{"id":1688,"date":"2024-12-09T14:20:22","date_gmt":"2024-12-09T05:20:22","guid":{"rendered":"https:\/\/www.yanagichiaki.jp\/?p=1688"},"modified":"2024-12-21T22:01:22","modified_gmt":"2024-12-21T13:01:22","slug":"computer-architecture-labs-guidebook-hitsz","status":"publish","type":"post","link":"https:\/\/yanagichiaki.jp\/index.php\/2024\/12\/09\/computer-architecture-labs-guidebook-hitsz\/","title":{"rendered":"Computer Architecture Labs Guidebook HITSZ"},"content":{"rendered":"\n<figure class=\"wp-block-image size-full is-style-shadow\"><img decoding=\"async\" width=\"1988\" height=\"1180\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALabGuideBookCover.png\" alt=\"\" class=\"wp-image-1758\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALabGuideBookCover.png 1988w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALabGuideBookCover-300x178.png 300w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALabGuideBookCover-1024x608.png 1024w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALabGuideBookCover-768x456.png 768w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALabGuideBookCover-1536x912.png 1536w\" sizes=\"(max-width: 1988px) 100vw, 1988px\" \/><\/figure>\n\n\n\n<p class=\"is-style-big_icon_point\">\u30cf\u30eb\u30d3\u30f3\u5de5\u696d\u5927\u5b66\uff08\u6df1\u5733\uff09\u2022 2024 \u2022 \u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30fb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3 Lab Guidebook 2024\u30fb\u65e5\u672c\u8a9e\u8a33\u7248<br>HITSZ \u8ba1\u7b97\u673a\u4f53\u7cfb\u7ed3\u6784\u5b9e\u9a8c\u6307\u5bfc\u4e66 2024<\/p>\n\n\n\n<p class=\"has-border -border03 is-style-icon_info\">\u5fa1\u8cea\u554f\u304c\u5fa1\u5ea7\u3044\u307e\u3057\u305f\u3089\u3001\u3053\u306e\u30da\u30fc\u30b8\u306e\u4e0b\u90e8\u306b\u3042\u308b\u30b3\u30e1\u30f3\u30c8\u6b04\u3092\u5fa1\u5229\u7528\u304f\u3060\u3055\u3044\u3002<br><span class=\"swl-marker mark_yellow\">\u4ef0\u305b\u4e8b\u6709\u4e4b\u5019\u30cf\u30cf<\/span>\u3001<span class=\"swl-marker mark_blue\">\u6b64\u4e01\u4e4b\u4e0b\u30cb\u30a2\u30eb\u610f\u898b\u4e4b\u6b04\u30f2\u7528\u30f0\u7d66\u30d8<\/span>\u3002<\/p>\n\n\n\n<p class=\"is-style-big_icon_batsu\">\u5f53\u30b5\u30a4\u30c8\u5185\u306e\u30b3\u30f3\u30c6\u30f3\u30c4\u306e\u7121\u65ad\u8ee2\u8f09\u3001\u5f15\u7528\u3001\u30b3\u30d4\u30fc\u306f\u7981\u6b62\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n\n<div class=\"swell-block-button green_ is-style-btn_shiny\"><a href=\"https:\/\/comp-arch.pages.dev\/\" class=\"swell-block-button__link\" data-has-icon=\"1\"><svg class=\"__icon\" height=\"1em\" width=\"1em\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" aria-hidden=\"true\" viewBox=\"0 0 48 48\"><path d=\"M21.2 30.2c-.5 0-1-.2-1.4-.6l-.7-.7c-2.3-2.3-3.5-5.3-3.5-8.5s1.2-6.2 3.5-8.5l7.1-7.1c2.3-2.3 5.3-3.5 8.5-3.5s6.2 1.2 8.5 3.5c4.7 4.7 4.7 12.3 0 17l-3.5 3.5c-.8.8-2 .8-2.8 0-.8-.8-.8-2 0-2.8l3.5-3.5c3.1-3.1 3.1-8.2 0-11.3-1.5-1.5-3.5-2.3-5.7-2.3-2.1 0-4.2.8-5.7 2.3l-7.1 7.1c-1.5 1.5-2.3 3.5-2.3 5.7s.8 4.2 2.3 5.7l.7.7c.8.8.8 2 0 2.8-.4.3-.9.5-1.4.5z\"><\/path><path d=\"M13.4 46.6c-3.1 0-6.1-1.2-8.5-3.5-2.3-2.3-3.5-5.3-3.5-8.5s1.2-6.2 3.5-8.5l3.5-3.5c.8-.8 2-.8 2.8 0 .8.8.8 2 0 2.8l-3.5 3.5c-1.5 1.5-2.3 3.5-2.3 5.7 0 2.1.8 4.2 2.3 5.7 3.1 3.1 8.2 3.1 11.3 0l7.1-7.1c1.5-1.5 2.3-3.5 2.3-5.7 0-2.1-.8-4.2-2.3-5.7l-.7-.7c-.8-.8-.8-2 0-2.8.8-.8 2-.8 2.8 0l.7.7c2.3 2.3 3.5 5.3 3.5 8.5s-1.2 6.2-3.5 8.5l-7.1 7.1c-2.3 2.3-5.3 3.5-8.4 3.5z\"><\/path><\/svg><span>\u539f\u7248GuideBook\u3078<\/span><\/a><\/div>\n\n\n\n<div class=\"swell-block-button blue_ is-style-btn_shiny\"><a href=\"https:\/\/www.yanagichiaki.jp\/index.php\/2024\/11\/14\/computer-architecture-labs\/\" class=\"swell-block-button__link\" data-has-icon=\"1\"><svg class=\"__icon\" height=\"1em\" width=\"1em\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" aria-hidden=\"true\" viewBox=\"0 0 48 48\"><path d=\"m24 8.8-1.4-1.4c-2.9-2.8-7.2-4.1-11.7-3-5.1 1.2-8.5 5.8-8.9 11C1 31.7 24 44 24 44s23-12.3 22-28.6c-.3-5.2-3.8-9.8-8.9-11-4.5-1.1-8.8.3-11.7 3L24 8.8z\"><\/path><\/svg><span>\u53c2\u8003\u89e3\u7b54\u3078<\/span><\/a><\/div>\n\n\n\n<h3 class=\"wp-block-heading\">\u30b3\u30fc\u30b9\u6982\u8981<\/h3>\n\n\n\n<p>\u672cGuideBook\u306f\u3001\u30cf\u30eb\u30d3\u30f3\u5de5\u696d\u5927\u5b66\uff08\u6df1\u5733\uff09\u300c\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u300d\u306eLab\u6307\u5c0e\u8cc7\u6599\u3067\u3059\u3002PC\u3067\u30da\u30fc\u30b8\u5de6\u5074\u306b\u5404\u5c0f\u7bc0\u306e\u76ee\u6b21\u3001\u7d22\u5f15\u304c\u8868\u793a\u3055\u308c\u307e\u3059\u3002<br><strong>\u5fc5\u305a\u9806\u756a\u306b\u6307\u5c0e\u66f8\u3092\u8aad\u307f\u9032\u3081\u3001\u7591\u554f\u70b9\u304c\u3042\u308c\u3070\u7a4d\u6975\u7684\u306b\u8cea\u554f\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/strong><\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u6ce8\u610f\u4e8b\u9805 <\/h3>\n\n\n\n<p class=\"is-style-big_icon_batsu\">\u3053\u306e\u30b3\u30fc\u30b9\u8cc7\u6599\u306f\u3001<strong>\u30cf\u30eb\u30d3\u30f3\u5de5\u696d\u5927\u5b66\uff08\u6df1\u5733\uff09\u300c\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\uff08\u5b9f\u9a13\uff09\u300d2024\u5e74\u79cb\u5b66\u671f<\/strong>\u306e\u6388\u696d\u306b\u9650\u308a\u4f7f\u7528\u53ef\u80fd\u3067\u3059\u3002<br>\u8cc7\u6599\u306e\u7121\u65ad\u8ee2\u8f09\u3084\u4ed6\u306e\u7528\u9014\u3067\u306e\u4f7f\u7528\u3092\u56fa\u304f\u7981\u3058\u307e\u3059\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">\u5185\u5bb9\u306e\u69cb\u6210<\/h3>\n\n\n\n<p>\u672c\u30b3\u30fc\u30b9\u5b9f\u9a13\u306e\u76ee\u7684\u306f\u3001\u53d7\u8b1b\u8005\u304c\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u304a\u3051\u308b <strong>\u30ad\u30e3\u30c3\u30b7\u30e5\u6700\u9069\u5316\u3001\u547d\u4ee4\u30ec\u30d9\u30eb\u306e\u6700\u9069\u5316\u3001\u304a\u3088\u3073\u4e26\u5217\u6700\u9069\u5316<\/strong> \u3092\u7fd2\u5f97\u3059\u308b\u3053\u3068\u306b\u3042\u308a\u307e\u3059\u3002\u5177\u4f53\u7684\u306b\u306f\u3001\u4ee5\u4e0b\u306e\u5185\u5bb9\u3092\u542b\u307f\u307e\u3059\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u57fa\u672c\u7684\u306a\u5b9f\u9a13\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306e\u4f5c\u6210<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u30a2\u30bb\u30f3\u30d6\u30ea\u30b3\u30fc\u30c9\u3092\u4f7f\u7528\u3057\u3066\u884c\u5217\u7a4d\u3068Llama\u63a8\u8ad6\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u3092\u5b9f\u88c5\u3002<\/li>\n\n\n\n<li>Llama\u63a8\u8ad6\u30d5\u30ec\u30fc\u30e0\u30ef\u30fc\u30af\u306b\u304a\u3051\u308b\u884c\u5217\u7a4d\uff08\u307e\u305f\u306f\u884c\u5217-\u30d9\u30af\u30c8\u30eb\u7a4d\uff09\u304c\u63a8\u8ad6\u6642\u9593\u306b\u5360\u3081\u308b\u5272\u5408\u3092\u5206\u6790\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u30ad\u30e3\u30c3\u30b7\u30e5\u6700\u9069\u5316\u30a2\u30bb\u30f3\u30d6\u30ea\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u4f5c\u6210<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u30d7\u30ed\u30bb\u30c3\u30b5\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u968e\u5c64\u3068\u30b5\u30a4\u30ba\u3092\u7406\u89e3\u3002<\/li>\n\n\n\n<li>\u884c\u5217\u7a4d\u306b\u304a\u3051\u308b\u30c7\u30fc\u30bf\u306e\u8aad\u307f\u66f8\u304d\u304c\u30ad\u30e3\u30c3\u30b7\u30e5\u306b\u4e0e\u3048\u308b\u5f71\u97ff\u3092\u628a\u63e1\u3002<\/li>\n\n\n\n<li>\u30c7\u30fc\u30bf\u3068\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u7279\u6027\u306b\u57fa\u3065\u304d\u3001\u30c7\u30fc\u30bf\u306e\u8aad\u307f\u66f8\u304d\u65b9\u6cd5\u3092\u6700\u9069\u5316\u3057\u3001\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u5229\u7528\u7387\u3092\u5411\u4e0a\u3055\u305b\u3001\u884c\u5217\u7a4d\u306e\u6027\u80fd\u3092\u6539\u5584\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u5316CPU\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u4f5c\u6210<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u5316\u306b\u3088\u308b\u884c\u5217\u7a4d\u306e\u9ad8\u901f\u5316\u3092\u7406\u89e3\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>CUDA\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u57fa\u672c\u7684\u306aGPU\u884c\u5217\u7a4d\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u4f5c\u6210\u3002<\/li>\n\n\n\n<li>\u5171\u6709\u30e1\u30e2\u30ea\u3092\u4f7f\u7528\u3057\u305f\u6700\u9069\u5316\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>Llama.c\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u9ad8\u901f\u5316<\/strong>\n<ul class=\"wp-block-list\">\n<li>\u81ea\u4e3b\u63a2\u7d22\u30c6\u30fc\u30de\u3068\u3057\u3066\u63d0\u4f9b<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">Labs\u4e00\u89a7<\/h3>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u5185\u5bb9\u30926\u3064\u306eLabs\u306b\u5206\u5272\u3057\u3066\u9032\u884c\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-table is-all-centered\"><table class=\"has-fixed-layout\"><thead><tr><th><strong>\u30c6\u30fc\u30de<\/strong><\/th><th><strong>Lab<\/strong><\/th><th><strong>\u5b9f\u9a13\u6642\u9593<\/strong><\/th><\/tr><\/thead><tbody><tr><td rowspan=\"4\">CPU\u6700\u9069\u5316<\/td><td>\u5b9f\u9a131\uff1a\u884c\u5217\u7a4d\u306e\u4f5c\u6210<\/td><td>4\u6642\u9593<\/td><\/tr><tr><td>\u5b9f\u9a132\uff1a\u30ad\u30e3\u30c3\u30b7\u30e5\u3092\u6700\u9069\u5316\u3057\u305f\u884c\u5217\u7a4d\u306e\u4f5c\u6210<\/td><td>4\u6642\u9593<\/td><\/tr><tr><td>\u5b9f\u9a133\uff1a\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u5316\u3001\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u3001\u4e26\u5217\u6700\u9069\u5316\u3092\u4f7f\u7528\u3057\u305f\u884c\u5217\u7a4d\u4f5c\u6210<\/td><td>8\u6642\u9593<\/td><\/tr><tr><td colspan=\"2\">\ud83d\udcdd \u5b9f\u9a13\u30ec\u30dd\u30fc\u30c8\u4f5c\u6210<\/td><\/tr><tr><td rowspan=\"3\">GPU\u6700\u9069\u5316<\/td><td>\u5b9f\u9a134\uff1aC\/CUDA\u3092\u4f7f\u7528\u3057\u305fGPU\u30b3\u30fc\u30c9\u4f5c\u6210<\/td><td>4\u6642\u9593<\/td><\/tr><tr><td>\u5b9f\u9a135\uff1a\u5171\u6709\u30e1\u30e2\u30ea\u3092\u7528\u3044\u305fGPU\u30b3\u30fc\u30c9\u306e\u6700\u9069\u5316<\/td><td>4\u6642\u9593<\/td><\/tr><tr><td colspan=\"2\">\ud83d\udcdd \u5b9f\u9a13\u30ec\u30dd\u30fc\u30c8\u4f5c\u6210<\/td><\/tr><tr><td rowspan=\"2\">Llama\u6700\u9069\u5316<\/td><td>\u5b9f\u9a136\uff1aLlama\u30e2\u30c7\u30eb\u63a8\u8ad6\u6027\u80fd\u306e\u8a55\u4fa1<\/td><td>4\u6642\u9593<\/td><\/tr><tr><td colspan=\"2\">\ud83d\udcdd \u5b9f\u9a13\u30ec\u30dd\u30fc\u30c8\u4f5c\u6210<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<h3 class=\"wp-block-heading\">\u6388\u696d\u30b9\u30b1\u30b8\u30e5\u30fc\u30eb<\/h3>\n\n\n\n<p>\u6388\u696d\u306f\u3001\u4ee5\u4e0b\u306e3\u3064\u306e\u30af\u30e9\u30b9\u306b\u5206\u304b\u308c\u3066\u884c\u308f\u308c\u307e\u3059\u3002\u6388\u696d\u306f\u3059\u3079\u3066<strong>T2210<\/strong>\u3067\u5b9f\u65bd\u3055\u308c\u307e\u3059\u3002<br>\u7279\u5b9a\u306e\u9031\u9650\u3067\u4ed6\u306e\u6388\u696d\u3084\u8a66\u9a13\u306a\u3069\u304c\u91cd\u8907\u3059\u308b\u5834\u5408\u306f\u3001\u4ee5\u4e0b\u306e\u30b9\u30b1\u30b8\u30e5\u30fc\u30eb\u3092\u53c2\u7167\u3057\u3066\u3054\u81ea\u7531\u306b\u4ed6\u306e\u30af\u30e9\u30b9\u3067\u884c\u3046\u3053\u3068\u304c\u53ef\u80fd\u3067\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-table is-all-centered--ta\"><table class=\"has-fixed-layout\"><thead><tr><th><strong>\u9031\u6b21<\/strong><\/th><th><strong>\u6388\u696d\u5185\u5bb9<\/strong><\/th><th><strong>A\u30af\u30e9\u30b9<\/strong><\/th><th><strong>B\u30af\u30e9\u30b9<\/strong><\/th><th><strong>C\u30af\u30e9\u30b9<\/strong><\/th><\/tr><\/thead><tbody><tr><td>11\u9031\u76ee<\/td><td>\u5b9f\u9a131<\/td><td>\u6c34\u66dc 9-12\u9650<\/td><td>\u6728\u66dc 9-12\u9650<\/td><td>\u571f\u66dc 5-8\u9650<\/td><\/tr><tr><td>12\u9031\u76ee<\/td><td>\u5b9f\u9a132<\/td><td>\u6c34\u66dc 9-12\u9650<\/td><td>\u6728\u66dc 9-12\u9650<\/td><td>\u91d1\u66dc 9-12\u9650<\/td><\/tr><tr><td>13\u9031\u76ee<\/td><td>\u5b9f\u9a133-1<\/td><td>\u6c34\u66dc 9-12\u9650<\/td><td>\u91d1\u66dc 5-8\u9650<\/td><td>\u571f\u66dc 5-8\u9650<\/td><\/tr><tr><td>14\u9031\u76ee<\/td><td>\u5b9f\u9a133-2<\/td><td>\u6728\u66dc 1-4\u9650<\/td><td>\u6728\u66dc 9-12\u9650<\/td><td>\u91d1\u66dc 9-12\u9650<\/td><\/tr><tr><td>15\u9031\u76ee<\/td><td>\u5b9f\u9a134<\/td><td>\u6728\u66dc 1-4\u9650<\/td><td>\u6728\u66dc 9-12\u9650<\/td><td>\u91d1\u66dc 9-12\u9650<\/td><\/tr><tr><td>16\u9031\u76ee-1<\/td><td>\u5b9f\u9a135<\/td><td>\u6708\u66dc 1-4\u9650<\/td><td>\u6728\u66dc 1-4\u9650<\/td><td>\u6708\u66dc 5-8\u9650<\/td><\/tr><tr><td>16\u9031\u76ee-2<\/td><td>\u5b9f\u9a136<\/td><td>\u6c34\u66dc 1-4\u9650<\/td><td>\u91d1\u66dc 9-12\u9650<\/td><td>\u6728\u66dc 9-12\u9650<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<h3 class=\"wp-block-heading\">\u305d\u306e\u4ed6<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\">1. \u5b9f\u9a13\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9<\/h4>\n\n\n\n<p><strong>\u30ad\u30e3\u30f3\u30d1\u30b9\u30cd\u30c3\u30c8<\/strong>\u3092\u5229\u7528\u3057\u305f<a href=\"http:\/\/10.249.10.96:3011\/\">\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u30ea\u30f3\u30af<\/a>\u304c\u63d0\u4f9b\u3055\u308c\u307e\u3059\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">2. Lab\u306e\u63d0\u51fa\u65b9\u6cd5<\/h4>\n\n\n\n<p><strong>1. \u30ed\u30b0\u30a4\u30f3<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u751f\u5f92\u7528\u30ed\u30b0\u30a4\u30f3URL<\/strong>\uff1a<a href=\"http:\/\/labgrader.hitsz.edu.cn:8000\">http:\/\/labgrader.hitsz.edu.cn:8000<\/a><\/li>\n\n\n\n<li>\u63a8\u5968\u30d6\u30e9\u30a6\u30b6\u306f<strong>Chrome<\/strong>\u3067\u3059\u3002\u521d\u671f\u306e\u30e6\u30fc\u30b6\u30fc\u540d\u3068\u30d1\u30b9\u30ef\u30fc\u30c9\u306f\u3069\u3061\u3089\u3082\u5b66\u7c4d\u756a\u53f7\u3067\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u521d\u56de\u30ed\u30b0\u30a4\u30f3\u5f8c\u306f\u3001<strong>\u30d1\u30b9\u30ef\u30fc\u30c9\u306e\u5909\u66f4<\/strong>\u3092\u304a\u3059\u3059\u3081\u3057\u307e\u3059\u3002\u624b\u9806\u306f\u4e0b\u306e\u300c5. \u30d1\u30b9\u30ef\u30fc\u30c9\u5909\u66f4\u300d\u3092\u53c2\u7167\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>2. \u30b3\u30fc\u30b9\u306e\u78ba\u8a8d<\/strong><\/p>\n\n\n\n<p>\u30ed\u30b0\u30a4\u30f3\u5f8c\u3001\u81ea\u5206\u304c\u767b\u9332\u3057\u3066\u3044\u308b\u30b3\u30fc\u30b9\u3092\u78ba\u8a8d\u3067\u304d\u307e\u3059\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u30b3\u30fc\u30b9\u540d\u3092\u30af\u30ea\u30c3\u30af\u3059\u308b\u3068\u3001\u30b3\u30fc\u30b9\u306e\u8a73\u7d30\u304c\u8868\u793a\u3055\u308c\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u63d0\u51fa\u671f\u9650<\/strong>\u306b\u7279\u306b\u6ce8\u610f\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>3. \u6210\u679c\u306e\u63d0\u51fa<\/strong><\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u63d0\u51fa\u958b\u59cb<\/strong>\u3092\u30af\u30ea\u30c3\u30af\u3057\u3066\u8ab2\u984c\u306e\u89e3\u7b54\u3084\u30d5\u30a1\u30a4\u30eb\u306e\u63d0\u51fa\u3092\u884c\u3044\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>File