\u5f53\u7136\u662f\u4f7f\u7528\u6c47\u7f16\u4e86\uff0c\u4f46\u662f\u6ca1\u591a\u5c11\u4eba\u6709\u5174\u8da3\u548c\u80fd\u529b\u53bb\u6298\u817e\u6c47\u7f16\u5427\uff0c\u6216\u8005\u4eba\u624b\u5199\u51fa\u6765\u7684\u6c47\u7f16\u4e0d\u4e00\u5b9a\u771f\u6709\u7a0b\u5e8f\u81ea\u52a8\u751f\u6210\u7684\u597d\u3002\u4e3a\u4e86\u65b9\u4fbfC\/C++\u7a0b\u5e8f\u5458\u4e5f\u80fd\u4f7f\u7528NEON\u6280\u672f\uff0c\u4e8e\u662f\u5c31\u6709\u4e86NEON Intrinsics\u5b58\u5728\u7684\u9700\u6c42\u3002\u6211\u4e0d\u77e5\u9053NEON Intrinsics\u7684\u4e2d\u6587\u7ffb\u8bd1\u662f\u4ec0\u4e48\uff0c\u5927\u81f4\u5c31\u662f\u5185\u8054\u51fd\u6570\u4e00\u6837\u7684\u4e1c\u897f\uff0c\u901a\u8fc7\u7c7b\u4f3c\u51fd\u6570\u4e00\u6837\u7684\u63a5\u53e3\u53bb\u8c03\u7528\u76f8\u5173\u7684NEON\u6307\u4ee4\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"%E5%8C%85%E5%90%AB%E5%BF%85%E8%A6%81%E7%9A%84%E5%A4%B4%E6%96%87%E4%BB%B6\"><span class=\"ez-toc-section\" id=\"%E5%8C%85%E5%90%AB%E5%BF%85%E8%A6%81%E7%9A%84%E5%A4%B4%E6%96%87%E4%BB%B6\"><\/span>\u5305\u542b\u5fc5\u8981\u7684\u5934\u6587\u4ef6<span class=\"ez-toc-section-end\"><\/span><\/h4>\n\n\n\n<p>\u60f3\u8981\u4f7f\u7528NEON Intrinsics\u9700\u8981\u5305\u542b<em>arm_neon.h<\/em>\u8fd9\u4e2a\u5934\u6587\u4ef6\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"%E5%B0%86%E6%95%B0%E6%8D%AE%E6%94%BE%E5%85%A5d%E6%88%96%E8%80%85q%E5%AF%84%E5%AD%98%E5%99%A8\"><span class=\"ez-toc-section\" id=\"%E5%B0%86%E6%95%B0%E6%8D%AE%E6%94%BE%E5%85%A5D%E6%88%96%E8%80%85Q%E5%AF%84%E5%AD%98%E5%99%A8\"><\/span>\u5c06\u6570\u636e\u653e\u5165<em>D<\/em>\u6216\u8005<em>Q<\/em>\u5bc4\u5b58\u5668<span class=\"ez-toc-section-end\"><\/span><\/h4>\n\n\n\n<p>\u53ea\u6709\u7279\u6b8a\u6570\u636e\u7c7b\u578b\u7684\u53d8\u91cf\u624d\u53ef\u4ee5\u653e\u5165<em>D<\/em>\u6216\u8005<em>Q<\/em>\u5bc4\u5b58\u5668\u91cc\u8fdb\u884c\u64cd\u4f5c\uff0c\u8fd9\u4e9b\u7279\u6b8a\u7684\u6570\u636e\u7c7b\u578b\u662f\u8fd9\u6837\u5b50\u7684\uff1a<code>&lt;type&gt;&lt;size&gt;x&lt;number_of_lanes&gt;_t<\/code>\u3002<\/p>\n\n\n\n<p>\u6bd4\u5982:<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><em>int16x4_t<\/em>\u8868\u793a4\u4e2a16\u4f4d\u7684short\uff0c\u8fd94\u4e2a16\u4f4d\u7684short\u53ef\u4ee5\u4e00\u8d77\u653e\u5728\u4e00\u4e2a<em>D<\/em>\u5bc4\u5b58\u5668\u4e2d<\/li><li><em>float32x4_t<\/em>\u8868\u793a4\u4e2a32\u4f4d\u7684float\uff0c\u8fd94\u4e2afloat\u53ef\u4ee5\u4e00\u8d77\u653e\u5728\u4e00\u4e2a<em>Q<\/em>\u5bc4\u5b58\u5668\u4e2d<\/li><\/ul>\n\n\n\n<p>\u5728\u521d\u59cb\u5316\u4e0a\u8ff0\u6570\u636e\u7c7b\u578b\u7684\u65f6\u5019\uff0c\u662f\u4e0d\u80fd\u76f4\u63a5\u521d\u59cb\u5316\u7684\uff0c\u9700\u8981\u8c03\u7528\u76f8\u5e94\u7684NEON Intrinsics\u51fd\u6570\u3002<br>NEON