input<\/strong>\u3092\u30af\u30ea\u30c3\u30af\u3057\u3001\u63d0\u51fa\u3059\u308b\u30d5\u30a1\u30a4\u30eb\u3092\u9078\u629e\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u30d5\u30a1\u30a4\u30eb\u3092\u9078\u629e\u3057\u305f\u3089\u3001\u753b\u9762\u53f3\u4e0b\u306e<strong>\u63d0\u51fa<\/strong>\u30dc\u30bf\u30f3\u3092\u30af\u30ea\u30c3\u30af\u3057\u3066\u63d0\u51fa\u3092\u5b8c\u4e86\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30da\u30fc\u30b8\u304c\u9077\u79fb<\/strong>\u3057\u305f\u3089\u3001\u63d0\u51fa\u6210\u529f\u3092\u793a\u3059\u306e\u5834\u5408\u3001\u63d0\u51fa\u6210\u529f\u3067\u3059\u3002<br>\u203b\u63d0\u51fa\u4e2d\u306b\u30d6\u30e9\u30a6\u30b6\u3092\u9589\u3058\u308b\u3068\u5931\u6557\u3059\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u306e\u3067\u6ce8\u610f\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/li>\n<\/ol>\n\n\n\n<p><strong>4. \u518d\u63d0\u51fa<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u63d0\u51fa\u5f8c\u3067\u3082<strong>\u63d0\u51fa\u671f\u9650\u5185<\/strong>\u3067\u3042\u308c\u3070\u3001\u518d\u63d0\u51fa\u304c\u53ef\u80fd\u3067\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>5. \u30d1\u30b9\u30ef\u30fc\u30c9\u306e\u5909\u66f4<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u753b\u9762\u53f3\u4e0a\u306e\u30a2\u30ab\u30a6\u30f3\u30c8\u60c5\u5831\u3092\u30af\u30ea\u30c3\u30af\u3059\u308b\u3068\u3001\u30d1\u30b9\u30ef\u30fc\u30c9\u5909\u66f4\u753b\u9762\u306b\u79fb\u52d5\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>6. \u6ce8\u610f\u4e8b\u9805<\/strong><\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u63d0\u51fa\u671f\u9650\u5185<\/strong>\u3067\u3042\u308c\u3070\u3001\u518d\u63d0\u51fa\u304c\u53ef\u80fd\u3067\u3059\uff08\u63d0\u51fa\u56de\u6570\u306e\u5236\u9650\u306f\u3042\u308a\u307e\u305b\u3093\uff09\u3002<\/li>\n\n\n\n<li>\u63d0\u51fa\u671f\u9650\u5f8c\u306f<strong>\u63d0\u51fa\u4e0d\u53ef<\/strong>\u3068\u306a\u308a\u307e\u3059\u306e\u3067\u6ce8\u610f\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/li>\n\n\n\n<li>\u63d0\u51fa\u6642\u3001<strong>\u30d5\u30a1\u30a4\u30eb\u306e\u62e1\u5f35\u5b50\u306f\u5404\u63d0\u51fa\u90e8\u5206\u3067\u6307\u5b9a\u3055\u308c\u305f\u5834\u5408\u306f\u305d\u306e\u30d5\u30a1\u30a4\u30eb<strong>\u62e1\u5f35\u5b50<\/strong>\u3092\u6301\u3064\u30d5\u30a1\u30a4\u30eb\u3092\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3057\u306a\u3055\u3044<\/strong>\u3002<\/li>\n\n\n\n<li><strong>256MB\u3092\u8d85\u3048\u308b\u30d5\u30a1\u30a4\u30eb<\/strong>\u306e\u63d0\u51fa\u306f\u73fe\u5728\u5bfe\u5fdc\u3057\u3066\u3044\u307e\u305b\u3093\u3002<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">3. \u5b9f\u9a13\u5ba4\u306e\u30cd\u30c3\u30c8\u30ef\u30fc\u30af\u306b\u3064\u3044\u3066<\/h4>\n\n\n\n<p>T2210\u306eUbuntu\u304c\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u306b\u63a5\u7d9a\u3067\u304d\u306a\u3044\u5834\u5408\u306f\u3001\u4ee5\u4e0b\u306e\u8a2d\u5b9a\u3092\u884c\u3063\u3066\u304f\u3060\u3055\u3044\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-table aligncenter\"><table class=\"has-fixed-layout\"><thead><tr><th><strong>\u9805\u76ee<\/strong><\/th><th><strong>\u8a2d\u5b9a\u5024<\/strong><\/th><\/tr><\/thead><tbody><tr><td>IP<\/td><td>10.251.137.[\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u756a\u53f7]<\/td><\/tr><tr><td>\u30b5\u30d6\u30cd\u30c3\u30c8\u30de\u30b9\u30af<\/td><td>255.255.255.0<\/td><\/tr><tr><td>\u30b2\u30fc\u30c8\u30a6\u30a7\u30a4<\/td><td>10.251.137.254<\/td><\/tr><tr><td>DNS<\/td><td>10.248.98.30<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\">Lab1 \u884c\u5217\u4e57\u7b97<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u76ee\u7684<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u306e\u547d\u4ee4\u30bb\u30c3\u30c8\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3084\u30a2\u30c9\u30ec\u30b9\u30e2\u30fc\u30c9\u306e\u7406\u89e3\u3002<\/li>\n\n\n\n<li>\u884c\u5217\u7a4d\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u5b9f\u88c5\u3059\u308b\u3053\u3068\u3067\u3001x86-64\u30a2\u30bb\u30f3\u30d6\u30ea\u8a00\u8a9e\u306e\u3055\u307e\u3056\u307e\u306a\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u30e2\u30fc\u30c9\u3092\u5b66\u3073\u5fdc\u7528\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u30cf\u30fc\u30c9\u30a6\u30a7\u30a2\u30ec\u30d9\u30eb\u3067\u30e1\u30e2\u30ea\u306b\u30a2\u30af\u30bb\u30b9\u3057\u64cd\u4f5c\u3059\u308b\u4ed5\u7d44\u307f\u306e\u7406\u89e3\u3002<\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u524d\u306e\u6e96\u5099<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u30a2\u30bb\u30f3\u30d6\u30ea\u8a00\u8a9e\u306e\u57fa\u672c\u6982\u5ff5\u3092\u89e3\u660e\u3057\u3001\u7279\u306b\u4ee5\u4e0b\u306e\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u30e2\u30fc\u30c9\u3092\u91cd\u70b9\u7684\u306b\u7406\u89e3\u3059\u308b\uff1a\n<ul class=\"wp-block-list\">\n<li>\u30ec\u30b8\u30b9\u30bf\u9593\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/li>\n\n\n\n<li>\u30d9\u30fc\u30b9+\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/li>\n\n\n\n<li>\u5373\u5024\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/li>\n\n\n\n<li>\u6bd4\u4f8b\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>Lab\u74b0\u5883<\/strong>\uff1ax86\u30a2\u30bb\u30f3\u30d6\u30e9\u3068\u30c7\u30d0\u30c3\u30ac\u304c\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3055\u308c\u305fLinux\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u3002<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u539f\u7406<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\">\u5206\u5272\u884c\u5217\u7a4d\u3068\u4e57\u7b97\u8d70\u67fb\u6cd5<\/h4>\n\n\n\n<p>\u884c\u5217$A$\u304c$M \\times K$\u306e\u30b5\u30a4\u30ba\u3001\u884c\u5217$B$\u304c$K \\times N$\u306e\u30b5\u30a4\u30ba\u3067\u3042\u308b\u3068\u3059\u308b\u3002\u3053\u306e\u3068\u304d\u3001\u884c\u5217$A$\u3068\u884c\u5217$B$\u306e\u7a4d$C$\u306f$M \\times N$\u306e\u30b5\u30a4\u30ba\u3068\u306a\u308a\u3001\u884c\u5217$C$\u306e\u5404\u8981\u7d20\u306f\u6b21\u306e\u5f0f\u3067\u8a08\u7b97\u3055\u308c\u308b\uff1a$$C_{ij} = \\sum_{k=1}^{K} A_{ik} \\cdot B_{kj}, \\quad i=1,2,\\dots,M; \\, j=1,2,\\dots,N$$<\/p>\n\n\n\n<p>\u884c\u5217$A$\u3001\u884c\u5217$B$\u3001\u304a\u3088\u3073\u884c\u5217$C$\u3092\u5206\u5272\u884c\u5217\u7a4d\u306e\u5b9a\u7fa9\u306b\u57fa\u3065\u304d\u56f32-1\u306e\u3088\u3046\u306b\u5206\u5272\u3059\u308b\u3002\u5b9a\u7fa9\u3088\u308a\u3001\u6b21\u306e\u3088\u3046\u306b\u8a08\u7b97\u3055\u308c\u308b\uff1a$$C_{00} = A_{00} \\cdot B_{00} + A_{01} \\cdot B_{10} + A_{02} \\cdot B_{20}$$<\/p>\n\n\n\n<p>\u3053\u306e$C_{00}$\u3092\u8a08\u7b97\u3059\u308b\u306b\u306f\u3001\u884c\u5217$A$\u3068\u884c\u5217$B$\u306e\u3059\u3079\u3066\u306e\u5206\u5272\u90e8\u5206\u306b\u3064\u3044\u3066\u3001$K$\u6b21\u5143\u306b\u6cbf\u3063\u305f\u8a08\u7b97\u3092\u884c\u3046\u5fc5\u8981\u304c\u3042\u308b\u3002<\/p>\n\n\n\n<p>\u4e00\u822c\u7684\u306b\u3001\u7d50\u679c\u884c\u5217\u306e2\u6b21\u5143\u3092\u5148\u306b\u8d70\u67fb\u3057\u3001\u6b8b\u308a\u306e1\u6b21\u5143\u3092\u6700\u5f8c\u306b\u8d70\u67fb\u3059\u308b\u65b9\u6cd5\u304c\u3088\u304f\u7528\u3044\u3089\u308c\u308b\u3002\u305f\u3068\u3048\u3070\u3001\u4e0a\u8a18\u306e\u884c\u5217$C$\u306e\u5404\u8981\u7d20\u3092\u8a08\u7b97\u3059\u308b\u5f0f\u306f\u3001\u5178\u578b\u7684\u306a <strong>ijk\u8d70\u67fb\u65b9\u5f0f<\/strong> \u3067\u3042\u308b\u3002\u3053\u306e\u65b9\u5f0f\u3067\u306f\u3001\u6700\u5f8c\u306e\u6b21\u5143\u3092\u8d70\u67fb\u3057\u7d42\u3048\u305f\u6642\u70b9\u3067\u3001\u884c\u5217$C$\u306e1\u3064\u306e\u8981\u7d20\u306e\u8a08\u7b97\u304c\u5b8c\u4e86\u3059\u308b\u3068\u3044\u3046\u5229\u70b9\u304c\u3042\u308b\u3002<\/p>\n\n\n\n<p>\u4e00\u65b9\u3001<strong>kij\u8d70\u67fb\u65b9\u5f0f<\/strong>\uff08\u307e\u305f\u306f\u30e9\u30f3\u30afK\u66f4\u65b0\uff09\u3082\u3088\u304f\u4f7f\u7528\u3055\u308c\u308b\u3002ijk\u8d70\u67fb\u65b9\u5f0f\u3068\u306f\u7570\u306a\u308a\u3001kij\u8d70\u67fb\u65b9\u5f0f\u3067\u306f\u884c\u5217$C$\u306e\u4e2d\u9593\u7d50\u679c\u3092\u4fdd\u6301\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u3002\u56f31-1\u306b\u793a\u3055\u308c\u308b\u3088\u3046\u306b\u3001\u884c\u5217$A$\u306e1\u5217\u3068\u884c\u5217$B$\u306e1\u884c\u3092\u8d70\u67fb\u3059\u308b\u3060\u3051\u3067\u306f\u3001\u884c\u5217$C$\u306e\u5404\u8981\u7d20\u306e\u4e00\u90e8\u306e\u7d50\u679c\u3057\u304b\u5f97\u3089\u308c\u306a\u3044\u3002\u884c\u5217$A$\u3068\u884c\u5217$B$\u306e\u3059\u3079\u3066\u306e\u5206\u5272\u90e8\u5206\u3092\u8d70\u67fb\u3057\u7d42\u3048\u305f\u6642\u70b9\u3067\u3001\u3088\u3046\u3084\u304f\u884c\u5217$C$\u306e\u6700\u7d42\u7d50\u679c\u304c\u5f97\u3089\u308c\u308b\u3002kij\u8d70\u67fb\u65b9\u5f0f\u306f\u3001\u7a7a\u9593\u30a2\u30af\u30bb\u30b9\u306e\u5c40\u6240\u6027\u3092\u3088\u308a\u78ba\u4fdd\u3057\u3001\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u5229\u7528\u52b9\u7387\u3092\u5411\u4e0a\u3055\u305b\u308b\u70b9\u3067\u512a\u308c\u3066\u3044\u308b\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized is-style-border\"><img decoding=\"async\" width=\"663\" height=\"732\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-1.png\" alt=\"\" class=\"wp-image-1734\" style=\"width:299px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-1.png 663w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-1-272x300.png 272w\" sizes=\"(max-width: 663px) 100vw, 663px\" \/><figcaption class=\"wp-element-caption\">\u56f31-1<\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">x86-64\u6982\u8981<\/h4>\n\n\n\n<p>x86-64\u306fx86\u306e64\u30d3\u30c3\u30c8\u62e1\u5f35\u7248\u3067\u3042\u308a\u3001\u9ad8\u3044\u4e92\u63db\u6027\u3092\u4fdd\u3061\u306a\u304c\u3089\u3001\u30ec\u30b8\u30b9\u30bf\u3068\u4eee\u60f3\u30a2\u30c9\u30ec\u30b9\u7a7a\u9593\u309264\u30d3\u30c3\u30c8\u306b\u62e1\u5f35\u3057\u3001\u547d\u4ee4\u30bb\u30c3\u30c8\u306e\u6a5f\u80fd\u3068\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u3066\u3044\u308b\u3002<\/p>\n\n\n\n<p>x86-64\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306f\u3001\u5236\u5fa1\u3068\u72b6\u614b\u30ec\u30b8\u30b9\u30bf\u3001\u30d7\u30ed\u30b0\u30e9\u30e0\u30dd\u30a4\u30f3\u30bf\u30ec\u30b8\u30b9\u30bf\u3001\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u3001\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf\u306a\u3069\u3001\u3055\u307e\u3056\u307e\u306a\u30bf\u30a4\u30d7\u306e\u30ec\u30b8\u30b9\u30bf\u3092\u5099\u3048\u3066\u3044\u308b\uff08\u88681-1\u53c2\u7167\uff09\u3002<\/p>\n\n\n\n<figure class=\"wp-block-table is-all-centered is-style-regular has-small-font-size\"><table class=\"has-swl-pale-04-background-color has-background\"><thead><tr><th>\u7a2e\u985e<\/th><th>\u540d\u524d<\/th><th>\u30d3\u30c3\u30c8\u5e45<\/th><th>\u8aac\u660e<\/th><\/tr><\/thead><tbody><tr><td>\u5236\u5fa1\u3068\u72b6\u614b\u30ec\u30b8\u30b9\u30bf<\/td><td>RFLAGS<\/td><td>64<\/td><td>\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u30d5\u30e9\u30b0\u60c5\u5831\uff08\u4f8b\uff1a\u30bc\u30ed\u30d5\u30e9\u30b0\uff08ZF\uff09\u3001\u7b26\u53f7\u30d3\u30c3\u30c8\uff08SF\uff09\u3001\u30ad\u30e3\u30ea\u30fc\u30d3\u30c3\u30c8\uff08CF\uff09\u3001\u30aa\u30fc\u30d0\u30fc\u30d5\u30ed\u30fc\u30d3\u30c3\u30c8\uff08OF\uff09\u3001\u5272\u308a\u8fbc\u307f\u8a31\u53ef\u30d3\u30c3\u30c8\uff08IF\uff09\u306a\u3069\uff09<\/td><\/tr><tr><td>\u30dd\u30a4\u30f3\u30bf\u30ec\u30b8\u30b9\u30bf<\/td><td>RIP<\/td><td>64<\/td><td>\u73fe\u5728\u306e\u547d\u4ee4\u306e\u30a2\u30c9\u30ec\u30b9<\/td><\/tr><tr><td rowspan=\"9\">\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf<\/td><td>RAX<\/td><td>64<\/td><td>\u7d2f\u7a4d\u7528\u3001\u7a4d\u3084\u88ab\u9664\u6570\u306e\u4e0b\u4f4d64\u30d3\u30c3\u30c8\u3001\u5546\u306e\u683c\u7d0d\u3001\u307e\u305f\u306f\u30b5\u30d6\u30eb\u30fc\u30c1\u30f3\u306e\u623b\u308a\u5024\u306e\u683c\u7d0d\u306b\u4f7f\u7528<\/td><\/tr><tr><td>RBX<\/td><td>64<\/td><td>\u30d9\u30fc\u30b9\u30ec\u30b8\u30b9\u30bf\u3068\u3057\u3066\u4f7f\u7528<\/td><\/tr><tr><td>RCX<\/td><td>64<\/td><td>\u30ab\u30a6\u30f3\u30bf\u3001\u307e\u305f\u306f\u30b5\u30d6\u30eb\u30fc\u30c1\u30f3\u547c\u3073\u51fa\u3057\u6642\u306e\u7b2c4\u5f15\u6570\u3092\u683c\u7d0d<\/td><\/tr><tr><td>RDX<\/td><td>64<\/td><td>\u30c7\u30fc\u30bf\u30ec\u30b8\u30b9\u30bf\u3001\u7a4d\u3084\u88ab\u9664\u6570\u306e\u4e0a\u4f4d64\u30d3\u30c3\u30c8\u3001\u4f59\u308a\u306e\u683c\u7d0d\u3001\u307e\u305f\u306f\u30b5\u30d6\u30eb\u30fc\u30c1\u30f3\u547c\u3073\u51fa\u3057\u6642\u306e\u7b2c3\u5f15\u6570\u3092\u683c\u7d0d<\/td><\/tr><tr><td>RSI<\/td><td>64<\/td><td>\u6587\u5b57\u5217\u64cd\u4f5c\u306e\u30bd\u30fc\u30b9\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u3001\u307e\u305f\u306f\u30b5\u30d6\u30eb\u30fc\u30c1\u30f3\u547c\u3073\u51fa\u3057\u6642\u306e\u7b2c2\u5f15\u6570<\/td><\/tr><tr><td>RDI<\/td><td>64<\/td><td>\u6587\u5b57\u5217\u64cd\u4f5c\u306e\u30c7\u30b9\u30c6\u30a3\u30cd\u30fc\u30b7\u30e7\u30f3\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u3001\u307e\u305f\u306f\u30b5\u30d6\u30eb\u30fc\u30c1\u30f3\u547c\u3073\u51fa\u3057\u6642\u306e\u7b2c1\u5f15\u6570<\/td><\/tr><tr><td>RSP<\/td><td>64<\/td><td>\u30b9\u30bf\u30c3\u30af\u30dd\u30a4\u30f3\u30bf<\/td><\/tr><tr><td>RBP<\/td><td>64<\/td><td>\u30b9\u30bf\u30c3\u30af\u306e\u57fa\u6e96\u30dd\u30a4\u30f3\u30bf<\/td><\/tr><tr><td>R8\uff5eR15<\/td><td>64<\/td><td>\u6c4e\u7528\uff08R8\u3001R9\u306f\u30b5\u30d6\u30eb\u30fc\u30c1\u30f3\u547c\u3073\u51fa\u3057\u6642\u306e\u7b2c5\u3001\u7b2c6\u5f15\u6570\u3068\u3057\u3066\u4f7f\u7528\uff09<\/td><\/tr><tr><td rowspan=\"2\">\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf<\/td><td>XMM0\uff5eXMM15<\/td><td>128<\/td><td>SSE\uff08\u30b9\u30c8\u30ea\u30fc\u30df\u30f3\u30b0SIMD\u62e1\u5f35\uff09\u30ec\u30b8\u30b9\u30bf\uff1bYMM\u306e\u4e0b\u4f4d128\u30d3\u30c3\u30c8<\/td><\/tr><tr><td>YMM0\uff5eYMM15<\/td><td>256<\/td><td>AVX\uff08\u30a2\u30c9\u30d0\u30f3\u30b9\u30c9\u30d9\u30af\u30c8\u30eb\u62e1\u5f35\uff09\u30ec\u30b8\u30b9\u30bf<\/td><\/tr><\/tbody><\/table><figcaption class=\"wp-element-caption\"><strong>\u88681-1\uff1ax86-64\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306e\u4e00\u90e8\u30ec\u30b8\u30b9\u30bf<\/strong><\/figcaption><\/figure>\n\n\n\n<h6 class=\"wp-block-heading\"><\/h6>\n\n\n\n<p>x86-64\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306e\u4e00\u822c\u7684\u306a\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u30e2\u30fc\u30c9\u306b\u306f\u3001\u5373\u5024\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3001\u30ec\u30b8\u30b9\u30bf\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3001\u76f4\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3001\u30ec\u30b8\u30b9\u30bf\u9593\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3001\u30d9\u30fc\u30b9\u5909\u6570\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3001\u304a\u3088\u3073\u6bd4\u4f8b\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u304c\u542b\u307e\u308c\u308b\uff08\u88681-2\u53c2\u7167\uff09\u3002<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u30e2\u30fc\u30c9<\/th><th>\u4f7f\u7528\u4f8b<\/th><th>\u8aac\u660e<\/th><\/tr><\/thead><tbody><tr><td>\u5373\u5024\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/td><td><code>mov $0x1234, %rax<\/code><\/td><td>RAX \u2190 0x1234<\/td><\/tr><tr><td>\u30ec\u30b8\u30b9\u30bf\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/td><td><code>mov %r8, %rax<\/code><\/td><td>RAX \u2190 R8<\/td><\/tr><tr><td>\u76f4\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/td><td><code>mov 0x1234, %rax<\/code><\/td><td>RAX \u2190 MEM[0x1234]<\/td><\/tr><tr><td>\u30ec\u30b8\u30b9\u30bf\u9593\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/td><td><code>mov (%rsp), %rax<\/code><\/td><td>RAX \u2190 MEM[RSP]<\/td><\/tr><tr><td>\u30d9\u30fc\u30b9\u5909\u6570\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/td><td><code>mov -20(%rsp), %rax<\/code><\/td><td>RAX \u2190 MEM[RSP &#8211; 20]<\/td><\/tr><tr><td>\u6bd4\u4f8b\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0<\/td><td><code>mov -20(%rsp, %rcx, 4), %rax<\/code><\/td><td>RAX \u2190 MEM[RSP &#8211; 20 + RCX*4]<\/td><\/tr><\/tbody><\/table><figcaption class=\"wp-element-caption\"><strong>\u88681-2\uff1ax86-64\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306e\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u30e2\u30fc\u30c9<\/strong><\/figcaption><\/figure>\n\n\n\n<p>\u88681-2\u304b\u3089\u308f\u304b\u308b\u3088\u3046\u306b\u3001\u76f4\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3001\u30ec\u30b8\u30b9\u30bf\u9593\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3001\u304a\u3088\u3073\u30d9\u30fc\u30b9\u5909\u6570\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u306f\u3001\u3044\u305a\u308c\u3082\u6bd4\u4f8b\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u306e\u7279\u6b8a\u306a\u30b1\u30fc\u30b9\u3068\u898b\u306a\u305b\u308b\u3002<\/p>\n\n\n\n<div data-wp-interactive=\"core\/file\" class=\"wp-block-file\"><object data-wp-bind--hidden=\"!state.hasPdfPreview\" hidden class=\"wp-block-file__embed\" data=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-pdf-1-1.pdf\" type=\"application\/pdf\" style=\"width:100%;height:600px\" aria-label=\"PDF \u57cb\u3081\u8fbc\u307f\"><\/object><\/div>\n\n\n\n<h4 class=\"wp-block-heading\">x87 FPU<\/h4>\n\n\n\n<p>x87 FPU\uff08\u6d6e\u52d5\u5c0f\u6570\u70b9\u6f14\u7b97\u30e6\u30cb\u30c3\u30c8\uff09\u306f\u3001\u6b21\u306e\u3088\u3046\u306a\u30ec\u30b8\u30b9\u30bf\u3092\u5099\u3048\u3066\u3044\u308b\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>8\u500b\u306e80\u30d3\u30c3\u30c8\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\uff08R0\uff5eR7\uff09<\/li>\n\n\n\n<li>3\u500b\u306e16\u30d3\u30c3\u30c8\u72b6\u614b\u304a\u3088\u3073\u5236\u5fa1\u30ec\u30b8\u30b9\u30bf<\/li>\n\n\n\n<li>2\u500b\u306e48\u30d3\u30c3\u30c8\u30dd\u30a4\u30f3\u30bf\u30ec\u30b8\u30b9\u30bf<\/li>\n\n\n\n<li>1\u500b\u306e11\u30d3\u30c3\u30c8\u30aa\u30da\u30b3\u30fc\u30c9\u30ec\u30b8\u30b9\u30bf<\/li>\n<\/ul>\n\n\n\n<p>\u56f31-2\u306b\u3053\u308c\u3092\u793a\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized is-style-border\"><img decoding=\"async\" width=\"713\" height=\"561\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-2.png\" alt=\"\" class=\"wp-image-1738\" style=\"width:315px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-2.png 713w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-2-300x236.png 300w\" sizes=\"(max-width: 713px) 100vw, 713px\" \/><figcaption class=\"wp-element-caption\">\u56f31-2<\/figcaption><\/figure>\n\n\n\n<p>\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u306f\u3001\u8a08\u7b97\u30c7\u30fc\u30bf\u306e\u683c\u7d0d\u306b\u4f7f\u7528\u3055\u308c\u307e\u3059\u3002\u30c7\u30fc\u30bf\u304c\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u306b\u66f8\u304d\u8fbc\u307e\u308c\u308b\u3068\u3001\u81ea\u52d5\u7684\u306b80\u30d3\u30c3\u30c8\u306e<strong>\u62e1\u5f35\u500d\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\uff08Double Extended-Precision Floating-Point\uff09<\/strong>\u306b\u5909\u63db\u3055\u308c\u307e\u3059\u3002<br>\u4e00\u65b9\u3001\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u304b\u3089\u30e1\u30e2\u30ea\u306b\u30c7\u30fc\u30bf\u3092\u66f8\u304d\u8fbc\u3080\u969b\u306f\u3001\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\u306e\u3055\u307e\u3056\u307e\u306a\u7cbe\u5ea6\u3001\u6574\u6570\u3001\u307e\u305f\u306fBCD\u30b3\u30fc\u30c9\u3068\u3044\u3063\u305f\u5f62\u5f0f\u306e\u4e2d\u304b\u30891\u3064\u3092\u67d4\u8edf\u306b\u9078\u629e\u3057\u3066\u66f8\u304d\u8fbc\u307f\u3092\u884c\u3046\u3053\u3068\u304c\u53ef\u80fd\u3067\u3059\u3002<\/p>\n\n\n\n<p><strong>x87 FPU\u547d\u4ee4<\/strong>\u3067\u306f\u3001\u30b9\u30bf\u30c3\u30af\u5f62\u5f0f\u30678\u500b\u306e\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u306b\u30a2\u30af\u30bb\u30b9\u3067\u304d\u307e\u3059\u3002\u30b9\u30bf\u30c3\u30af\u306e\u6700\u4e0a\u90e8\u306b\u3042\u308b\u30c7\u30fc\u30bf\u306f\u5e38\u306b <code>st(0)<\/code> \u3068\u3057\u3066\u6271\u308f\u308c\u307e\u3059\uff08\u56f31-3\u306e(a) \uff5e (d)\u53c2\u7167\uff09\u3002\u307e\u305f\u3001<code>st(0)<\/code> \u304b\u3089\u306e<strong>\u76f8\u5bfe\u30aa\u30d5\u30bb\u30c3\u30c8<\/strong>\u3067\u4efb\u610f\u306e\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u306b\u30a2\u30af\u30bb\u30b9\u3059\u308b\u3053\u3068\u3082\u53ef\u80fd\u3067\u3059\uff08\u56f31-3\u306e(e)\u53c2\u7167\uff09\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-large is-resized is-style-border\"><img decoding=\"async\" width=\"1024\" height=\"535\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-3-1024x535.png\" alt=\"\" class=\"wp-image-1739\" style=\"width:461px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-3-1024x535.png 1024w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-3-300x157.png 300w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-3-768x401.png 768w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-1-3.png 1062w\" sizes=\"(max-width: 1024px) 100vw, 1024px\" \/><\/figure>\n\n\n\n<p>x87 FPU\u306f\u3001\u4ee5\u4e0b\u306e\u7a2e\u985e\u306e\u547d\u4ee4\u3092\u30b5\u30dd\u30fc\u30c8\u3057\u3066\u3044\u307e\u3059\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u30ed\u30fc\u30c9\u547d\u4ee4<\/strong>\uff1a\u30c7\u30fc\u30bf\u3092\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u30b9\u30bf\u30c3\u30af\u306b\u30d7\u30c3\u30b7\u30e5\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30b9\u30c8\u30a2\u547d\u4ee4<\/strong>\uff1a\u30b9\u30bf\u30c3\u30af\u5185\u306e\u30c7\u30fc\u30bf\u3092\u30dd\u30c3\u30d7\u3057\u3001\u4e3b\u8a18\u61b6\u88c5\u7f6e\u306b\u4fdd\u5b58\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u6f14\u7b97\u547d\u4ee4<\/strong>\uff1a\u6d6e\u52d5\u5c0f\u6570\u70b9\u30c7\u30fc\u30bf\u306e\u52a0\u6e1b\u4e57\u9664\u3001\u7b26\u53f7\u53cd\u8ee2\u306a\u3069\u3092\u5b9f\u73fe\u3057\u307e\u3059\u3002<\/li>\n<\/ol>\n\n\n\n<p>x87 FPU\u547d\u4ee4\u306f\u3001\u4ee5\u4e0b\u306e\u5f62\u5f0f\u3067\u547d\u540d\u3055\u308c\u307e\u3059\uff1a<\/p>\n\n\n\n<p><strong><code>f<\/code> + <code>i<\/code>\uff08\u7701\u7565\u53ef\u80fd\uff09 + \u64cd\u4f5c\u540d + <code>p\/s\/l\/t<\/code>\uff08\u7701\u7565\u53ef\u80fd\uff09<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong><code>f<\/code><\/strong>\uff1a\u547d\u4ee4\u306e\u30d7\u30ec\u30d5\u30a3\u30c3\u30af\u30b9\u3002<\/li>\n\n\n\n<li><strong><code>i<\/code>\uff08\u7701\u7565\u53ef\u80fd\uff09<\/strong>\uff1a\u30aa\u30da\u30e9\u30f3\u30c9\u304c\u6574\u6570\u3067\u3042\u308b\u3053\u3068\u3092\u793a\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong><code>p\/s\/l\/t<\/code>\uff08\u7701\u7565\u53ef\u80fd\uff09<\/strong>\uff1a\u4ee5\u4e0b\u3092\u793a\u3057\u307e\u3059\uff1a\n<ul class=\"wp-block-list\">\n<li><strong><code>s<\/code><\/strong>\uff1a\u5358\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\uff08Single Precision Floating-Point\uff09\u3002<\/li>\n\n\n\n<li><strong><code>l<\/code><\/strong>\uff1a\u500d\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\uff08Double Precision Floating-Point\uff09\u3002<\/li>\n\n\n\n<li><strong><code>t<\/code><\/strong>\uff1a\u62e1\u5f35\u500d\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\uff08Double Extended-Precision Floating-Point\uff09\u3002<\/li>\n\n\n\n<li><strong><code>p<\/code><\/strong>\uff1a\u6f14\u7b97\u5f8c\u306b\u6c4e\u7528\u30ec\u30b8\u30b9\u30bf\u30b9\u30bf\u30c3\u30af\u304b\u3089\u30c7\u30fc\u30bf\u3092\u30dd\u30c3\u30d7\u3059\u308b\u3053\u3068\u3092\u793a\u3057\u307e\u3059\u3002<br><\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>fistpl (%rax)<\/strong>\uff1ast(0)\u306e\u62e1\u5f35\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\u3092\u6574\u6570\u306b\u5909\u63db\u3057\u3066%rax\u304c\u6307\u3059\u30e1\u30e2\u30ea\u306b\u683c\u7d0d\u3057\u3001FPU\u30b9\u30bf\u30c3\u30af\u304b\u3089\u30c7\u30fc\u30bf\u3092\u30dd\u30c3\u30d7\u3059\u308b\u3002<\/li>\n\n\n\n<li><strong>faddp %st(3), %st(0)<\/strong>\uff1ast(3) + st(0)\u306e\u7d50\u679c\u3092st(3)\u306b\u683c\u7d0d\u3057\u3001st(0)\u3092\u30b9\u30bf\u30c3\u30af\u304b\u3089\u30dd\u30c3\u30d7\u3059\u308b\u3002<\/li>\n<\/ul>\n\n\n\n<div data-wp-interactive=\"core\/file\" class=\"wp-block-file\"><object data-wp-bind--hidden=\"!state.hasPdfPreview\" hidden class=\"wp-block-file__embed\" data=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-pdf-1-2.pdf\" type=\"application\/pdf\" style=\"width:100%;height:600px\" aria-label=\"PDF \u57cb\u3081\u8fbc\u307f\"><\/object><\/div>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u5185\u5bb9<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u7c21\u6613\u30a2\u30bb\u30f3\u30d6\u30ea\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u4f5c\u6210\u3057\u3001\u30a2\u30bb\u30f3\u30d6\u30ea\u3092\u4f53\u9a13\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u63d0\u4f9b\u3055\u308c\u305f\u30a2\u30bb\u30f3\u30d6\u30ea\u30b3\u30fc\u30c9\u3092\u7528\u3044\u3066\u884c\u5217\u7a4d\u3092\u5b9f\u88c5\u3057\u3001\u7570\u306a\u308b\u30b5\u30a4\u30ba\u306e\u884c\u5217\u306b\u5bfe\u5fdc\u3059\u308b\u3088\u3046\u306b\u30b3\u30fc\u30c9\u3092\u8abf\u6574\u3059\u308b\u3002<\/li>\n\n\n\n<li>OS\u306e\u30b3\u30de\u30f3\u30c9\u3001\u30d5\u30a1\u30a4\u30eb\u3001\u304a\u3088\u3073\u8a55\u4fa1\u30c4\u30fc\u30eb\u3092\u4f7f\u7528\u3057\u3066\u3001\u30b7\u30b9\u30c6\u30e0\u76ee\u6a19\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u60c5\u5831\u3092\u53d6\u5f97\u3059\u308b\u3002<\/li>\n\n\n\n<li><strong>perf<\/strong> \u3092\u5229\u7528\u3057\u3066\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u6027\u80fd\u3092\u89b3\u5bdf\u3059\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">\u74b0\u5883\u8a2d\u5b9a\u3068\u4e8b\u524d\u30c6\u30b9\u30c8<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\">\u7c21\u6613\u30a2\u30bb\u30f3\u30d6\u30ea\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u4f5c\u6210\u3057\u3001\u30a2\u30bb\u30f3\u30d6\u30ea\u3092\u4f53\u9a13\u3059\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u30c4\u30fc\u30eb\u30e9\u30a4\u30d6\u30e9\u30ea\u304c\u6b63\u3057\u304f\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u304a\u3088\u3073\u8a2d\u5b9a\u3055\u308c\u3066\u3044\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u30a2\u30bb\u30f3\u30d6\u30ea\u8a00\u8a9e\u306e\u69cb\u6587\u3068\u30b3\u30f3\u30d1\u30a4\u30eb\u30d7\u30ed\u30bb\u30b9\u306b\u6163\u308c\u308b\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u5b9f\u9a13\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3068\u89e3\u51cd<\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u5b9f\u884c\u3059\u308b\u524d\u306b\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u304c\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u306b\u6b63\u5e38\u306b\u63a5\u7d9a\u3055\u308c\u3066\u3044\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br><code>lab1.tar.gz<\/code> \u3092<a href=\"https:\/\/www.yanagichiaki.jp\/index.php\/2024\/12\/09\/computer-architecture-labs-guidebook-hitsz\/#index_id6\">\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9<\/a>\u3057\u3001\u30e6\u30fc\u30b6\u30fc\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u30b3\u30d4\u30fc\u3057\u3066\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<p><code>tar -zxvf lab1.tar.gz<\/code><\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u5fc5\u8981\u306a\u74b0\u5883\u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb<\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>sudo apt update\nsudo apt install build-essential net-tools git vim cmake gdb make gfortran libnuma-dev libtirpc-dev<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u6b63\u3057\u304f\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3055\u308c\u305f\u304b\u78ba\u8a8d<\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>cmake --version\ngcc --version\ngdb --version<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><code>src\/lab1\/print_integer.S<\/code> \u306e <code>convert_loop<\/code>\u306b\u6b63\u3057\u3044\u7d42\u4e86\u547d\u4ee4\u3092\u8ffd\u52a0<\/strong><\/div><div class=\"swell-block-step__body\">\n\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><code>src\/lab1\/print_integer.S<\/code> \u306e <code>print_string<\/code> (48\uff5e55\u884c\u76ee) \u306b\u3042\u308b\u30a8\u30e9\u30fc\u3092\u4fee\u6b63<\/strong><\/div><div class=\"swell-block-step__body\">\n\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u306e\u30eb\u30fc\u30c8\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u79fb\u52d5\u3057\u3066\u30b3\u30fc\u30c9\u3092\u30b3\u30f3\u30d1\u30a4\u30eb<\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>cd lab1\nmkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab1_print_integer<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u30b3\u30fc\u30c9\u3092\u5b9f\u884c\u3057\u3066\u6b63\u3057\u3044\u51fa\u529b\u3092\u5f97\u308b<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>cd dist\/bins\/ &amp;&amp; .\/lab1_print_integer<\/code><\/p>\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">\u884c\u5217\u4e57\u7b97\u306e\u30b3\u30fc\u30c9\u88dc\u5b8c<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\">\u63d0\u4f9b\u3055\u308c\u305f\u30a2\u30bb\u30f3\u30d6\u30ea\u30b3\u30fc\u30c9\u3092\u7528\u3044\u3066\u884c\u5217\u7a4d\u3092\u5b9f\u88c5\u3057\u3001\u7570\u306a\u308b\u30b5\u30a4\u30ba\u306e\u884c\u5217\u306b\u5bfe\u5fdc\u3059\u308b\u3088\u3046\u306b\u30b3\u30fc\u30c9\u3092\u8abf\u6574\u3059\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u30ec\u30b8\u30b9\u30bf\u9593\u63a5\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3068\u30d9\u30fc\u30b9+\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u30a2\u30c9\u30ec\u30c3\u30b7\u30f3\u30b0\u3092\u30c7\u30e2\u30f3\u30b9\u30c8\u30ec\u30fc\u30b7\u30e7\u30f3\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u884c\u5217\u30c7\u30fc\u30bf\u3092\u521d\u671f\u5316\u3057\u3001\u30a2\u30bb\u30f3\u30d6\u30ea\u3067\u30dd\u30a4\u30f3\u30bf\u3092\u4f7f\u7528\u3059\u308b\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><code>src\/lab1\/gemm_kernel.S<\/code> \u306e <code>GEMM_INIT<\/code> \u306b\u9069\u5207\u306a\u884c\u5217B\u306e\u30a2\u30c9\u30ec\u30b9\u4fdd\u5b58\u547d\u4ee4\u3092\u8ffd\u52a0<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><code>src\/lab1\/gemm_kernel.S<\/code> \u306e <code>DO_GEMM<\/code>\u306b <code>A[m][k]<\/code> \u3092FPU\u30ec\u30b8\u30b9\u30bf\u30b9\u30bf\u30c3\u30af\u306b\u30ed\u30fc\u30c9\u3059\u308b\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><code>src\/lab1\/gemm_kernel.S<\/code> \u306e <code>DO_GEMM<\/code>\u306b<strong><strong> <\/strong><code>B[k][n]<\/code><\/strong> \u3092FPU\u30ec\u30b8\u30b9\u30bf\u30b9\u30bf\u30c3\u30af\u306b\u30ed\u30fc\u30c9\u3059\u308b\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><code>src\/lab1\/gemm_kernel.S<\/code> \u306e <code>DO_GEMM<\/code>\u306b <code>C[m][n]<\/code> \u3092FPU\u30ec\u30b8\u30b9\u30bf\u30b9\u30bf\u30c3\u30af\u306b\u30ed\u30fc\u30c9\u3059\u308b\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab1_test_gemm_kernel.unittest\n.\/dist\/bins\/lab1_test_gemm_kernel.unittest --gtest_filter=gemm_kernel.test0<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong>\u4e0a\u4f4d\u30b3\u30fc\u30c9\u3092\u30b3\u30f3\u30d1\u30a4\u30eb\u3057\u3066\u5b9f\u884c\u53ef\u80fd\u306a\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u751f\u6210<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab1_gemm<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u30b3\u30fc\u30c9\u3092\u5b9f\u884c\u3057\u3066\u51fa\u529b\u3092\u5f97\u308b<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>.