Intrinsics\u51fd\u6570\u662f\u8fd9\u6837\u5b50\u7684\uff1a<code>&lt;opname&gt;&lt;flags&gt;_&lt;type&gt;<\/code><\/p>\n\n\n\n<p><em>opname<\/em>\u662f\u8fd0\u7b97\u6216\u8005\u64cd\u4f5c\u7c7b\u578b\uff0c<em>flags<\/em>\u5982\u679c\u6ca1\u6709\u7684\u8bdd\uff0c\u5c31\u8868\u793a\u4e00\u822c\u7684\u3001\u5728<em>D<\/em>\u5bc4\u5b58\u5668\u4e0a\u8fdb\u884c\u7684\u64cd\u4f5c\uff0c\u6709\u7684\u8bdd\uff0c\u4e00\u822c\u662f<em>q<\/em>(\u8868\u793a\u662f\u5bf9<em>Q<\/em>\u5bc4\u5b58\u5668\u8fdb\u884c\u64cd\u4f5c)\u6216\u8005<em>l<\/em>(\u7ed3\u679c\u6570\u636e\u7c7b\u578b\u6bd4\u6e90\u64cd\u4f5c\u6570\u6570\u636e\u7c7b\u578b\u957f)\u4e4b\u7c7b\u7684\u503c\u3002<em>type<\/em>\u8868\u793a\u5bc4\u5b58\u5668\u91cc\u5355\u4e2a\u6570\u636e\u7c7b\u578b(\u6bcf\u4e2a<em>lane<\/em>\u7684\u6570\u636e\u7c7b\u578b)\u3002<\/p>\n\n\n\n<p>\u6bd4\u5982\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li><em>vmul_s16<\/em>\u662f\u5c06\u4e24\u4e2a<em>D<\/em>\u5bc4\u5b58\u5668\u91cc\u5bf9\u5e94\u7684short\u76f8\u4e58<\/li><li><em>vaddl_u8<\/em>\u628a\u4e24\u4e2a<em>D<\/em>\u5bc4\u5b58\u5668\u91cc\u7684uint8x8_t\u76f8\u52a0\uff0c\u7ed3\u679c\u7ed9\u653e\u5728\u4e86<em>Q<\/em>\u5bc4\u5b58\u5668\u7684uint16x8_t\u91cc\u9762\u3002<\/li><\/ul>\n\n\n\n<p>\u56de\u5230\u5177\u4f53\u7684\u521d\u59cb\u5316\u8fd9\u4e9b\u7279\u6b8a\u7c7b\u578b\u53d8\u91cf\u7684\u95ee\u9898\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\"><li>\u5c06\u521d\u59cb\u503c\u90fd\u8bbe\u4e3a0<code>uint8x8 a; a = vdup_n_u8(0);<\/code><\/li><li>\u5c06\u521d\u59cb\u503c\u8bbe\u4e3a\u7279\u5b9a\u7684\u503c<code>uintix8 a = vreinterpret_u8_u64(vcreate_u64(0x123456789ABCDEFULL));<\/code><\/li><li>\u8bbe\u7f6e\u67d0\u4e2a<em>lane<\/em>\u7684\u503c \u4f7f\u7528<code>VSET_LANE<\/code>\uff1aResult_t vset_lane_type(Scalar_t N, Vector_t M, int n); Result_t vsetq_lane_type(scalar_t N, Vector_t M, int n);<\/li><li>\u6839\u636e\u5df2\u6709\u67d0\u4e2a\u6570\u7ec4\u7684\u503c\/\u67d0\u5757\u5185\u5b58\u521d\u59cb\u5316 \u4f7f\u7528<code>VLD1<\/code>Intrinsic\uff1aResult_t vld1_type(Scalar_t *N); Result_t vld1q_type(Scalar_t *N);<\/li><\/ul>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"%E8%BF%9B%E8%A1%8C%E6%89%80%E9%9C%80%E8%A6%81%E7%9A%84%E8%BF%90%E7%AE%97\"><span class=\"ez-toc-section\" id=\"%E8%BF%9B%E8%A1%8C%E6%89%80%E9%9C%80%E8%A6%81%E7%9A%84%E8%BF%90%E7%AE%97\"><\/span>\u8fdb\u884c\u6240\u9700\u8981\u7684\u8fd0\u7b97<span class=\"ez-toc-section-end\"><\/span><\/h4>\n\n\n\n<p>\u6bd4\u5982\u5bf9\u4e8e\u4e09\u4e2a<code>int16x4_t<\/code>\u7c7b\u578b\u7684\u53d8\u91cf<code>sum<\/code>\uff0c<code>a<\/code>\uff0c<code>x<\/code>\uff0c\u8981\u8ba1\u7b97<code>sum = a * x + sum<\/code>\uff0c\u53ef\u4ee5\u8fd9\u6837\u5199\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code lang=\"c\" class=\"language-c\">sum = vmla_s16(sum, a, x);\n<\/code><\/pre>\n\n\n\n<p>\u8fd9\u5b8c\u5168\u53d6\u51b3\u4e8e\u81ea\u5df1\u7684\u9700\u8981\u800c\u9009\u62e9\u5408\u9002\u7684NEON Intrinsic\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\" id=\"%E4%BB%8Ed%E6%88%96%E8%80%85q%E5%AF%84%E5%AD%98%E5%99%A8%E4%B8%AD%E5%8F%96%E5%87%BA%E8%BF%90%E7%AE%97%E7%BB%93%E6%9E%9C\"><span class=\"ez-toc-section\" id=\"%E4%BB%8ED%E6%88%96%E8%80%85Q%E5%AF%84%E5%AD%98%E5%99%A8%E4%B8%AD%E5%8F%96%E5%87%BA%E8%BF%90%E7%AE%97%E7%BB%93%E6%9E%9C\"><\/span>\u4ece<em>D<\/em>\u6216\u8005<em>Q<\/em>\u5bc4\u5b58\u5668\u4e2d\u53d6\u51fa\u8fd0\u7b97\u7ed3\u679c<span class=\"ez-toc-section-end\"><\/span><\/h4>\n\n\n\n<ul class=\"wp-block-list\"><li>\u53ef\u4ee5\u4f7f\u7528<code>VST1<\/code>Intrinsic\u628a\u7ed3\u679c\u5b58\u5230\u5185\u5b58\u4e2d\uff1avoid vst1_type(Scalar_t *N, Vector_t M); void vst1q_type(Scalar_t *N, Vector_t M);<\/li><li>\u4e5f\u53ef\u4ee5\u7528<code>VGET_LANE<\/code>Intrinsic\u76f4\u63a5\u83b7\u53d6\u6bcf\u4e00\u4e2a<em>lane<\/em>\u7684\u503c\uff1aResult_t vget_lane_type(Vector_t N, int n); Result_t vgetq_lane_type(Vector_t N, int n);<\/li><\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"%E7%BC%96%E8%AF%91%E5%99%A8%E9%80%89%E9%A1%B9%E8%AE%BE%E7%BD%AE\"><span class=\"ez-toc-section\" id=\"%E7%BC%96%E8%AF%91%E5%99%A8%E9%80%89%E9%A1%B9%E8%AE%BE%E7%BD%AE\"><\/span>\u7f16\u8bd1\u5668\u9009\u9879\u8bbe\u7f6e<span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<ul class=\"wp-block-list\"><li>\u5728\u7f16\u8bd1\u5668\u7684\u547d\u4ee4\u9009\u9879\u4e2d\u6307\u5b9a\u8981\u4f7f\u7528NEON\u6280\u672f\uff1a<ul><li><code>-mfpu=neon<\/code><\/li><li><code>-mcpu=&lt;\u4e00\u4e2a\u652f\u6301NEON\u7684\u5904\u7406\u5668&gt;<\/code><\/li><li><code>-ftree-vectorize<\/code><\/li><li><code>-mfloat-abi=softfp<\/code>&nbsp;<code>-mfloat-abi<\/code>\u9009\u9879\u6709\u4e09\u4e2a\u53ef\u53d6\u7684\u503c\uff0c<em>soft<\/em>,<em>softfp<\/em>,<em>hard<\/em>\uff0c\u6211\u53ea\u7528\u8fc7<em>softfp<\/em>\uff0c\u5173\u4e8e\u8fd9\u4e09\u4e2a\u503c\u7684\u5177\u4f53\u542b\u4e49\u53ef\u4ee5\u67e5\u770bmanual\u3002<\/li><\/ul>\u5728\u5b9e\u9645\u4f7f\u7528\u8fc7\u7a0b\u4e2d\uff0c\u6211\u6ca1\u7528\u8fc7<code>-ftree-vectorize<\/code>\u8fd9\u4e2a\u9009\u9879\uff0c\u4f7f\u7528<code>-O3<\/code>\u9009\u9879\u5c31\u5df2\u7ecf\u9ed8\u8ba4\u5305\u542b\u4e86\u8fd9\u4e2a\u9009\u9879\u3002<\/li><\/ul>\n\n\n\n<h3 class=\"wp-block-heading\" id=\"%E4%B8%80%E4%B8%AA%E7%9F%A9%E9%98%B5%E4%B9%98%E6%B3%95%E7%9A%84%E4%BE%8B%E5%AD%90\"><span