\/dist\/bins\/lab1_gemm 256 256 256<\/code><\/p>\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">CPU\u30a4\u30f3\u30d5\u30a9<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\">OS\u306e\u30b3\u30de\u30f3\u30c9\u3001\u30d5\u30a1\u30a4\u30eb\u3001\u304a\u3088\u3073\u8a55\u4fa1\u30c4\u30fc\u30eb\u3092\u4f7f\u7528\u3057\u3066\u3001\u30b7\u30b9\u30c6\u30e0\u76ee\u6a19\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u60c5\u5831\u3092\u53d6\u5f97\u3059\u308b\u3002<\/p>\n\n\n\n<p>\u672a\u77e5\u306e\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u3092\u6301\u3064\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u60c5\u5831\u3092\u53d6\u5f97\u3059\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<\/p>\n\n\n\n<p class=\"is-style-sticky_box\">\u76ee\u6a19\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u968e\u5c64\u3068\u5404\u30ec\u30d9\u30eb\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u30b5\u30a4\u30ba\u3092\u53d6\u5f97\u3059\u308b\u3002<br>\u5404\u30ec\u30d9\u30eb\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u30bb\u30c3\u30c8\u76f8\u95a2\u6570\u3001\u30ad\u30e3\u30c3\u30b7\u30e5\u30e9\u30a4\u30f3\u30b5\u30a4\u30ba\u3092\u53d6\u5f97\u3059\u308b\u3002<\/p>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><code>lscpu<\/code> \u30b3\u30de\u30f3\u30c9\u3067\u30d7\u30ed\u30bb\u30c3\u30b5\u30e2\u30c7\u30eb\u3068\u30ad\u30e3\u30c3\u30b7\u30e5\u968e\u5c64\u60c5\u5831\u3092\u78ba\u8a8d<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>lscpu<\/code><\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>CPU0\u306eL1D\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u30bb\u30c3\u30c8\u6570\u3001\u30bb\u30c3\u30c8\u76f8\u95a2\u6570\u3001\u30ad\u30e3\u30c3\u30b7\u30e5\u30e9\u30a4\u30f3\u30b5\u30a4\u30ba\u3092\u78ba\u8a8d<\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>cd \/sys\/devices\/system\/cpu\/cpu0\/cache\ncd index0\n# \u30ad\u30e3\u30c3\u30b7\u30e5\u30e9\u30a4\u30f3\u30b5\u30a4\u30ba\u3092\u78ba\u8a8d\ncat coherency_line_size\n# \u30bb\u30c3\u30c8\u6570\u3092\u78ba\u8a8d\ncat number_of_sets\n# \u30bb\u30c3\u30c8\u76f8\u95a2\u6570\u3092\u78ba\u8a8d\ncat ways_of_associativity<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number\" style=\"background-color:var(--color_main)\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>CPU0\u306e<span style=\"font-size: var(--swl-fz--large); background-color: var(--color_content_bg); color: var(--color_text); font-family: var(--swl-font_family); letter-spacing: var(--swl-letter_spacing,normal);\"><span class=\"swl-fz u-fz-xs\">L2\u304a\u3088\u3073L3\u30ad\u30e3\u30c3\u30b7\u30e5\u306b\u3064\u3044\u3066\u3082\u540c\u69d8\u3067\u78ba\u8a8d<\/span><\/span><\/strong><\/div><div class=\"swell-block-step__body\">\n\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">Perf\u306e\u4f7f\u7528<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\"><strong>perf<\/strong> \u3092\u5229\u7528\u3057\u3066\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u6027\u80fd\u3092\u89b3\u5bdf\u3059\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<\/p>\n\n\n\n<p><code>perf<\/code> \u306e\u57fa\u672c\u7684\u306a\u4f7f\u7528\u65b9\u6cd5\u3092\u7fd2\u5f97\u3059\u308b\u3002<\/p>\n\n\n\n<p class=\"is-style-sticky_box\"><code>perf list<\/code> \u30b3\u30de\u30f3\u30c9\u3067\u5bfe\u5fdc\u3059\u308b\u6027\u80fd\u30a4\u30d9\u30f3\u30c8\u3092\u78ba\u8a8d\u3059\u308b\u3002<br><code>perf stat<\/code> \u3092\u4f7f\u7528\u3057\u3066\u7df4\u7fd22\u306e\u884c\u5217\u7a4d\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u5229\u7528\u72b6\u6cc1\u3092\u78ba\u8a8d\u3059\u308b\u3002<\/p>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><code>perf<\/code> \u3092\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>sudo apt install linux-tools-5.4.0-26-generic<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong>Perf \u304c\u30b5\u30dd\u30fc\u30c8\u3059\u308b\u6027\u80fd\u30a4\u30d9\u30f3\u30c8\u3092\u78ba\u8a8d<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>perf list<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number\" style=\"background-color:var(--color_main)\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><code>lab1_gemm<\/code> \u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u5229\u7528\u72b6\u6cc1\u3092\u78ba\u8a8d<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>\u57fa\u672c\u7684\u306a\u6027\u80fd\u30a4\u30d9\u30f3\u30c8\u3092\u78ba\u8a8d\nperf stat .\/dist\/bins\/lab1_gemm 256 256 256\n\n# \u6307\u5b9a\u3057\u305f\u6027\u80fd\u30a4\u30d9\u30f3\u30c8 (-e) \u3092\u78ba\u8a8d\nperf stat -e L1-dcache-loads,L1-dcache-load-misses,dTLB-loads,dTLB-load-misses .\/lab1_gemm 256 256 256<\/code><\/pre>\n<\/div><\/div>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading\">\u6b21\u306e\u5b9f\u9a13\u306e\u4e88\u7fd2<\/h3>\n\n\n\n<p>\u4ee5\u4e0b\u3092\u8aad\u307f\u3001\u9ad8\u6027\u80fd\u306a\u884c\u5217\u7a4d\u8a08\u7b97\u30ab\u30fc\u30cd\u30eb\u3092\u8a2d\u8a08\u3059\u308b\u57fa\u672c\u539f\u5247\u3092\u7406\u89e3\u3057\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>Anatomy of High-Performance Matrix Multiplication<\/li>\n\n\n\n<li>Analytical modeling is enough for high-performance BLIS<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">Lab2 \u30ad\u30e3\u30c3\u30b7\u30e5\u3001\u30eb\u30fc\u30d7\u3001\u304a\u3088\u3073\u30d6\u30ed\u30c3\u30ad\u30f3\u30b0\u3092\u4f7f\u7528\u3057\u3066\u884c\u5217\u4e57\u7b97\u3092\u6700\u9069\u5316<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u76ee\u7684<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u306e<strong>\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u4ed5\u7d44\u307f<\/strong>\u3092\u7406\u89e3\u3057\u3001\u3053\u306e\u4ed5\u7d44\u307f\u3092\u6d3b\u7528\u3057\u3066\u884c\u5217\u7a4d\u8a08\u7b97\u3092\u6700\u9069\u5316\u3059\u308b\u65b9\u6cd5\u3092\u7fd2\u5f97\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u30ad\u30e3\u30c3\u30b7\u30e5\u3092\u52b9\u679c\u7684\u306b\u5229\u7528\u3057\u3066\u884c\u5217\u7a4d\u8a08\u7b97\u306e\u30c7\u30fc\u30bf\u30a2\u30af\u30bb\u30b9\u7d4c\u8def\u3092\u6700\u9069\u5316\u3057\u3001\u547d\u4ee4\u30ec\u30d9\u30eb\u3067\u306e\u6700\u9069\u5316\u624b\u9806\u3092\u5b66\u3073\u5b9f\u8df5\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u6027\u80fd\u30dc\u30c8\u30eb\u30cd\u30c3\u30af\u3092\u5206\u6790\u3057\u3001\u547d\u4ee4\u30ec\u30d9\u30eb\u3067\u306e\u6700\u9069\u5316\u624b\u6cd5\u3092\u7406\u89e3\u3059\u308b\u3002<\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u524d\u306e\u6e96\u5099<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ee5\u4e0b\u306e\u6587\u3092\u8aad\u3093\u3067\u3001\u9ad8\u6027\u80fd\u306a\u884c\u5217\u7a4d\u8a08\u7b97\u30ab\u30fc\u30cd\u30eb\u3092\u8a2d\u8a08\u3059\u308b\u57fa\u672c\u539f\u5247\u3092\u7406\u89e3\u3057\u3066\u304f\u3060\u3055\u3044\uff1a\n<ul class=\"wp-block-list\">\n<li><strong>\u300aAnatomy of High-Performance Matrix Multiplication\u300b<\/strong><\/li>\n\n\n\n<li><strong>\u300aAnalytical modeling is enough for high-performance BLIS\u300b<\/strong><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u539f\u7406<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\">\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u547d\u4ee4 <code>prefetch<\/code><\/h4>\n\n\n\n<p>x86-64\u547d\u4ee4\u30bb\u30c3\u30c8\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306f\u3001SSE\uff08Streaming SIMD Extension\uff09\u62e1\u5f35\u547d\u4ee4\u3092\u542b\u307f\u307e\u3059\u3002\u3053\u306eSSE\u62e1\u5f35\u547d\u4ee4\u306fSIMD\u65b9\u5f0f\u3067\u30c7\u30fc\u30bf\u306e\u4e26\u5217\u51e6\u7406\u3092\u5b9f\u73fe\u3057\u3001CPU\u306e\u30c7\u30fc\u30bf\u51e6\u7406\u52b9\u7387\u3092\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002<\/p>\n\n\n\n<p><strong>\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u547d\u4ee4<\/strong>\u306f\u3001\u30c7\u30fc\u30bf\u3092\u30ad\u30e3\u30c3\u30b7\u30e5\u306b\u4e8b\u524d\u306b\u8aad\u307f\u8fbc\u3080\u305f\u3081\u306b\u4f7f\u7528\u3055\u308c\u3001\u4e3b\u306a\u547d\u4ee4\u306f\u4ee5\u4e0b\u306e\u901a\u308a\u3067\u3059\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th><strong>\u547d\u4ee4\u540d<\/strong><\/th><th><strong>\u6a5f\u80fd<\/strong><\/th><\/tr><\/thead><tbody><tr><td><code>prefetcht0<\/code><\/td><td>\u30c7\u30fc\u30bf\u3092\u3059\u3079\u3066\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u5c64\uff08L1\u3001L2\u3001L3\uff09\u306b\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u3059\u308b<\/td><\/tr><tr><td><code>prefetcht1<\/code><\/td><td>\u30c7\u30fc\u30bf\u3092L1\u30ad\u30e3\u30c3\u30b7\u30e5\u3092\u9664\u304f\u3059\u3079\u3066\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u5c64\u306b\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u3059\u308b<\/td><\/tr><tr><td><code>prefetcht2<\/code><\/td><td>\u30c7\u30fc\u30bf\u3092L1\u3001L2\u30ad\u30e3\u30c3\u30b7\u30e5\u3092\u9664\u304f\u3059\u3079\u3066\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u5c64\u306b\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u3059\u308b<\/td><\/tr><tr><td><code>prefetchnta<\/code><\/td><td>\u65e2\u5b58\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u30c7\u30fc\u30bf\u306b\u5f71\u97ff\u3092\u4e0e\u3048\u305a\u3001\u7279\u5b9a\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u5c64\u306b\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u3059\u308b<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<p><strong>\u4f7f\u7528\u4f8b\uff1a<\/strong><\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>prefetch&lt;t0\/t1\/t2\/nta&gt; (&lt;byte_addr&gt;)<\/code><\/pre>\n\n\n\n<ul class=\"wp-block-list\">\n<li><code>byte_addr<\/code> \u306f\u30d0\u30a4\u30c8\u30a2\u30c9\u30ec\u30b9\u3092\u8868\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u547d\u4ee4\u306f\u975e\u30d6\u30ed\u30c3\u30ad\u30f3\u30b0\u3067\u3042\u308a\u3001\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u64cd\u4f5c\u304c\u5b8c\u4e86\u3059\u308b\u306e\u3092\u5f85\u305f\u305a\u306b\u5f8c\u7d9a\u306e\u547d\u4ee4\u3092\u5b9f\u884c\u3067\u304d\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">\u5206\u5272\u884c\u5217\u7a4d\u3068\u30ad\u30e3\u30c3\u30b7\u30e5\u6700\u9069\u5316<\/h4>\n\n\n\n<p>\u73fe\u5728\u3001\u4e3b\u6d41\u306e\u7dda\u5f62\u4ee3\u6570\u30e9\u30a4\u30d6\u30e9\u30ea\u3067\u306f\u3001\u591a\u304f\u306e\u5834\u5408\u3001\u56f32-1\u306b\u793a\u3055\u308c\u308b<strong>Goto\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0<\/strong>\u3084\u305d\u306e\u6d3e\u751f\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u4f7f\u7528\u3057\u3066\u884c\u5217\u7a4d\u3092\u8a08\u7b97\u3057\u3066\u3044\u307e\u3059\u3002<strong>Goto\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0<\/strong>\u306e\u6838\u5fc3\u306f\u3001\u5206\u5272\u884c\u5217\u7a4d\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306e\u8a08\u7b97\u624b\u9806\u306b\u5f93\u3063\u3066\u3001\u8a08\u7b97\u306b\u5fc5\u8981\u306a\u30c7\u30fc\u30bf\u30d6\u30ed\u30c3\u30af\u3092\u7570\u306a\u308b\u30ad\u30e3\u30c3\u30b7\u30e5\u306b\u914d\u7f6e\u3059\u308b\u3053\u3068\u3067\u3001\u30c7\u30fc\u30bf\u306e\u30a2\u30af\u30bb\u30b9\u6642\u9593\u3092\u5206\u5272\u8a08\u7b97\u5185\u306b\u96a0\u853d\u3059\u308b\u3053\u3068\u306b\u3042\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized\"><img decoding=\"async\" width=\"546\" height=\"833\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-2-1.png\" alt=\"\" class=\"wp-image-1742\" style=\"width:421px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-2-1.png 546w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-2-1-197x300.png 197w\" sizes=\"(max-width: 546px) 100vw, 546px\" \/><figcaption class=\"wp-element-caption\">\u56f32-1<\/figcaption><\/figure>\n\n\n\n<p><strong>Goto\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0<\/strong>\u306e\u7279\u5fb4\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u56f32-1\u306b\u793a\u3055\u308c\u308b\u3088\u3046\u306b\u3001<strong>\u516d\u91cd\u30eb\u30fc\u30d7\u69cb\u9020<\/strong>\u3092\u6301\u3061\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u7b2c4\u5c64\u304b\u3089\u7b2c6\u5c64\u307e\u3067\u306e\u30eb\u30fc\u30d7\u306f\u4e00\u822c\u306b\u300cKernel\uff08\u30ab\u30fc\u30cd\u30eb\uff09\u300d\u3068\u547c\u3070\u308c\u3001\u8a08\u7b97\u52b9\u7387\u3092\u78ba\u4fdd\u3059\u308b\u305f\u3081\u3001\u901a\u5e38\u306f\u624b\u66f8\u304d\u306e\u30a2\u30bb\u30f3\u30d6\u30ea\u3067\u5b9f\u88c5\u3055\u308c\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u7b2c6\u5c64\u30eb\u30fc\u30d7\u306e <code>mr<\/code>\u3001<code>nr<\/code><\/strong>\n<ul class=\"wp-block-list\">\n<li>\u3053\u308c\u3089\u306f\u30d7\u30ed\u30bb\u30c3\u30b5\u30b3\u30a2\u306e\u4f7f\u7528\u53ef\u80fd\u306a\u30ec\u30b8\u30b9\u30bf\u6570\u3068\u5bc6\u63a5\u306b\u95a2\u9023\u3057\u3066\u3044\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u4e00\u822c\u306b\u3001\u307b\u3068\u3093\u3069\u306e\u30c7\u30fc\u30bf\u30ec\u30b8\u30b9\u30bf\u306f\u884c\u5217 <strong>C<\/strong> \u306e\u5206\u5272\uff08<code>mr \u00d7 nr<\/code> \u306e\u30b5\u30a4\u30ba\uff09\u306b\u5272\u308a\u5f53\u3066\u3089\u308c\u307e\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u3001\u8a08\u7b97\u306b\u5fc5\u8981\u306a\u884c\u5217 <strong>A<\/strong> \u306e\u30c7\u30fc\u30bf\u3092 <strong>L2\u30ad\u30e3\u30c3\u30b7\u30e5<\/strong> \u304b\u3089\u884c\u5358\u4f4d\u3067\u30ed\u30fc\u30c9\u3057\u3001\u884c\u5217 <strong>B<\/strong> \u306e\u30c7\u30fc\u30bf\u3092 <strong>L1D\u30ad\u30e3\u30c3\u30b7\u30e5<\/strong> \u304b\u3089\u30ed\u30fc\u30c9\u3059\u308b\u305f\u3081\u306e\u5341\u5206\u306a\u6642\u9593\u3092\u78ba\u4fdd\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u7b2c5\u5c64\u30eb\u30fc\u30d7\u306e <code>kc<\/code><\/strong>\n<ul class=\"wp-block-list\">\n<li>\u3053\u3053\u3067\u306f\u3001<code>kc \u00d7 nr<\/code> \u306e\u30b5\u30a4\u30ba\u306e\u884c\u5217 <strong>B<\/strong> \u306e\u30c7\u30fc\u30bf\u304c <strong>L1D\u30ad\u30e3\u30c3\u30b7\u30e5<\/strong> \u306e\u5927\u90e8\u5206\u3092\u5360\u3081\u308b\u3088\u3046\u306b\u8a2d\u5b9a\u3055\u308c\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u6b8b\u308a\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u5bb9\u91cf\u306f\u3001\u884c\u5217 <strong>C<\/strong> \u3068\u884c\u5217 <strong>A<\/strong> \u306b\u5272\u308a\u5f53\u3066\u3089\u308c\u307e\u3059\u3002\u3053\u306e\u3088\u3046\u306b\u3059\u308b\u3068\u3001\u884c\u5217 <strong>B<\/strong> \u306e\u30c7\u30fc\u30bf\u306f\u7e70\u308a\u8fd4\u3057\u5229\u7528\u3055\u308c\u3001<strong>L1D\u30ad\u30e3\u30c3\u30b7\u30e5<\/strong> \u306b\u7559\u307e\u308a\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u305f\u3060\u3057\u3001<code>kc<\/code> \u304c\u5c0f\u3055\u3059\u304e\u308b\u5834\u5408\u3001\u7b2c2\u5c64\u30eb\u30fc\u30d7\u3067\u884c\u5217 <strong>C<\/strong> \u306e\u90e8\u5206\u548c\u306e\u7d2f\u7a4d\u56de\u6570\u304c\u5897\u3048\u307e\u3059\u3002\u884c\u5217 <strong>C<\/strong> \u304c\u30e1\u30e2\u30ea\u306b\u683c\u7d0d\u3055\u308c\u3066\u3044\u308b\u305f\u3081\u3001\u904e\u5270\u306a\u7d2f\u7a4d\u306f\u8a08\u7b97\u901f\u5ea6\u3092\u5927\u5e45\u306b\u4f4e\u4e0b\u3055\u305b\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u7b2c3\u5c64\u30eb\u30fc\u30d7\u306e <code>mc<\/code><\/strong>\n<ul class=\"wp-block-list\">\n<li><code>mc \u00d7 kc<\/code> \u306e\u30b5\u30a4\u30ba\u306e\u884c\u5217 <strong>A<\/strong> \u306e\u30c7\u30fc\u30bf\u304c <strong>L2\u30ad\u30e3\u30c3\u30b7\u30e5<\/strong> \u306e\u5927\u90e8\u5206\u3092\u5360\u3081\u308b\u3088\u3046\u306b\u8a2d\u5b9a\u3055\u308c\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u8a08\u7b97\u4e2d\u3001\u3053\u306e\u30c7\u30fc\u30bf\u306f <strong>L2\u30ad\u30e3\u30c3\u30b7\u30e5<\/strong> \u304b\u3089\u30ec\u30b8\u30b9\u30bf\u3078\u76f4\u63a5\u6d41\u308c\u8fbc\u307f\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>\u7b2c1\u5c64\u30eb\u30fc\u30d7\u306e <code>nc<\/code><\/strong>\n<ul class=\"wp-block-list\">\n<li>\u7b2c3\u5c64\u30eb\u30fc\u30d7\u3067\u884c\u5217 <strong>B<\/strong> \u306e\u30c7\u30fc\u30bf\u304c <strong>L3\u30ad\u30e3\u30c3\u30b7\u30e5<\/strong> \u306e\u5927\u90e8\u5206\u3092\u5360\u3081\u308b\u3088\u3046\u306b\u9078\u629e\u3055\u308c\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ol>\n\n\n\n<p><strong>\u5206\u5272\u3068Packing\u306b\u3064\u3044\u3066<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u884c\u5217\u306e\u5206\u5272\u5f8c\u3001\u5206\u5272\u8981\u7d20\u306e\u30a2\u30af\u30bb\u30b9\u304c\u4e0d\u9023\u7d9a\u306b\u306a\u308b\u554f\u984c\u304c\u751f\u3058\u308b\u53ef\u80fd\u6027\u304c\u3042\u308a\u307e\u3059\u3002\u305d\u306e\u305f\u3081\u3001\u901a\u5e38\u306f\u7b2c3\u5c64\u30eb\u30fc\u30d7\u3068\u7b2c4\u5c64\u30eb\u30fc\u30d7\u3067\u30c7\u30fc\u30bf\u3092<strong>Packing<\/strong>\u3057\u3066\u3001Kernel\u306b\u3088\u308b\u30a2\u30af\u30bb\u30b9\u306e\u9023\u7d9a\u6027\u3068\u7a7a\u9593\u5c40\u6240\u6027\u3092\u78ba\u4fdd\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>Packing<\/strong> \u306f\u30b3\u30b9\u30c8\u3092\u4f34\u3046\u305f\u3081\u3001\u5e38\u306b\u6709\u76ca\u3068\u3044\u3046\u308f\u3051\u3067\u306f\u3042\u308a\u307e\u305b\u3093\u3002\u67d4\u8edf\u306aPacking\u6226\u7565\u304c\u3001\u884c\u5217\u7a4d\u8a08\u7b97\u306e\u6027\u80fd\u5411\u4e0a\u306b\u5bc4\u4e0e\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4e3b\u6d41\u30d7\u30ed\u30bb\u30c3\u30b5\u3067\u306f\u3001\u8a08\u7b97\u30e6\u30cb\u30c3\u30c8\u3068\u30e1\u30e2\u30ea\u30a2\u30af\u30bb\u30b9\u30e6\u30cb\u30c3\u30c8\u304c\u72ec\u7acb\u3057\u3066\u52d5\u4f5c\u3057\u307e\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u3001\u8a08\u7b97\u4e2d\u306b\u6b21\u56de\u5fc5\u8981\u306a\u30c7\u30fc\u30bf\u3092\u9069\u5207\u306a\u4f4d\u7f6e\u306b\u4e8b\u524d\u306b\u6e96\u5099\u3059\u308b\u3053\u3068\u3067\u3001\u30c7\u30fc\u30bf\u8aad\u307f\u8fbc\u307f\u6642\u9593\u3092\u5927\u5e45\u306b\u524a\u6e1b\u3057\u3001\u9ad8\u3044\u8a08\u7b97\u6027\u80fd\u3092\u5f97\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u305f\u3060\u3057\u3001\u30c7\u30fc\u30bf\u306e\u4e8b\u524d\u6e96\u5099\u306b\u3088\u3063\u3066<strong>way\u306e\u7af6\u5408<\/strong>\u3084\u6df1\u523b\u306a<strong>\u30ad\u30e3\u30c3\u30b7\u30e5\u30a8\u30d3\u30af\u30b7\u30e7\u30f3<\/strong>\uff08\u8ffd\u3044\u51fa\u3057\uff09\u304c\u767a\u751f\u3057\u306a\u3044\u3088\u3046\u306b\u6ce8\u610f\u304c\u5fc5\u8981\u3067\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u9ad8\u6027\u80fd\u884c\u5217\u7a4d\u8a08\u7b97\u306e\u8a2d\u8a08\u3068\u6700\u9069\u5316\u65b9\u6cd5\u3092\u4f53\u7cfb\u7684\u306b\u7406\u89e3\u3059\u308b\u306b\u306f\u3001\u4ee5\u4e0b\u306e\u8cc7\u6599\u3092\u53c2\u7167\u3057\u3066\u304f\u3060\u3055\u3044\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/www.cs.utexas.edu\/~flame\/pubs\/GotoTOMS_revision.pdf\">Anatomy of High-Performance Matrix Multiplication<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/repositories.lib.utexas.edu\/server\/api\/core\/bitstreams\/289b65bc-85ae-461e-a15e-4c78537e10eb\/content\">Theory and Practice of Classical Matrix-Matrix Multiplication for Hierarchical Memory Architectures<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/www.cs.utexas.edu\/~flame\/pubs\/TOMS-BLIS-Analytical.pdf\">Analytical modeling is enough for high-performance BLIS<\/a><\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u5185\u5bb9<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u6027\u80fd\u5206\u6790\u30c4\u30fc\u30eb<\/strong>\u3092\u4f7f\u7528\u3057\u3066\u6027\u80fd\u30dc\u30c8\u30eb\u30cd\u30c3\u30af\u3092\u5206\u6790\u3059\u308b\u3002<\/li>\n\n\n\n<li><strong>\u30c7\u30fc\u30bf\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u6a5f\u69cb<\/strong>\u3092\u5229\u7528\u3057\u3066\u884c\u5217\u7a4d\u306e\u6027\u80fd\u3092\u6700\u9069\u5316\u3059\u308b\u3002<\/li>\n\n\n\n<li><strong>\u30eb\u30fc\u30d7\u5c55\u958b\u3068\u30d6\u30ed\u30c3\u30ad\u30f3\u30b0<\/strong>\u3092\u6d3b\u7528\u3057\u3066\u8a08\u7b97\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u3002<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">Perf\u3067\u884c\u5217\u4e57\u7b97\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u306e\u30dc\u30c8\u30eb\u30cd\u30c3\u30af\u3092\u7279\u5b9a<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\"><strong>\u6027\u80fd\u5206\u6790\u30c4\u30fc\u30eb<\/strong>\u3092\u4f7f\u7528\u3057\u3066\u6027\u80fd\u30dc\u30c8\u30eb\u30cd\u30c3\u30af\u3092\u5206\u6790\u3059\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><code>perf<\/code> \u3092\u4f7f\u7528\u3057\u3066\u30b5\u30f3\u30d7\u30eb\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u30d2\u30c3\u30c8\u7387\u3092\u5206\u6790\u3059\u308b\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u5b9f\u9a13\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3068\u89e3\u51cd<\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u5b9f\u884c\u3059\u308b\u524d\u306b\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u304c\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u306b\u6b63\u5e38\u306b\u63a5\u7d9a\u3055\u308c\u3066\u3044\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br><code>lab2.tar.gz<\/code> \u3092<a href=\"https:\/\/www.yanagichiaki.jp\/index.php\/2024\/12\/09\/computer-architecture-labs-guidebook-hitsz\/#index_id6\">\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9<\/a>\u3057\u3001\u30e6\u30fc\u30b6\u30fc\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u30b3\u30d4\u30fc\u3057\u3066\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<p><code>tar -zxvf lab2.tar.gz<\/code><\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u30d0\u30a4\u30ca\u30ea\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u751f\u6210<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>mkdir -p build &amp;&amp; cd build <\/code><br><code>cmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab2_gemm_baseline<\/code><\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><code>perf<\/code> \u3092\u4f7f\u7528\u3057\u3066\u6027\u80fd\u3092\u5206\u6790\uff1a<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>perf stat -e l2_rqsts.code_rd_hit,l1d.replacement,L1-dcache-loads,L1-dcache-load-misses .\/dist\/bins\/lab2_gemm_baseline 256 1024 256<\/code><\/p>\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">Prefetch\u3092\u5229\u7528\u3057\u3066\u884c\u5217\u4e57\u7b97\u306e\u6027\u80fd\u3092\u6700\u9069\u5316\u3059\u308b<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\"><strong>\u30c7\u30fc\u30bf\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u6a5f\u69cb<\/strong>\u3092\u5229\u7528\u3057\u3066\u884c\u5217\u7a4d\u306e\u6027\u80fd\u3092\u6700\u9069\u5316\u3059\u308b\u3002<\/p>\n\n\n\n<p>\u30d7\u30ed\u30bb\u30c3\u30b5\u30e1\u30fc\u30ab\u30fc\u304c\u63d0\u4f9b\u3059\u308b<a href=\"https:\/\/www.intel.co.jp\/content\/www\/jp\/ja\/content-details\/782158\/intel-64-and-ia-32-architectures-software-developer-s-manual-combined-volumes-1-2a-2b-2c-2d-3a-3b-3c-3d-and-4.html\">\u958b\u767a\u30de\u30cb\u30e5\u30a2\u30eb<\/a>\u3092\u53c2\u7167\u3057\u3001\u9069\u5207\u306a\u547d\u4ee4\u3092\u9078\u629e\u3059\u308b\u65b9\u6cd5\u3092\u5b66\u3076\uff0810.4.6\u7bc0\u306e\u30c7\u30fc\u30bf\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u95a2\u9023\u5185\u5bb9\uff09\u3002<br>\u547d\u4ee4\u30b9\u30c8\u30ea\u30fc\u30e0\u5185\u3067\u30c7\u30fc\u30bf\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u547d\u4ee4\u306e\u9069\u5207\u306a\u914d\u7f6e\u3092\u9078\u5b9a\u3059\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>lmbench3\u3092\u4f7f\u7528\u3057\u3066\u5404\u30ec\u30d9\u30eb\u306e\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u30a2\u30af\u30bb\u30b9\u30ec\u30a4\u30c6\u30f3\u30b7\u3092\u53d6\u5f97\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u30a2\u30af\u30bb\u30b9\u30ec\u30a4\u30c6\u30f3\u30b7\u60c5\u5831\u3092\u57fa\u306b\u3001\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u547d\u4ee4\u306e\u9069\u5207\u306a\u4f4d\u7f6e\u3092\u9078\u5b9a\u3059\u308b\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><code>lmbench3<\/code>\u3092\u4f7f\u7528\u3057\u3066\u30ad\u30e3\u30c3\u30b7\u30e5\u306e\u30a2\u30af\u30bb\u30b9\u9045\u5ef6\u3092\u6e2c\u5b9a<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p><strong>\u6ce8\u610f<\/strong>\uff1a\u5e73\u5747\u5024\u3092\u53d6\u5f97\u3059\u308b\u305f\u3081\u3001\u8907\u6570\u56de\u306e\u5b9f\u884c\u304c\u5fc5\u8981\u3067\u3059\u30021\u56de\u306e\u5b9f\u884c\u306b\u306f\u7d047\u6642\u9593\u304b\u304b\u308b\u305f\u3081\u3001\u518d\u8a66\u884c\u56de\u6570\u3092\u5408\u7406\u7684\u306b\u8abf\u6574\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>cd tools &amp;&amp; tar xf lmbench.tgz &amp;&amp; cd lmbench\nmake results<\/code><\/pre>\n\n\n\n<p>\u8a2d\u5b9a\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u9078\u629e\u4f8b\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>MULTIPLE COPIES [default 1]<\/strong>: Enter\u30ad\u30fc\u3092\u62bc\u3059\u3002<\/li>\n\n\n\n<li><strong>Job placement selection: 4<\/strong>\u3002<\/li>\n\n\n\n<li><strong>MB [default 10956]<\/strong>: Enter\u30ad\u30fc\u3092\u62bc\u3059\u3002<\/li>\n\n\n\n<li><strong>SUBSET (ALL|HARWARE|OS|DEVELOPMENT) [default all]<\/strong>: <code>HARWARE<\/code>\u3092\u5165\u529b\u3002<\/li>\n\n\n\n<li><strong>FASTMEM [default no]<\/strong>: Enter\u30ad\u30fc\u3092\u62bc\u3059\u3002<\/li>\n\n\n\n<li><strong>SLOWFS [default no]<\/strong>: <code>yes<\/code>\u3092\u5165\u529b\u3002<\/li>\n\n\n\n<li><strong>DISKS [default none]<\/strong>: Enter\u30ad\u30fc\u3092\u62bc\u3059\u3002<\/li>\n\n\n\n<li><strong>REMOTE [default none]<\/strong>: Enter\u30ad\u30fc\u3092\u62bc\u3059\u3002<\/li>\n\n\n\n<li><strong>Processor mhz [default 2688 MHz, 0.3720 nanosec clock]<\/strong>: Enter\u30ad\u30fc\u3092\u62bc\u3059\u3002<\/li>\n\n\n\n<li><strong>FSDIR [default \/var\/tmp]<\/strong>: Enter\u30ad\u30fc\u3092\u62bc\u3059\u3002<\/li>\n\n\n\n<li><strong>Status output file [default \/dev\/tty]<\/strong>: <code>\/tmp\/lmbench.test<\/code>\u3092\u5165\u529b\u3002<\/li>\n\n\n\n<li><strong>Mail results [default yes]<\/strong>: <code>no<\/code>\u3092\u5165\u529b\u3002<\/li>\n<\/ul>\n\n\n\n<p><em><code>tools<\/code>\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u5185\u306e\u30d5\u30a1\u30a4\u30eb\u304c<code>make results<\/code>\u30b3\u30de\u30f3\u30c9\u306e\u51fa\u529b\u3067\u3059\u3002<\/em><\/p>\n\n\n\n<p><strong>\u5b9f\u884c\u304c\u5931\u6557\u3057\u305f\u5834\u5408\u306e\u5bfe\u51e6\u624b\u9806<\/strong>\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><code>LMBENCH3<\/code>\u306f\u65b0\u3057\u3044\u30b7\u30b9\u30c6\u30e0\u3068\u306e\u4e92\u63db\u6027\u306b\u554f\u984c\u304c\u3042\u308b\u53ef\u80fd\u6027\u304c\u3042\u308b\u305f\u3081\u3001\u3082\u3057\u4ee5\u4e0a\u306e\u9078\u629e\u3092\u7279\u5b9a\u5f8c\u3001\u30d7\u30ed\u30b0\u30e9\u30e0\u304c\u7570\u5e38\u30c0\u30a6\u30f3\u767a\u751f\u3002\u4ee5\u4e0b\u3092\u5b9f\u884c\u3057\u3066\u7d9a\u884c\u3057\u307e\u3059\u3002<br><code>cp -a results .\/bin\/ make rerun<\/code><\/li>\n\n\n\n<li>\u5b9f\u884c\u7d42\u4e86\u3092\u5f85\u3064\u3002<\/li>\n\n\n\n<li>\u7d50\u679c\u3092\u4fdd\u5b58\u3057\u518d\u5b9f\u884c\uff1a<br><code>mv results results.bak<\/code><br><code>cp -a .\/bin\/results .\/<\/code><\/li>\n\n\n\n<li>\u30ec\u30a4\u30c6\u30f3\u30b7\u60c5\u5831\u3092\u78ba\u8a8d\uff1a<br><code>cd results &amp;&amp; make LIST=$(..\/scripts\/os)\/*<\/code><\/li>\n<\/ol>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><code>src\/lab2\/gemm_kernel.S<\/code>\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u547d\u4ee4\u3092\u8ffd\u52a0\u3057\u3066\u6700\u9069\u5316<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>gemm_kernel_baseline.S<\/code>\u5185\u306e<code>DO_GEMM<\/code>\u30b3\u30fc\u30c9\u3092<code>gemm_kernel_opt_prefetch.S<\/code>\u306b\u7f6e\u304d\u63db\u3048\u3001\u305d\u306e\u4e0a\u3067\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u6700\u9069\u5316\u30b3\u30fc\u30c9\u3092\u8ffd\u52a0\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u4ee5\u4e0b\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab2_gemm_kernel_opt_prefetch.unittest\ncd dist\/bins &amp;&amp; .\/lab2_gemm_kernel_opt_prefetch.unittest<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong>\u6027\u80fd\u306e\u6bd4\u8f03\u3068\u5206\u6790<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab2_gemm_opt_prefetch\ncd dist\/bins &amp;&amp; .\/lab2_gemm_opt_prefetch 1024 128 4<\/code><\/pre>\n\n\n\n<p>\u7d50\u679c\u3092\u7528\u3044\u3066\u3001\u3053\u306e\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u6700\u9069\u5316\u306e\u52b9\u679c\u3092\u8a55\u4fa1\u3057\u3066\u3002\u305d\u306e\u539f\u56e0\u3092\u8aac\u660e\u3057\u306a\u3055\u3044\u3002<\/p>\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">\u30eb\u30fc\u30d7\u3068\u30d6\u30ed\u30c3\u30ad\u30f3\u30b0\u3092\u5229\u7528\u3057\u3066\u884c\u5217\u4e57\u7b97\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b<\/h4>\n\n\n\n<p class=\"is-style-big_icon_hatena\">\u3053\u306e\u90e8\u5206\u306f\u3001\u304a\u597d\u304d\u306a\u3088\u3046\u306b\u9078\u3093\u3067\u3001\u5b9f\u88c5\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p class=\"is-style-icon_pen\"><strong>\u30eb\u30fc\u30d7\u5c55\u958b\u3068\u30d6\u30ed\u30c3\u30ad\u30f3\u30b0<\/strong>\u3092\u6d3b\u7528\u3057\u3066\u8a08\u7b97\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u3002<\/p>\n\n\n\n<p>\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306e\u6700\u9069\u5316\u3092\u901a\u3058\u3066\u30ad\u30e3\u30c3\u30b7\u30e5\u30d2\u30c3\u30c8\u7387\u3092\u5411\u4e0a\u3055\u305b\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<br>\u30ad\u30e3\u30c3\u30b7\u30e5\u30b5\u30a4\u30ba\u306b\u57fa\u3065\u304d\u884c\u5217\u7a4d\u306e\u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba\u3092\u8abf\u6574\u3057\u3066\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u30d7\u30ea\u30d5\u30a7\u30c3\u30c1\u306e\u30b3\u30fc\u30c9\u3092\u4fee\u6b63\u3057\u3001\u9069\u5207\u306a\u30eb\u30fc\u30d7\u9806\u5e8f\u3068\u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba\u3092\u8a2d\u8a08\u3057\u3066\u3001\u884c\u5217\u7a4d\u306e\u6027\u80fd\u3092\u3055\u3089\u306b\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u672c\u7df4\u7fd2\u306f\u4efb\u610f\u9078\u629e\u3067\u3059\u3002\u5177\u4f53\u7684\u306a\u5b9f\u88c5\u5185\u5bb9\u3084\u52a0\u901f\u52b9\u679c\u306b\u57fa\u3065\u304d\u52a0\u70b9\u8a55\u4fa1\u3055\u308c\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u884c\u5217\u8a08\u7b97\u30ed\u30b8\u30c3\u30af\u3092\u5b9f\u88c5<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>src\/lab2\/gemm_kernel_opt_loop.S<\/code>\u5185\u306e<code>DO_GEMM<\/code>\u30d7\u30ed\u30bb\u30b9\u306e\u884c\u5217\u8a08\u7b97\u30ed\u30b8\u30c3\u30af\u3092\u5b8c\u6210\u3055\u305b\u307e\u3059\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u306e\u30eb\u30fc\u30c8\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u3067\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab2_gemm_kernel_opt_loop.unittest\n.\/dist\/bins\/lab2_gemm_kernel_opt_loop.unittest<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number\" style=\"background-color:var(--color_main)\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong>\u6027\u80fd\u306e\u6bd4\u8f03\u3068\u5206\u6790<\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab2_gemm_opt_loop\n.