class=\"ez-toc-section\" id=\"%E4%B8%80%E4%B8%AA%E7%9F%A9%E9%98%B5%E4%B9%98%E6%B3%95%E7%9A%84%E4%BE%8B%E5%AD%90\"><\/span>\u4e00\u4e2a\u77e9\u9635\u4e58\u6cd5\u7684\u4f8b\u5b50<span class=\"ez-toc-section-end\"><\/span><\/h3>\n\n\n\n<p>\u6700\u540e\u76f4\u63a5\u8d34\u4e0a\u4e00\u4e2aNEON\u52a0\u901f\u7684\u77e9\u9635\u4e58\u6cd5\u7684\u4f8b\u5b50\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code lang=\"c\" class=\"language-c\">void neon_float_matrix_vec_mul(const float *matrix, const float *vec,\n                             float *vec_ret, int row, int col) {\n\n    float32x4_t sum, a, x;\n    for (int i = 0; i &lt; row; i++) {\n        sum = vdupq_n_s32(0);\n        for (int j = 0; j &lt; col; j += 4) {\n            a = vld1q_f32(matrix + i*col + j);\n            x = vld1q_f32(vec + j);\n            sum = vmlaq_f32(sum, a, x);\n        }\n        vec_ret[i] = vgetq_lane_f32(sum, 0)\n                     +vgetq_lane_f32(sum, 1)\n                     +vgetq_lane_f32(sum, 2)\n                     +vgetq_lane_f32(sum, 3);\n    }\n}\n<\/code><\/pre>\n\n\n\n<p>\u5982\u679c\u9700\u8981\u5904\u7406\u56fe\u50cf\u6216\u8005\u58f0\u97f3\u8fd9\u79cd\u591a\u8def\u6570\u636e\u4ea4\u66ff\u5b58\u50a8\u5728\u4e00\u5757\u7684\u4e1c\u897f\uff0c\u53ef\u4ee5\u53c2\u8003\u624b\u518c\u4e2d<strong>Constructing multiple vectors from interleaved memory<\/strong>\u8fd9\u4e00\u90e8\u5206\u5185\u5bb9\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>NEON\u6280\u672f \u5173\u4e8eARM\u5904\u7406\u5668NEON\u6280\u672f\u7684\u4ecb\u7ecd\u53ef\u4ee5\u53c2\u8003\u7ef4\u57fa\u767e\u79d1\u3001\u767e\u5ea6\u767e\u79d1\u8bcd\u6761\u3001ARM\u5b98\u7f51\uff0c\u5173\u4e8eNEON\u6280\u672f\u7684\u4f7f\u7528\u53ef\u4ee5\u53c2\u8003\u7f51\u4e0a\u7684\u5404\u79cd\u535a\u5ba2\uff0c\u4f46\u662f\u4f3c\u4e4e\u6765\u81ea\u4e8eARM\u5b98\u7f51\u7684\u8fd9\u4efdNEON Programmer&#038;&#46;&#46;&#46;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10],"tags":[67],"class_list":["post-697","post","type-post","status-publish","format-standard","hentry","category-10","tag-neon"],"_links":{"self":[{"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=\/wp\/v2\/posts\/697","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=697"}],"version-history":[{"count":2,"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=\/wp\/v2\/posts\/697\/revisions"}],"predecessor-version":[{"id":701,"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=\/wp\/v2\/posts\/697\/revisions\/701"}],"wp:attachment":[{"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=697"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=697"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.cauchyschwarz.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=697"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}