\/dist\/bins\/lab2_gemm_opt_loop 4 32768 4<\/code><\/pre>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u6700\u9069\u5316\u5f8c\u306e\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3068\u57fa\u6e96\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0(Baseline)\u3092\u6bd4\u8f03<\/strong>\u3057\u3001\u6027\u80fd\u3092\u6e2c\u5b9a\u3057\u307e\u3059\uff08\u30c6\u30b9\u30c8\u30b1\u30fc\u30b9\u306f\u6700\u9069\u5316\u306e\u5185\u5bb9\u306b\u5fdc\u3058\u3066\u9078\u629e\u3057\u3066\u304f\u3060\u3055\u3044\uff09\u3002<\/li>\n\n\n\n<li><strong>\u6027\u80fd\u7d50\u679c\u306b\u57fa\u3065\u304d\u3001\u6700\u9069\u5316\u624b\u6cd5\u306e\u52b9\u679c\u7bc4\u56f2\u3092\u5206\u6790<\/strong>\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading\">\u6b21\u306e\u5b9f\u9a13\u306e\u4e88\u7fd2<\/h3>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u8cc7\u6599\u3092\u8aad\u307f\u3001\u9ad8\u6027\u80fd\u884c\u5217\u7a4d\u8a08\u7b97\u30ab\u30fc\u30cd\u30eb\u306e\u8a2d\u8a08\u539f\u5247\u3068Intel\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u7279\u6027\u3092\u7406\u89e3\u3057\u3066\u304f\u3060\u3055\u3044\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>LIBXSMM: Accelerating Small Matrix Multiplications by Runtime Code Generation<\/li>\n\n\n\n<li><a href=\"https:\/\/www.intel.co.jp\/content\/www\/jp\/ja\/content-details\/782158\/intel-64-and-ia-32-architectures-software-developer-s-manual-combined-volumes-1-2a-2b-2c-2d-3a-3b-3c-3d-and-4.html\">Intel\u00ae 64 and IA-32 Architectures Software Developer&#8217;s Manual<\/a>\uff08Volume 1 &amp; 2, Chapter 5: FPU, Chapter 14: AVX instructions\uff09<\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">Lab3 \u547d\u4ee4\u30ec\u30d9\u30eb\u306e\u4e26\u5217\u6027\u3001\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u3001\u304a\u3088\u3073\u4e26\u5217\u51e6\u7406\u3092\u4f7f\u7528\u3057\u3066\u884c\u5217\u4e57\u7b97\u3092\u6700\u9069\u5316<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u76ee\u7684<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30b7\u30b9\u30c6\u30e0\u306b\u304a\u3051\u308b\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u6027\uff08Instruction Level Parallelism, ILP\uff09\u3092\u7406\u89e3\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u30eb\u30fc\u30d7\u5c55\u958b\u3001\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u3001\u591a\u30b9\u30ec\u30c3\u30c9\u6280\u8853\u3092\u6d3b\u7528\u3057\u3001\u73fe\u4ee3\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u591a\u69d8\u306a\u7279\u6027\u3092\u7dcf\u5408\u7684\u306b\u5229\u7528\u3057\u3066\u30d7\u30ed\u30b0\u30e9\u30e0\u6027\u80fd\u3092\u6700\u9069\u5316\u3059\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<\/li>\n\n\n\n<li>\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u6027\u3001\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u3001\u591a\u30b9\u30ec\u30c3\u30c9\u6280\u8853\u3092\u6d3b\u7528\u3057\u3066\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u4e26\u5217\u6027\u80fd\u3092\u3055\u3089\u306b\u5f15\u304d\u51fa\u3059\u624b\u6cd5\u3092\u7fd2\u5f97\u3059\u308b\u3002<\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u524d\u306e\u6e96\u5099<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ee5\u4e0b\u306e\u6587\u3092\u8aad\u3093\u3067\u3001\u9ad8\u6027\u80fd\u306a\u884c\u5217\u7a4d\u8a08\u7b97\u30ab\u30fc\u30cd\u30eb\u3092\u8a2d\u8a08\u3059\u308b\u57fa\u672c\u539f\u5247\u3092\u7406\u89e3\u3057\u3066\u304f\u3060\u3055\u3044\uff1a\n<ul class=\"wp-block-list\">\n<li>LIBXSMM: Accelerating Small Matrix Multiplications by Runtime Code Generation<\/li>\n\n\n\n<li><a href=\"https:\/\/www.intel.co.jp\/content\/www\/jp\/ja\/content-details\/782158\/intel-64-and-ia-32-architectures-software-developer-s-manual-combined-volumes-1-2a-2b-2c-2d-3a-3b-3c-3d-and-4.html\">Intel\u00ae 64 and IA-32 Architectures Software Developer&#8217;s Manual<\/a>\uff08Volume 1 &amp; 2, Chapter 5: FPU, Chapter 14: AVX instructions\uff09<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u539f\u7406<\/h3>\n\n\n\n<p><strong>\u30c7\u30fc\u30bf\u4e26\u5217\u51e6\u7406<\/strong><\/p>\n\n\n\n<p>\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u304c\u9ad8\u901f\u5316\u3092\u5b9f\u73fe\u3059\u308b\u539f\u7406\u306f\u3001\u4e3b\u306b<strong>SIMD\uff08Single Instruction Multiple Data\uff09\u6280\u8853<\/strong>\u306b\u57fa\u3065\u3044\u3066\u3044\u307e\u3059\u3002SIMD\u306f\u30011\u3064\u306eCPU\u547d\u4ee4\u3067\u8907\u6570\u306e\u30c7\u30fc\u30bf\u306b\u5bfe\u3057\u3066\u540c\u6642\u306b\u64cd\u4f5c\u3092\u884c\u3046\u3053\u3068\u3092\u53ef\u80fd\u306b\u3057\u3001\u5927\u91cf\u306e\u30c7\u30fc\u30bf\u3092\u51e6\u7406\u3059\u308b\u969b\u306e\u52b9\u7387\u3092\u98db\u8e8d\u7684\u306b\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u304c\u9ad8\u901f\u5316\u3092\u53ef\u80fd\u306b\u3059\u308b\u4e3b\u306a\u539f\u7406<\/h4>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u8907\u6570\u306e\u30c7\u30fc\u30bf\u8981\u7d20\u3092\u4e26\u5217\u51e6\u7406<\/strong><br>\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u306f\u3001\u30d7\u30ed\u30bb\u30c3\u30b5\u304c\u8907\u6570\u306e\u30c7\u30fc\u30bf\u8981\u7d20\u3092\u540c\u6642\u306b\u64cd\u4f5c\u3067\u304d\u308b\u3088\u3046\u306b\u3057\u307e\u3059\u3002\u3053\u308c\u3089\u306e\u30c7\u30fc\u30bf\u8981\u7d20\u306f\u901a\u5e38\u30011\u3064\u306e\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf\u306b\u683c\u7d0d\u3055\u308c\u307e\u3059\u3002\u4f8b\u3048\u3070\u30011\u3064\u306e\u30d9\u30af\u30c8\u30eb\u52a0\u7b97\u547d\u4ee4\u3067\u306f\u30012\u3064\u306e\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf\u5185\u306e\u5bfe\u5fdc\u3059\u308b\u8981\u7d20\u3092\u540c\u6642\u306b\u52a0\u7b97\u3057\u3001\u7d50\u679c\u3092\u5225\u306e\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf\u306b\u683c\u7d0d\u3067\u304d\u307e\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u30011\u3064\u306e\u547d\u4ee4\u3067\u8907\u6570\u306e\u30b9\u30ab\u30e9\u64cd\u4f5c\u304c\u5b9f\u884c\u3055\u308c\u3001\u30c7\u30fc\u30bf\u51e6\u7406\u901f\u5ea6\u304c\u5927\u5e45\u306b\u5411\u4e0a\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u52b9\u7387\u7684\u306a\u30e1\u30e2\u30ea\u30a2\u30af\u30bb\u30b9<\/strong><br>\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u306f\u901a\u5e38\u3001\u9023\u7d9a\u3059\u308b\u30e1\u30e2\u30ea\u30c7\u30fc\u30bf\u3092\u52b9\u7387\u7684\u306b\u51e6\u7406\u3067\u304d\u308b\u3088\u3046\u306b\u8a2d\u8a08\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u30e1\u30e2\u30ea\u5185\u306e\u30c7\u30fc\u30bf\u304c\u9023\u7d9a\u3057\u3066\u914d\u7f6e\u3055\u308c\u3066\u3044\u308b\u5834\u5408\u3001\u30d9\u30af\u30c8\u30eb\u30ed\u30fc\u30c9\u304a\u3088\u3073\u30b9\u30c8\u30a2\u547d\u4ee4\u3092\u4f7f\u3046\u3053\u3068\u3067\u3001\u4e00\u5ea6\u306b\u8907\u6570\u306e\u9023\u7d9a\u30c7\u30fc\u30bf\u3092\u8aad\u307f\u66f8\u304d\u3067\u304d\u307e\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u30e1\u30e2\u30ea\u30a2\u30af\u30bb\u30b9\u56de\u6570\u304c\u524a\u6e1b\u3055\u308c\u3001\u5e2f\u57df\u5e45\u306e\u5229\u7528\u52b9\u7387\u304c\u5411\u4e0a\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u6700\u9069\u5316\u3055\u308c\u305f\u30c7\u30fc\u30bf\u30d1\u30b9<\/strong><br>\u73fe\u4ee3\u306e\u30d7\u30ed\u30bb\u30c3\u30b5\u306b\u306f\u3001\u30d9\u30af\u30c8\u30eb\u8a08\u7b97\u5c02\u7528\u306b\u8a2d\u8a08\u3055\u308c\u305f\u30c7\u30fc\u30bf\u30d1\u30b9\u304c\u7d44\u307f\u8fbc\u307e\u308c\u3066\u3044\u307e\u3059\u3002\u3053\u308c\u3089\u306e\u30c7\u30fc\u30bf\u30d1\u30b9\u306f\u901a\u5e38\u3001\u9ad8\u3044\u30b9\u30eb\u30fc\u30d7\u30c3\u30c8\u3092\u6301\u3061\u3001\u30d9\u30af\u30c8\u30eb\u52a0\u7b97\u3084\u4e57\u7b97\u306a\u3069\u306e\u64cd\u4f5c\u306b\u7279\u5316\u3057\u305f\u6700\u9069\u5316\u304c\u65bd\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u305d\u306e\u7d50\u679c\u3001\u5f93\u6765\u306e\u30b9\u30ab\u30e9\u6f14\u7b97\u3088\u308a\u3082\u9ad8\u901f\u306b\u5b9f\u884c\u3067\u304d\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u5206\u5c90\u4e88\u6e2c\u30df\u30b9\u306e\u8efd\u6e1b<\/strong><br>\u975e\u30d9\u30af\u30c8\u30eb\u30b3\u30fc\u30c9\u3092\u5b9f\u884c\u3059\u308b\u5834\u5408\u3001\u5206\u5c90\u4e88\u6e2c\u30df\u30b9\u304c\u767a\u751f\u3059\u308b\u3068\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u304c\u505c\u6b62\u3057\u3001\u6027\u80fd\u304c\u4f4e\u4e0b\u3057\u307e\u3059\u3002\u4e00\u65b9\u3067\u3001\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u306f\u901a\u5e38\u3001\u5358\u7d14\u306a\u7b97\u8853\u8ad6\u7406\u64cd\u4f5c\u306b\u95a2\u4e0e\u3059\u308b\u305f\u3081\u3001\u8907\u96d1\u306a\u6761\u4ef6\u5224\u65ad\u3092\u4f34\u308f\u305a\u3001\u5206\u5c90\u4e88\u6e2c\u30df\u30b9\u306b\u3088\u308b\u6027\u80fd\u4f4e\u4e0b\u304c\u5c11\u306a\u304f\u306a\u308a\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u6027\uff08ILP\uff09\u306e\u5411\u4e0a<\/strong><br>\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u306f\u30011\u30af\u30ed\u30c3\u30af\u30b5\u30a4\u30af\u30eb\u5185\u306b\u3088\u308a\u591a\u304f\u306e\u64cd\u4f5c\u3092\u5b9f\u884c\u3059\u308b\u3053\u3068\u3067\u3001\u30d7\u30ed\u30bb\u30c3\u30b5\u306e\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u6027\u3092\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u3001\u540c\u6642\u306b\u3088\u308a\u591a\u304f\u306e\u8a08\u7b97\u30bf\u30b9\u30af\u3092\u51e6\u7406\u3067\u304d\u3001\u5168\u4f53\u7684\u306a\u6027\u80fd\u304c\u5411\u4e0a\u3057\u307e\u3059\u3002<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">SIMD\u547d\u4ee4\u306e\u30b5\u30dd\u30fc\u30c8<\/h4>\n\n\n\n<p>\u73fe\u4ee3\u306e\u30d7\u30ed\u30bb\u30c3\u30b5\u306f\u307b\u3068\u3093\u3069\u304cSIMD\u547d\u4ee4\u3092\u30b5\u30dd\u30fc\u30c8\u3057\u3066\u3044\u307e\u3059\u3002\u4f8b\u3048\u3070\u3001Intel\u306eSSE\u3084AVX\u3001ARM\u306eNEON\u3084SVE\u306a\u3069\u3067\u3059\u3002\u30d7\u30ed\u30bb\u30c3\u30b5\u304cSIMD\u547d\u4ee4\u3092\u3069\u306e\u3088\u3046\u306b\u30b5\u30dd\u30fc\u30c8\u3057\u3066\u3044\u308b\u304b\u306b\u3064\u3044\u3066\u306e\u8a73\u7d30\u306f\u3001\u30d7\u30ed\u30bb\u30c3\u30b5\u30e1\u30fc\u30ab\u30fc\u304c\u63d0\u4f9b\u3059\u308b\u30e6\u30fc\u30b6\u30fc\u30de\u30cb\u30e5\u30a2\u30eb\u3092\u53c2\u7167\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p>\u30c7\u30fc\u30bf\u4e26\u5217\u51e6\u7406\u3092\u6d3b\u7528\u3057\u305f\u9ad8\u901f\u5316\u306e\u4e00\u822c\u7684\u306a\u65b9\u6cd5\u306b\u306f\u4ee5\u4e0b\u304c\u3042\u308a\u307e\u3059\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u30eb\u30fc\u30d7\u5c55\u958b<\/strong><\/li>\n\n\n\n<li><strong>\u30d9\u30af\u30c8\u30eb\u547d\u4ee4\u306e\u5229\u7528<\/strong><\/li>\n\n\n\n<li><strong>\u5171\u6709\u30e1\u30e2\u30ea\u30d9\u30fc\u30b9\u306e\u4e26\u5217\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u3092\u5b9f\u73fe\u3059\u308bOpenMP\u30e9\u30a4\u30d6\u30e9\u30ea\u306e\u5229\u7528<\/strong><\/li>\n<\/ul>\n\n\n\n<p><strong>1. \u30eb\u30fc\u30d7\u5c55\u958b<\/strong><\/p>\n\n\n\n<p>\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3CPU\u306e\u6027\u80fd\u3092\u6700\u5927\u9650\u306b\u5f15\u304d\u51fa\u3059\u306b\u306f\u3001\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u306e\u505c\u6b62\u3092\u53ef\u80fd\u306a\u9650\u308a\u6e1b\u3089\u3057\u3001\u30d5\u30eb\u7a3c\u50cd\u3055\u305b\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002\u305d\u306e\u305f\u3081\u3001\u547d\u4ee4\u9593\u306e\u4e26\u5217\u6027\u3092\u5341\u5206\u306b\u6d3b\u7528\u3057\u3001\u4f9d\u5b58\u6027\u306e\u306a\u3044\u547d\u4ee4\u30b7\u30fc\u30b1\u30f3\u30b9\u3092\u898b\u3064\u3051\u3066\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3067\u91cd\u306d\u3066\u5b9f\u884c\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<p><strong>\u30eb\u30fc\u30d7\u5c55\u958b\uff08Loop Unrolling\uff09<\/strong>\u306f\u3001\u30eb\u30fc\u30d7\u672c\u4f53\u306e\u30b3\u30fc\u30c9\u3092\u8907\u6570\u56de\u30b3\u30d4\u30fc\u3057\u3066\u9806\u756a\u306b\u914d\u7f6e\u3057\u3001\u30eb\u30fc\u30d7\u7d42\u4e86\u6761\u4ef6\u3092\u8abf\u6574\u3059\u308b\u624b\u6cd5\u3067\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u3001\u30eb\u30fc\u30d7\u306e\u7570\u306a\u308b\u53cd\u5fa9\u9593\u3067\u5b58\u5728\u3059\u308b\u4e26\u5217\u6027\u3092\u5f15\u304d\u51fa\u305b\u307e\u3059\u3002\u30eb\u30fc\u30d7\u5c55\u958b\u5f8c\u3001\u30eb\u30fc\u30d7\u4f53\u306e\u30b3\u30fc\u30c9\u306f\u7d50\u5408\u3055\u308c\u3001\u30b3\u30f3\u30d1\u30a4\u30e9\u3084\u30d7\u30ed\u30bb\u30c3\u30b5\u306b\u3088\u308a\u591a\u304f\u306e\u30b9\u30b1\u30b8\u30e5\u30fc\u30ea\u30f3\u30b0\u304a\u3088\u3073\u6700\u9069\u5316\u306e\u4f59\u5730\u3092\u63d0\u4f9b\u3057\u307e\u3059\u3002\u307e\u305f\u3001\u30eb\u30fc\u30d7\u56de\u6570\u304c\u5927\u5e45\u306b\u6e1b\u5c11\u3059\u308b\u305f\u3081\u3001\u5206\u5c90\u547d\u4ee4\u3084\u30eb\u30fc\u30d7\u5236\u5fa1\u306b\u4f34\u3046\u30aa\u30fc\u30d0\u30fc\u30d8\u30c3\u30c9\u3082\u524a\u6e1b\u3055\u308c\u307e\u3059\u3002<\/p>\n\n\n\n<p><strong>2. AVX\u547d\u4ee4<\/strong><\/p>\n\n\n\n<p><strong>MMX\uff08MultiMedia eXtension\uff09<\/strong>\u3001<strong>SSE\uff08Streaming SIMD Extension\uff09<\/strong>\u3001\u304a\u3088\u3073<strong>AVX\uff08Advanced Vector Extensions\uff09<\/strong>\u306f\u3001Intel\u30d7\u30ed\u30bb\u30c3\u30b5\u3067\u4f7f\u7528\u3055\u308c\u308bSIMD\u547d\u4ee4\u30bb\u30c3\u30c8\u3067\u3059\u3002\u7279\u306bAVX\u306f\u3001MMX\u3084SSE\u547d\u4ee4\u30bb\u30c3\u30c8\u3092\u57fa\u306b\u3001128\u30d3\u30c3\u30c8\u306eSIMD\u30ec\u30b8\u30b9\u30bf\u3092256\u30d3\u30c3\u30c8\u4ee5\u4e0a\u306b\u62e1\u5f35\u3057\u3001\u975e\u69cb\u9020\u5316\u64cd\u4f5c\u30843\u30aa\u30da\u30e9\u30f3\u30c9\uff08\u5834\u5408\u306b\u3088\u3063\u3066\u306f4\u30aa\u30da\u30e9\u30f3\u30c9\uff09\u306e\u30b5\u30dd\u30fc\u30c8\u3092\u8ffd\u52a0\u3059\u308b\u3053\u3068\u3067\u3001\u547d\u4ee4\u306e\u67d4\u8edf\u6027\u3068\u6a5f\u80fd\u6027\u3092\u5411\u4e0a\u3055\u305b\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u307e\u305f\u3001AVX\u547d\u4ee4\u30bb\u30c3\u30c8\u306f\u3001\u4e57\u52a0\u878d\u5408\u547d\u4ee4\uff08Fused Multiply-Add, FMA\uff09\u3092\u30b5\u30dd\u30fc\u30c8\u3057\u3066\u304a\u308a\u30011\u3064\u306e\u547d\u4ee4\u3067<br>$$C += A \\times B$$<br>\u306e\u3088\u3046\u306a\u4e57\u52a0\u6f14\u7b97\u3092\u5b9f\u73fe\u3067\u304d\u307e\u3059\u3002<\/p>\n\n\n\n<p><strong>AVX\u306e\u74b0\u5883<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>16\u500b\u306e256\u30d3\u30c3\u30c8\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf\uff08YMM0\u301cYMM15\uff09\u3092\u6301\u3064\u3002<\/li>\n\n\n\n<li>1\u3064\u306eAVX\u30ec\u30b8\u30b9\u30bf\u306b\u30018\u500b\u306e32\u30d3\u30c3\u30c8\u5358\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\u3001\u307e\u305f\u306f4\u500b\u306e64\u30d3\u30c3\u30c8\u500d\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u70b9\u6570\u3092\u683c\u7d0d\u53ef\u80fd\u3002<\/li>\n\n\n\n<li>\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf\u5185\u306e\u5404\u30c7\u30fc\u30bf\u30d5\u30a3\u30fc\u30eb\u30c9\u306f\u300c\u8981\u7d20\uff08Element\uff09\u300d\u3068\u547c\u3070\u308c\u307e\u3059\uff08\u56f33-1\u53c2\u7167\uff09\u3002<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image size-full is-resized is-style-border\"><img decoding=\"async\" width=\"1972\" height=\"170\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-1.png\" alt=\"\" class=\"wp-image-1744\" style=\"width:617px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-1.png 1972w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-1-300x26.png 300w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-1-1024x88.png 1024w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-1-768x66.png 768w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-1-1536x132.png 1536w\" sizes=\"(max-width: 1972px) 100vw, 1972px\" \/><figcaption class=\"wp-element-caption\">\u56f33-1<\/figcaption><\/figure>\n\n\n\n<p><strong>AVX\u547d\u4ee4\u306e\u4f7f\u7528<\/strong><\/p>\n\n\n\n<p>AVX\u547d\u4ee4\u3092\u4f7f\u7528\u3057\u3066\u8a08\u7b97\u3092\u884c\u3046\u969b\u3001\u307e\u305a\u30c7\u30fc\u30bf\u3092\u30e1\u30e2\u30ea\u304b\u3089AVX\u30ec\u30b8\u30b9\u30bf\u306b\u30ed\u30fc\u30c9\u3057\u3001\u8a08\u7b97\u5f8c\u306b\u7d50\u679c\u3092\u518d\u3073\u30e1\u30e2\u30ea\u306b\u4fdd\u5b58\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<p><strong>AVX\u306e\u4ee3\u8868\u7684\u306a\u547d\u4ee4<\/strong><\/p>\n\n\n\n<figure class=\"wp-block-table\"><table class=\"has-fixed-layout\"><thead><tr><th>\u547d\u4ee4<\/th><th>\u8aac\u660e<\/th><th>\u4f7f\u7528\u4f8b<\/th><\/tr><\/thead><tbody><tr><td><code>vbroadcastss<\/code><br><code>vbroadcastsd<\/code><br><code>vbroadcastf128<\/code><\/td><td>\u30e1\u30e2\u30ea\u304b\u308932\/64\/128\u30d3\u30c3\u30c8\u306e\u30aa\u30da\u30e9\u30f3\u30c9\u3092\u8aad\u307f\u8fbc\u307f\u3001\u30d9\u30af\u30c8\u30eb\u30ec\u30b8\u30b9\u30bf\u5185\u306e\u3059\u3079\u3066\u306e\u8981\u7d20\u306b\u30d6\u30ed\u30fc\u30c9\u30ad\u30e3\u30b9\u30c8\u3059\u308b<\/td><td><code>vbroadcastss (%rax), %ymm0<\/code><br><br>rax\u306e\u30a2\u30c9\u30ec\u30b9\u3067\u306e\u6570\u5024\u3092ymm0\u30678\u3064\u540c\u3058\u5358\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u3068\u3057\u3066\u30b3\u30fc\u30d4<\/td><\/tr><tr><td><code>vmovups<\/code><br><code>vmovdqu<\/code><br><code>vmovapd<\/code><\/td><td>\u30e1\u30e2\u30ea\u3068AVX\u30ec\u30b8\u30b9\u30bf\u9593\u3067\u5358\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u3092\u975e\u30a2\u30e9\u30a4\u30f3\/\u500d\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u3092\u975e\u30a2\u30e9\u30a4\u30f3\/\u500d\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u3092\u30a2\u30e9\u30a4\u30f3\u3067\u79fb\u52d5\u3059\u308b<\/td><td><code>vmovups %ymm0, (%rax)<\/code><br><br>ymm0\u306e8\u3064\u5358\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570\u3092rax\u306e\u30a2\u30c9\u30ec\u30b9\u3078\u79fb\u52d5<\/td><\/tr><tr><td><code>vfmadd231ps<\/code><br><code>vaddps<\/code><br><code>vsubps<\/code><br><code>vmulps<\/code><br><code>vdivps<\/code><br><code>vsqrtps<\/code><\/td><td>\u5358\u7cbe\u5ea6\u6d6e\u52d5\u5c0f\u6570(s)\u3067256\u30d3\u30c3\u30c8\u306e\u4e57\u52a0\u6f14\u7b97\/\u52a0\/\u6e1b\/\u4e57\/\u9664\/\u30eb\u30fc\u30c8\u3092\u884c\u3046<\/td><td><code>vfmadd231ps %ymm0, %ymm1, %ymm2<\/code><br><br>ymm2 += ymm0 * ymm1<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<h4 class=\"wp-block-heading\">OpenMP<\/h4>\n\n\n\n<p><strong>OpenMP\uff08Open specifications for Multi Processing\uff09<\/strong>\u306f\u3001\u5171\u6709\u30e1\u30e2\u30ea\u65b9\u5f0f\u3067\u591a\u30b9\u30ec\u30c3\u30c9\u4e26\u5217\u51e6\u7406\u3092\u5b9f\u73fe\u3059\u308b\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u30e9\u30a4\u30d6\u30e9\u30ea\u3067\u3059\u3002OpenMP\u306f\u9ad8\u7d1a\u8a00\u8a9e\u306b\u57cb\u3081\u8fbc\u3080\u3053\u3068\u304c\u3067\u304d\u308b\u30b3\u30f3\u30d1\u30a4\u30eb\u6307\u793a\u3084\u30d7\u30ea\u30df\u30c6\u30a3\u30d6\u3092\u63d0\u4f9b\u3057\u3001\u30b7\u30ea\u30a2\u30eb\u30b3\u30fc\u30c9\u3092\u4e26\u5217\u5316\u3057\u307e\u3059\u3002OpenMP\u306f\u7c97\u7c92\u5ea6\u304b\u3089\u7d30\u7c92\u5ea6\u307e\u3067\u7570\u306a\u308b\u4e26\u5217\u5316\u3092\u30b5\u30dd\u30fc\u30c8\u3057\u307e\u3059\u3002\u307e\u305f\u3001OpenMP\u306f\u30b9\u30ec\u30c3\u30c9\u306e\u4f7f\u7528\u3092\u7c21\u7d20\u5316\u3057\u3001\u958b\u767a\u8005\u304c\u30b9\u30ec\u30c3\u30c9\u306e\u4f5c\u6210\u3001\u540c\u671f\u3001\u8ca0\u8377\u5206\u6563\u306a\u3069\u3092\u610f\u8b58\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u305b\u3093\u3002<\/p>\n\n\n\n<p><strong>OpenMP\u3092\u4f7f\u3063\u305f\u4e26\u5217\u5316\u306e\u57fa\u672c\u624b\u9806<\/strong><\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u8a08\u7b97\u30bf\u30b9\u30af\u306e\u5206\u5272<\/li>\n\n\n\n<li>\u30c7\u30fc\u30bf\u8ee2\u9001<\/li>\n\n\n\n<li>\u30b5\u30d6\u30bf\u30b9\u30af\u7d50\u679c\u306e\u7d71\u5408<\/li>\n\n\n\n<li>\u30b5\u30d6\u30bf\u30b9\u30af\u8a08\u7b97\u30ed\u30b8\u30c3\u30af\u306e\u5b9f\u88c5<\/li>\n\n\n\n<li>\u30b9\u30ec\u30c3\u30c9\u306e\u4f5c\u6210<\/li>\n<\/ol>\n\n\n\n<p>\u8a73\u7d30\u306f<a href=\"https:\/\/www.openmp.org\/\">OpenMP\u516c\u5f0f<\/a>\u30c9\u30ad\u30e5\u30e1\u30f3\u30c8\u3092\u3054\u53c2\u7167\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u5185\u5bb9<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li>\u30d7\u30ed\u30bb\u30c3\u30b5\u306eFPU\u3092\u5229\u7528\u3057\u3001\u30eb\u30fc\u30d7\u5c55\u958b\u3092\u7d44\u307f\u5408\u308f\u305b\u3066\u884c\u5217\u7a4d\u306e\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u3002<\/li>\n\n\n\n<li><strong>AVX\u547d\u4ee4<\/strong>\u3092\u57fa\u306b\u3001\u9ad8\u6027\u80fd\u306a\u884c\u5217\u7a4d\u8a08\u7b97\u30ab\u30fc\u30cd\u30eb\u3092\u8a2d\u8a08\u30fb\u5b9f\u88c5\u3059\u308b\u3002<\/li>\n\n\n\n<li><strong>OpenMP\u30e9\u30a4\u30d6\u30e9\u30ea<\/strong>\u3092\u6d3b\u7528\u3057\u3001\u4efb\u610f\u306e\u5f62\u72b6\u306e\u884c\u5217\u7a4d\u8a08\u7b97\u3092\u5b9f\u73fe\u3059\u308b\u3002<\/li>\n<\/ol>\n\n\n\n<h4 class=\"wp-block-heading\">x87 FPU\u3092\u57fa\u306b\u884c\u5217\u4e57\u7b97\u306e\u6027\u80fd\u3092\u6700\u9069\u5316\u3059\u308b<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\">\u30d7\u30ed\u30bb\u30c3\u30b5\u306eFPU\u3092\u5229\u7528\u3057\u3001\u30eb\u30fc\u30d7\u5c55\u958b\u3092\u7d44\u307f\u5408\u308f\u305b\u3066\u884c\u5217\u7a4d\u306e\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u3002<\/p>\n\n\n\n<p>\u6f14\u7b97\u30e6\u30cb\u30c3\u30c8\u306e\u7279\u6027\u3092\u6d3b\u7528\u3057\u305f\u30eb\u30fc\u30d7\u5c55\u958b\u624b\u6cd5\u3092\u7fd2\u5f97\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u30c7\u30fc\u30bf\u306e\u8aad\u307f\u53d6\u308a\u30fb\u4fdd\u5b58<\/strong>\uff1a<code>FLD<\/code>\u3001<code>FSTP<\/code> \u306a\u3069\u306e\u547d\u4ee4\u3092\u4f7f\u7528\u3002<\/li>\n\n\n\n<li><strong>\u8a08\u7b97<\/strong>\uff1a<code>FMUL<\/code>\u3001<code>FADDP<\/code> \u306a\u3069\u306ex87 FPU\u304c\u63d0\u4f9b\u3059\u308b\u547d\u4ee4\u3092\u4f7f\u7528\u3002<\/li>\n\n\n\n<li><strong>\u30eb\u30fc\u30d7\u5c55\u958b<\/strong>\uff1a\u6b21\u5143<code>N<\/code>\u306e\u30eb\u30fc\u30d7\u3092\u30b9\u30c6\u30c3\u30d7\u30b5\u30a4\u30ba2\u3067\u5c55\u958b\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u5b9f\u9a13\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3068\u89e3\u51cd<\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u5b9f\u884c\u3059\u308b\u524d\u306b\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u304c\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u306b\u6b63\u5e38\u306b\u63a5\u7d9a\u3055\u308c\u3066\u3044\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br><code>lab3.tar.gz<\/code> \u3092<a href=\"https:\/\/www.yanagichiaki.jp\/index.php\/2024\/12\/09\/computer-architecture-labs-guidebook-hitsz\/#index_id6\">\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9<\/a>\u3057\u3001\u30e6\u30fc\u30b6\u30fc\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u30b3\u30d4\u30fc\u3057\u3066\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<p><code>tar -zxvf lab3.tar.gz<\/code><\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><code>src\/lab3\/gemm_kernel_opt_loop_unrolling.S<\/code>\u3067\u8a08\u7b97\u30ed\u30b8\u30c3\u30af\u3092\u5b9f\u88c5<\/strong><\/div><div class=\"swell-block-step__body\">\n<ul class=\"wp-block-list\">\n<li>A[m][k] \u00d7 B[k][n+1] -&gt; st(0) \u306e\u8a08\u7b97\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/li>\n\n\n\n<li>C[m][n] \u3068 C[m][n+1] \u306e\u8aad\u307f\u8fbc\u307f\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/li>\n\n\n\n<li>C[m][n+1] + A[m][k] \u00d7 B[k][n+1]\u306e\u8a08\u7b97\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/li>\n\n\n\n<li>C[m][n] + A[m][k] \u00d7 B[k][n]\u306e\u8a08\u7b97\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/li>\n\n\n\n<li>C[m][n] \u306e\u4fdd\u5b58\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/li>\n\n\n\n<li>N\u6b21\u5143\u30eb\u30fc\u30d7\u306e\u66f4\u65b0\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab3_gemm_opt_loop_unrolling.unittest\n.\/dist\/bins\/lab3_gemm_opt_loop_unrolling.unittest<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong>\u6027\u80fd\u306e\u6bd4\u8f03\u3068\u5206\u6790<\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab3_gemm_opt_loop_unrolling\n.\/dist\/bins\/lab3_gemm_opt_loop_unrolling 256 256 256<\/code><\/pre>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u6700\u9069\u5316\u5f8c\u306e\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3068\u57fa\u6e96\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0(Baseline)\u3092\u6bd4\u8f03<\/strong>\u3057\u3001\u6027\u80fd\u3092\u6e2c\u5b9a\u3057\u307e\u3059\uff08\u30c6\u30b9\u30c8\u30b1\u30fc\u30b9\u306f\u6700\u9069\u5316\u306e\u5185\u5bb9\u306b\u5fdc\u3058\u3066\u9078\u629e\u3057\u3066\u304f\u3060\u3055\u3044\uff09\u3002<\/li>\n\n\n\n<li><strong>\u6027\u80fd\u7d50\u679c\u306b\u57fa\u3065\u304d\u3001\u6700\u9069\u5316\u624b\u6cd5\u306e\u52b9\u679c\u7bc4\u56f2\u3092\u5206\u6790<\/strong>\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">AVX\u547d\u4ee4\u306e(2m,32n,32k)\u9ad8\u6027\u80fd\u884c\u5217\u4e57\u7b97\u8a08\u7b97\u30ab\u30fc\u30cd\u30eb<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\"><strong>AVX\u547d\u4ee4<\/strong>\u3092\u57fa\u306b\u3001\u9ad8\u6027\u80fd\u306a\u884c\u5217\u7a4d\u8a08\u7b97\u30ab\u30fc\u30cd\u30eb\u3092\u8a2d\u8a08\u30fb\u5b9f\u88c5\u3059\u308b\u3002<\/p>\n\n\n\n<p>\u884c\u5217 $C$\u306e\u90e8\u5206\u884c\u5217 $C_{r}$\u200b\u306e\u8981\u7d20\u6570\u304c64\u4ee5\u4e0a\u5fc5\u8981\uff08\u3064\u307e\u308a$m_{r} \\cdot n_{r} \\geq 64$\uff09\u3002\u5404\u6b21\u5143\u306e\u8981\u4ef6\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>$m_{r} \\mod 2 = 0$<\/li>\n\n\n\n<li>$k_{r} \\mod 8 = 0$<\/li>\n\n\n\n<li>$n_{r} \\mod 8 = 0$<\/li>\n<\/ul>\n\n\n\n<p>\u30ab\u30fc\u30cd\u30eb\u306f\u3001\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u8a2d\u8a08\u3092\u63a1\u7528\u3057\u3001\u30c7\u30fc\u30bf\u306e\u8aad\u307f\u53d6\u308a\u3001\u8a08\u7b97\u3001\u66f8\u304d\u623b\u3057\u306e\u30d7\u30ed\u30bb\u30b9\u3092\u5b9f\u884c\u3059\u308b\u3002<\/p>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong>\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9<code>src\/lab3\/gemm_kernel_opt_avx.S<\/code>\u306e\u6982\u8981<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>src\/lab3\/gemm_kernel_opt_avx.S<\/code>\u306f\u3001\u4ee5\u4e0b\u306e\u56f3\u306b\u793a\u3059\u3088\u3046\u306a\u884c\u5217\u30d6\u30ed\u30c3\u30ad\u30f3\u30b0\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u4f7f\u7528\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized is-style-border\"><img decoding=\"async\" width=\"1724\" height=\"799\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-2.png\" alt=\"\" class=\"wp-image-1745\" style=\"width:663px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-2.png 1724w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-2-300x139.png 300w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-2-1024x475.png 1024w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-2-768x356.png 768w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-2-1536x712.png 1536w\" sizes=\"(max-width: 1724px) 100vw, 1724px\" \/><\/figure>\n\n\n\n<p>\u6700\u5185\u5c64\u30eb\u30fc\u30d7\u3067\u306f\u3001\u4ee5\u4e0b\u3092\u8a08\u7b97\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>$2 \\times 1$ \u306e\u30d6\u30ed\u30c3\u30af $A[m:m+2][k]$<\/li>\n\n\n\n<li>$1 \\times 32$ \u306e\u30d6\u30ed\u30c3\u30af $B[k][n:n+32]$<\/li>\n\n\n\n<li>\u7d50\u679c\u3068\u3057\u3066 $2 \\times 32$ \u306e\u30d6\u30ed\u30c3\u30af $C[m:m+2][n:n+32]$<\/li>\n<\/ul>\n\n\n\n<p>\u6700\u5185\u5c64\u306e\u30d6\u30ed\u30c3\u30af\u884c\u5217\u7a4d\u306f\u3001<strong>AVX\u30d9\u30af\u30c8\u30eb\u547d\u4ee4<\/strong>\u3092\u7528\u3044\u3066\u4e26\u5217\u8a08\u7b97\u3055\u308c\u307e\u3059\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong>\u30b3\u30fc\u30c9<code>src\/lab3\/gemm_kernel_opt_avx.S<\/code>\u3067AVX<\/strong><\/strong>\u547d\u4ee4\u3092\u8ffd\u52a0\u3057\u3066\u6700\u9069\u5316<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<ul class=\"wp-block-list\">\n<li>LOAD_MAT_C\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0: \u884c\u5217 $C[m:m+2][n:n+32]$ \u3092AVX\u30ec\u30b8\u30b9\u30bf\u306b\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>LOAD_MAT_A\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0: \u884c\u5217 $A[m+1][k]$ \u3092AVX\u30ec\u30b8\u30b9\u30bf\u306b\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>LOAD_MAT_B\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0: \u884c\u5217 $B[k][n:n+32]$ \u3092AVX\u30ec\u30b8\u30b9\u30bf\u306b\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>DO_COMPUTE\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0: $C[m:m+2][n:n+32]+=A[m:m+2][k]\u00d7B[k:k+8][n:n+32]$ \u3092\u8a08\u7b97\u3002<\/li>\n\n\n\n<li>STORE_MAT_C\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0: \u884c\u5217 $C[m][n:n+32]$ \u3068 $C[m+1][n:n+32]$ \u3092\u4fdd\u5b58\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u4ee5\u4e0b\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab3_gemm_opt_avx.unittest\n.\/dist\/bins\/lab3_gemm_opt_avx.unittest<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong>\u6027\u80fd\u306e\u6bd4\u8f03\u3068\u5206\u6790<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab3_gemm_opt_avx\n.\/dist\/bins\/lab3_gemm_opt_avx 256 256 256<\/code><\/pre>\n\n\n\n<p><strong>\u6700\u9069\u5316\u5f8c\u306e\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3068\u57fa\u6e96\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0(Baseline)\u3092\u6bd4\u8f03<\/strong>\u3057\u3001\u6027\u80fd\u3092\u6e2c\u5b9a\u3057\u307e\u3059\uff08\u30c6\u30b9\u30c8\u30b1\u30fc\u30b9\u306f\u6700\u9069\u5316\u306e\u5185\u5bb9\u306b\u5fdc\u3058\u3066\u9078\u629e\u3057\u3066\u304f\u3060\u3055\u3044\uff09\u3002<br><strong>\u6027\u80fd\u7d50\u679c\u306b\u57fa\u3065\u304d\u3001\u6700\u9069\u5316\u624b\u6cd5\u306e\u52b9\u679c\u7bc4\u56f2\u3092\u5206\u6790<\/strong>\u3057\u307e\u3059\u3002<\/p>\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">OpenMP\u3068AVX\u547d\u4ee4\u3067\u4efb\u610f\u5f62\u72b6\u306e\u884c\u5217\u4e57\u7b97\u3092\u5b9f\u73fe\u3059\u308b<\/h4>\n\n\n\n<p class=\"is-style-big_icon_hatena\">\u3053\u306e\u90e8\u5206\u306f\u3001\u304a\u597d\u304d\u306a\u3088\u3046\u306b\u9078\u3093\u3067\u3001\u5b9f\u88c5\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p class=\"is-style-icon_pen\"><strong>OpenMP\u30e9\u30a4\u30d6\u30e9\u30ea<\/strong>\u3092\u6d3b\u7528\u3057\u3001\u4efb\u610f\u306e\u5f62\u72b6\u306e\u884c\u5217\u7a4d\u8a08\u7b97\u3092\u5b9f\u73fe\u3059\u308b\u3002<\/p>\n\n\n\n<p>\u884c\u5217\u7a4d\u304c\u30de\u30eb\u30c1\u30b9\u30ec\u30c3\u30c9\u3067\u4ee5\u4e0b\u306e\u5f62\u72b6\u306b\u5bfe\u5fdc\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>$M \\geq 2$, $K \\geq 8$, $N \\geq 8$<\/li>\n<\/ul>\n\n\n\n<p>\u5206\u5272\u5f8c\u306b\u7df4\u7fd22\u30ab\u30fc\u30cd\u30eb\u306e\u8981\u4ef6\u3092\u6e80\u305f\u3055\u306a\u3044\u30c7\u30fc\u30bf\u30d6\u30ed\u30c3\u30af\u306b\u5bfe\u3057\u3066\u306f\u3001<strong>Padding<\/strong>\u307e\u305f\u306f\u5225\u9014\u8a2d\u8a08\u3057\u305f\u30ab\u30fc\u30cd\u30eb\u3067\u8a08\u7b97\u3092\u5b9f\u65bd\u3059\u308b\u3053\u3068\u304c\u53ef\u80fd\u3067\u3059\u3002<br>\u5404\u6b21\u5143\u306e\u4e26\u5217\u5316\u6226\u7565\u3092\u72ec\u81ea\u306b\u8a2d\u8a08\u3057\u307e\u3059\u3002\u305f\u3060\u3057\u3001\u4f7f\u7528\u3059\u308b\u30b9\u30ec\u30c3\u30c9\u6570\u306f\u95a2\u6570\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u3067\u6307\u5b9a\u3055\u308c\u305f\u6700\u5927\u30b9\u30ec\u30c3\u30c9\u6570\u3092\u8d85\u3048\u3066\u306f\u3044\u3051\u307e\u305b\u3093\u3002<br>\u8a08\u7b97\u6027\u80fd\u306f\u30d9\u30fc\u30b9\u30e9\u30a4\u30f3\u3092\u4e0a\u56de\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u7570\u306a\u308b\u30ec\u30d9\u30eb\u306e\u4e26\u5217\u5316\u6226\u7565\u3092\u6d3b\u7528\u3057\u3001\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u624b\u6cd5\u3092\u5b66\u3073\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u672c\u7df4\u7fd2\u306f\u4efb\u610f\u9078\u629e\u3067\u3059\u3002\u5177\u4f53\u7684\u306a\u5b9f\u88c5\u5185\u5bb9\u3084\u52a0\u901f\u52b9\u679c\u306b\u57fa\u3065\u304d\u52a0\u70b9\u8a55\u4fa1\u3055\u308c\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p class=\"is-style-icon_announce\">\u3053\u306e\u7df4\u7fd2\u3067\u884c\u3046\u4e26\u5217\u5316\u306e\u6700\u9069\u5316\u65b9\u6cd5\u306f\u3001\u81ea\u7531\u5ea6\u306e\u9ad8\u3044\u5185\u5bb9\u3067\u3042\u308a\u3001\u53b3\u5bc6\u306a\u610f\u5473\u3067\u306e\u6700\u9069\u89e3\u306f\u3042\u308a\u307e\u305b\u3093\u3002<\/p>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong>\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9<code>src\/lab3\/openmp_gemm_baseline.cpp<\/code>\u306e\u6982\u8981<\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>src\/lab3\/openmp_gemm_baseline.cpp<\/code>\u306f\u3001\u4ee5\u4e0b\u306b\u793a\u3059\u3088\u3046\u306a\u30de\u30eb\u30c1\u30b9\u30ec\u30c3\u30c9\u884c\u5217\u30d6\u30ed\u30c3\u30ad\u30f3\u30b0\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u4f7f\u7528\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized\"><img decoding=\"async\" width=\"1976\" height=\"615\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-3.png\" alt=\"\" class=\"wp-image-1746\" style=\"width:1038px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-3.png 1976w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-3-300x93.png 300w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-3-1024x319.png 1024w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-3-768x239.png 768w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-3-3-1536x478.png 1536w\" sizes=\"(max-width: 1976px) 100vw, 1976px\" \/><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>6\u884c\u76ee\uff5e13\u884c\u76ee<\/strong>: <code>get_parallel_thread_num<\/code>\u95a2\u6570\u306f\u3001\u6700\u5927\u30b9\u30ec\u30c3\u30c9\u6570<code>max_threads<\/code>\u3092\u4f7f\u7528\u3057\u3066\u3001\u884c\u5217A\u306e\u884c\u30d6\u30ed\u30c3\u30af\u6570<code>m_thread<\/code>\u304a\u3088\u3073\u884c\u5217B\u306e\u5217\u30d6\u30ed\u30c3\u30af\u6570<code>n_thread<\/code>\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002\u884c\u5217A\u306f\u6700\u59272\u3064\u306e\u30d6\u30ed\u30c3\u30af\u306b\u5206\u5272\u3055\u308c\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>15\u884c\u76ee<\/strong>: <code>openmp_gemm_baseline<\/code>\u95a2\u6570\u306e<code>thread_num<\/code>\u30d1\u30e9\u30e1\u30fc\u30bf\u306f\u3001\u8a08\u7b97\u306b\u4f7f\u7528\u3059\u308b\u6700\u5927\u30b9\u30ec\u30c3\u30c9\u6570\u3092\u8868\u3057\u307e\u3059\uff08\u30b3\u30de\u30f3\u30c9\u30e9\u30a4\u30f3\u5f15\u6570\u3067\u6307\u5b9a\u53ef\u80fd\uff09\u3002<\/li>\n\n\n\n<li><strong>22\u884c\u76ee\uff5e26\u884c\u76ee<\/strong>: OpenMP\u306e\u30c7\u30a3\u30ec\u30af\u30c6\u30a3\u30d6\u3092\u4f7f\u7528\u3057\u3066\u3001\u30b9\u30ec\u30c3\u30c9\u6570\u3001\u5171\u6709\u5909\u6570\uff08\u884c\u5217C\uff09\u3001\u30d7\u30e9\u30a4\u30d9\u30fc\u30c8\u5909\u6570\u306a\u3069\u306e\u60c5\u5831\u3092\u8a2d\u5b9a\u3057\u307e\u3059\u3002\n<ul class=\"wp-block-list\">\n<li><strong>27\u884c\u76ee\uff5e100\u884c\u76ee<\/strong>: \u5404\u30b9\u30ec\u30c3\u30c9\u304c\u5b9f\u884c\u3059\u308b\u30b3\u30fc\u30c9\u3067\u3059\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>29\u884c\u76ee<\/strong>: \u30b9\u30ec\u30c3\u30c9ID\u3092\u53d6\u5f97\u3057\u307e\u3059\uff08\u6700\u521d\u306e\u30b9\u30ec\u30c3\u30c9ID\u306f0\u3001\u6b21\u306f1\u3068\u9806\u306b\u5272\u308a\u632f\u3089\u308c\u307e\u3059\uff09\u3002<\/li>\n\n\n\n<li><strong>31\u884c\u76ee\uff5e32\u884c\u76ee<\/strong>: \u30b9\u30ec\u30c3\u30c9ID\u3092\u3001\u884c\u5217A\u306e\u884c\u30d6\u30ed\u30c3\u30af\u756a\u53f7<code>thread_id_m<\/code>\u3068\u884c\u5217B\u306e\u5217\u30d6\u30ed\u30c3\u30af\u756a\u53f7<code>thread_id_n<\/code>\u306b\u30de\u30c3\u30d4\u30f3\u30b0\u3057\u307e\u3059\u3002\n<ul class=\"wp-block-list\">\n<li>\u4f8b\u3048\u3070\u3001ID\u304c6\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u884c\u30d6\u30ed\u30c3\u30af\u756a\u53f71\u3068\u5217\u30d6\u30ed\u30c3\u30af\u756a\u53f72\u306b\u5bfe\u5fdc\u3059\u308b\u5834\u5408\u3001\u30b9\u30ec\u30c3\u30c96\u306f\u884c\u5217A\u306e1\u756a\u76ee\u306e\u884c\u30d6\u30ed\u30c3\u30af\u3068\u884c\u5217B\u306e2\u756a\u76ee\u306e\u5217\u30d6\u30ed\u30c3\u30af\u306e\u884c\u5217\u7a4d\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>36\u884c\u76ee\uff5e41\u884c\u76ee<\/strong>: \u884c\u5217A\u306e\u5404\u884c\u30d6\u30ed\u30c3\u30af\u306e\u884c\u6570<code>dim_m_per_thread<\/code>\u3001\u884c\u5217B\u306e\u5404\u5217\u30d6\u30ed\u30c3\u30af\u306e\u5217\u6570<code>dim_n_per_thread<\/code>\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>48\u884c\u76ee\uff5e52\u884c\u76ee<\/strong>: \u884c\u5217A\u306e<code>thread_id_m<\/code>\u756a\u76ee\u306e\u884c\u30d6\u30ed\u30c3\u30af\u306e\u958b\u59cb\u884c\u756a\u53f7<code>thread_m_start<\/code>\u3068\u7d42\u4e86\u884c\u756a\u53f7<code>thread_m_end<\/code>\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>55\u884c\u76ee\uff5e59\u884c\u76ee<\/strong>: \u884c\u5217B\u306e<code>thread_id_n<\/code>\u756a\u76ee\u306e\u5217\u30d6\u30ed\u30c3\u30af\u306e\u958b\u59cb\u5217\u756a\u53f7<code>thread_n_start<\/code>\u3068\u7d42\u4e86\u5217\u756a\u53f7<code>thread_n_end<\/code>\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>62\u884c\u76ee\uff5e67\u884c\u76ee<\/strong>: \u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba\u306b\u5fdc\u3058\u305f\u30e1\u30e2\u30ea\u3092\u5272\u308a\u5f53\u3066\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>70\u884c\u76ee\uff5e84\u884c\u76ee<\/strong>: \u30b9\u30ec\u30c3\u30c9ID\u306b\u5bfe\u5fdc\u3059\u308b\u884c\u30d6\u30ed\u30c3\u30af\u756a\u53f7<code>thread_id_m<\/code>\u3068\u5217\u30d6\u30ed\u30c3\u30af\u756a\u53f7<code>thread_id_n<\/code>\u3092\u57fa\u306b\u3001\u5bfe\u5fdc\u3059\u308b\u30d6\u30ed\u30c3\u30af\u30c7\u30fc\u30bf\u3092\u5143\u306e\u884c\u5217A\u3001B\u3001C\u304b\u3089\u4e00\u6642\u884c\u5217<code>A_padding<\/code>\u3001<code>B_padding<\/code>\u3001<code>C_padding<\/code>\u306b\u30b3\u30d4\u30fc\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>87\u884c\u76ee\uff5e95\u884c\u76ee<\/strong>: \u7df4\u7fd22\u3067\u5b9f\u88c5\u3057\u305f<code>gemm_kernel_opt_avx<\/code>\u30ab\u30fc\u30cd\u30eb\u3092\u547c\u3073\u51fa\u3057\u3001\u30d6\u30ed\u30c3\u30af\u884c\u5217\u7a4d\u3092\u8a08\u7b97\u3057\u3001\u8a08\u7b97\u7d50\u679c\u3092\u5143\u306e\u884c\u5217C\u306b\u4fdd\u5b58\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong>\u30b3\u30fc\u30c9<code>src\/lab3\/openmp_gemm_opt.cpp<\/code><\/strong><\/strong><\/strong>\u3067\u30ed\u30b8\u30c3\u30af\u3092\u5b9f\u88c5<\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p><strong>OpenMP\u30e9\u30a4\u30d6\u30e9\u30ea<\/strong>\u3092\u5229\u7528\u3057\u3066\u3001\u30de\u30eb\u30c1\u30b3\u30a2\u306b\u5bfe\u5fdc\u3057\u3001\u3088\u308a\u5927\u898f\u6a21\u306a\u884c\u5217\u7a4d\u8a08\u7b97\u3092\u30b5\u30dd\u30fc\u30c8\u3059\u308b\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u8a2d\u8a08\u3057\u307e\u3059\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u306e\u30eb\u30fc\u30c8\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u3067\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab3_gemm_opt_openmp.unittest\n.\/dist\/bins\/lab3_gemm_opt_openmp.unittest<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number\" style=\"background-color:var(--color_main)\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong>\u6027\u80fd\u306e\u6bd4\u8f03\u3068\u5206\u6790<\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>mkdir -p build &amp;&amp; cd build\ncmake -B . -S ..\/ &amp;&amp; cmake --build .\/ --target lab3_gemm_opt_openmp\n.\/dist\/bins\/lab3_gemm_opt_openmp 8 512 512 512<\/code><\/pre>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u6700\u9069\u5316\u5f8c\u306e\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3068\u57fa\u6e96\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0(Baseline)\u3092\u6bd4\u8f03<\/strong>\u3057\u3001\u6027\u80fd\u3092\u6e2c\u5b9a\u3057\u307e\u3059\uff08\u30c6\u30b9\u30c8\u30b1\u30fc\u30b9\u306f\u6700\u9069\u5316\u306e\u5185\u5bb9\u306b\u5fdc\u3058\u3066\u9078\u629e\u3057\u3066\u304f\u3060\u3055\u3044\uff09\u3002<\/li>\n\n\n\n<li><strong>\u6027\u80fd\u7d50\u679c\u306b\u57fa\u3065\u304d\u3001\u6700\u9069\u5316\u624b\u6cd5\u306e\u52b9\u679c\u7bc4\u56f2\u3092\u5206\u6790<\/strong>\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading\">\u30ec\u30dd\u30fc\u30c8\u306e\u8981\u4ef6<\/h3>\n\n\n\n<p>Lab1\uff5eLab3\u3092\u5b8c\u4e86\u3057\u3001\u4ee5\u4e0b\u306e\u8981\u4ef6\u306b\u5f93\u3063\u3066\u30ec\u30dd\u30fc\u30c8\u3092\u4f5c\u6210\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br>\u30ec\u30dd\u30fc\u30c8\u306e\u5f62\u5f0f\u306b\u306f\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u306f\u3042\u308a\u307e\u305b\u3093\u304c\u3001\u6b21\u306e\u5185\u5bb9\u3092\u542b\u3081\u308b\u3088\u3046\u306b\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>Labs\u306e\u30d7\u30ed\u30bb\u30b9<\/strong><\/p>\n\n\n\n<p>\u56f3\u8868\u3092\u7d44\u307f\u5408\u308f\u305b\u3001\u6587\u7ae0\u5f62\u5f0f\u3067\u8868\u73fe\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>\u30c6\u30b9\u30c8\u7d50\u679c\u3068\u539f\u7406\u306e\u5206\u6790<\/strong><\/p>\n\n\n\n<p>\u56f3\u8868\u3001\u6587\u7ae0\u3001\u30bd\u30fc\u30b9\u30b3\u30fc\u30c9\u3092\u7528\u3044\u3066\u4ee5\u4e0b\u3092\u6bd4\u8f03\u5206\u6790\u3057\u3066\u304f\u3060\u3055\u3044\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u30ad\u30e3\u30c3\u30b7\u30e5\u6700\u9069\u5316<\/strong>\u3068<strong>\u30c7\u30fc\u30bf\u4e26\u5217\u5316\u6700\u9069\u5316<\/strong>\u306e\u52b9\u679c\u3092\u6bd4\u8f03\u3059\u308b<\/li>\n\n\n\n<li>\u305d\u306e\u539f\u7406\u3092\u8aac\u660e\u3059\u308b<\/li>\n<\/ul>\n\n\n\n<p>\u30ec\u30dd\u30fc\u30c8\u3092<strong>PDF\u5f62\u5f0f<\/strong>\u3067\u3001\u5bbf\u984c\u30b7\u30b9\u30c6\u30e0\u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u30bd\u30fc\u30b9\u30d5\u30a1\u30a4\u30eb\uff083\u3064\u304b\u30895\u3064\uff09\u3092<code>.zip<\/code>\u5f62\u5f0f\u306b\u5727\u7e2e\u3057\u3001\u5bbf\u984c\u30b7\u30b9\u30c6\u30e0\u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>Lab2<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u7df4\u7fd22\uff1a<code>gemm_kernel_opt_prefetch.S<\/code><\/li>\n\n\n\n<li>\u7df4\u7fd23\uff08\u3042\u308b\u306a\u3089\uff09\uff1a<code>gemm_kernel_opt_loop.S<\/code><\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>Lab3<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li>\u7df4\u7fd21\uff1a<code>gemm_kernel_opt_loop_unrolling.S<\/code><\/li>\n\n\n\n<li>\u7df4\u7fd22\uff1a<code>gemm_kernel_opt_avx.S<\/code><\/li>\n\n\n\n<li>\u7df4\u7fd23\uff08\u3042\u308b\u306a\u3089\uff09\uff1a<code>openmp_gemm_opt.cpp<\/code><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">\u6b21\u306e\u5b9f\u9a13\u306e\u4e88\u7fd2<\/h3>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u8cc7\u6599\u3092\u8aad\u307f\u3001CUDA\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u3084GPU\u306e\u7279\u6027\u3092\u7406\u89e3\u3057\u3066\u304f\u3060\u3055\u3044\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/docs.nvidia.com\/cuda\/cuda-c-programming-guide\/\">CUDA C++ Programming Guide<\/a><\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">Lab4 CUDA\uff1aGPU\u306e\u884c\u5217\u4e57\u7b97<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u76ee\u7684<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>C\/CUDA\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0<\/strong>\u3092\u4f7f\u7528\u3057\u3066\u884c\u5217\u7a4d\u3092\u5b9f\u88c5\u3059\u308b\u65b9\u6cd5\u3092\u7fd2\u5f97\u3059\u308b\u3002<\/li>\n\n\n\n<li>GPU\u306e\u30de\u30eb\u30c1\u30b9\u30ec\u30c3\u30c9\u4e26\u5217\u8a08\u7b97\u80fd\u529b\u3092\u6d3b\u7528\u3057\u3001\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<\/li>\n\n\n\n<li><strong>CUDA\u306e\u57fa\u672c\u7684\u306a\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u30e2\u30c7\u30eb<\/strong>\u3068<strong>\u30b9\u30ec\u30c3\u30c9\u30b9\u30b1\u30b8\u30e5\u30fc\u30ea\u30f3\u30b0\u624b\u6cd5<\/strong>\u3092\u7fd2\u5f97\u3057\u3001GPU\u4e0a\u3067\u52b9\u7387\u7684\u306a\u884c\u5217\u6f14\u7b97\u3092\u5b9f\u73fe\u3059\u308b\u3053\u3068\u3067\u3001\u8a08\u7b97\u30c7\u30d0\u30a4\u30b9\u306e\u4e26\u5217\u51e6\u7406\u80fd\u529b\u3092\u3055\u3089\u306b\u5f15\u304d\u51fa\u3059\u65b9\u6cd5\u3092\u7406\u89e3\u3059\u308b\u3002<\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u524d\u306e\u6e96\u5099<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ee5\u4e0b\u306e\u8cc7\u6599\u3092\u8aad\u307f\u3001CUDA\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u306e\u57fa\u672c\u539f\u7406\u3068GPU\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306e\u7279\u6027\u3092\u7406\u89e3\u3057\u3066\u3001\u9ad8\u52b9\u7387\u306a\u884c\u5217\u7a4d\u8a08\u7b97\u3092\u5b9f\u73fe\u3057\u3066\u304f\u3060\u3055\u3044\u3002\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/docs.nvidia.com\/cuda\/cuda-c-programming-guide\/\">CUDA C++ Programming Guide<\/a><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u539f\u7406<\/h3>\n\n\n\n<p>\u3053\u306eLab\u3067\u306f\u3001CUDA\uff08Compute Unified Device Architecture\uff09\u3092\u4f7f\u7528\u3057\u305f\u4e26\u5217\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u3092\u5b66\u3073\u3001\u884c\u5217\u7a4d\u8a08\u7b97\u3092\u9ad8\u901f\u5316\u3057\u307e\u3059\u3002CUDA\u306fNVIDIA\u304c\u63d0\u4f9b\u3059\u308b\u4e26\u5217\u8a08\u7b97\u30d7\u30e9\u30c3\u30c8\u30d5\u30a9\u30fc\u30e0\u304a\u3088\u3073\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u30e2\u30c7\u30eb\u3067\u3042\u308a\u3001NVIDIA GPU\u306e\u8a08\u7b97\u80fd\u529b\u3092\u6d3b\u7528\u3057\u3066\u5927\u898f\u6a21\u306a\u4e26\u5217\u8a08\u7b97\u30bf\u30b9\u30af\u3092\u51e6\u7406\u3067\u304d\u307e\u3059\u3002CUDA\u3092\u4f7f\u7528\u3059\u308b\u3053\u3068\u3067\u3001\u884c\u5217\u7a4d\u306e\u8a08\u7b97\u6027\u80fd\u3092\u5927\u5e45\u306b\u5411\u4e0a\u3055\u305b\u3001\u5927\u898f\u6a21\u306a\u884c\u5217\u3092\u52b9\u7387\u7684\u306b\u51e6\u7406\u3059\u308b\u3053\u3068\u304c\u53ef\u80fd\u306b\u306a\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">\u884c\u5217\u7a4d\u306e\u6982\u8981<\/h4>\n\n\n\n<p>2\u3064\u306e\u884c\u5217 $A$ \u3068 $B$ \u306e\u7a4d\u884c\u5217 $C$ \u306e\u5404\u8981\u7d20\u306f\u3001\u4ee5\u4e0b\u306e\u5f0f\u3067\u8a08\u7b97\u3055\u308c\u307e\u3059\u3002$$C_{ij} = \\sum_{k=1}^n A_{ik} \\cdot B_{kj}$$<\/p>\n\n\n\n<p>\u3053\u3053\u3067\u3001$C_{ij}$ \u306f\u884c\u5217 $C$ \u306e $i$ \u884c $j$ \u5217\u306e\u8981\u7d20\u3092\u8868\u3057\u307e\u3059\u3002\u884c\u5217 $C$ \u306e\u5404\u8981\u7d20\u3092\u8a08\u7b97\u3059\u308b\u305f\u3081\u306b\u306f\u3001\u884c\u5217 $A$ \u306e\u884c\u3068\u884c\u5217 $B$ \u306e\u5217\u3092\u9806\u306b\u8d70\u67fb\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u3001\u305d\u306e\u8a08\u7b97\u91cf\u306e\u6642\u9593\u8a08\u7b97\u91cf\u306f $O(m \\times n \\times p)$ \u3067\u3059\u3002\u5927\u898f\u6a21\u306a\u884c\u5217\u306e\u5834\u5408\u3001\u8a08\u7b97\u91cf\u304c\u975e\u5e38\u306b\u591a\u304f\u306a\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">CUDA\u306b\u3088\u308b\u884c\u5217\u7a4d\u8a08\u7b97\u306e\u9ad8\u901f\u5316\u539f\u7406<\/h4>\n\n\n\n<p>CUDA\u306fGPU\u306e\u5927\u898f\u6a21\u4e26\u5217\u8a08\u7b97\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u57fa\u3065\u3044\u3066\u304a\u308a\u3001\u8907\u6570\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u5354\u529b\u3057\u3066\u52b9\u7387\u7684\u306b\u30c7\u30fc\u30bf\u3092\u51e6\u7406\u3057\u307e\u3059\u3002\u4ee5\u4e0b\u306fCUDA\u3092\u4f7f\u7528\u3057\u305f\u884c\u5217\u7a4d\u9ad8\u901f\u5316\u306e\u4e3b\u306a\u539f\u7406\u3067\u3059\u3002<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u591a\u30b9\u30ec\u30c3\u30c9\u306b\u3088\u308b\u4e26\u5217\u8a08\u7b97<\/strong><br>CUDA\u306fGPU\u4e0a\u306e\u6570\u5343\u304b\u3089\u6570\u4e07\u306e\u4e26\u5217\u30b9\u30ec\u30c3\u30c9\u3092\u4f7f\u7528\u3057\u3066\u8a08\u7b97\u3092\u9ad8\u901f\u5316\u3057\u307e\u3059\u3002\u884c\u5217\u7a4d\u3067\u306f\u5404\u8981\u7d20\u306e\u8a08\u7b97\u304c\u76f8\u4e92\u306b\u72ec\u7acb\u3057\u3066\u3044\u308b\u305f\u3081\u3001CUDA\u306f\u5404\u8a08\u7b97\u30bf\u30b9\u30af\u30921\u3064\u306e\u30b9\u30ec\u30c3\u30c9\u306b\u5272\u308a\u5f53\u3066\u3001\u540c\u6642\u306b\u8907\u6570\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u8a08\u7b97\u3092\u5b9f\u884c\u3059\u308b\u3053\u3068\u3067\u3001\u8a08\u7b97\u52b9\u7387\u3092\u5927\u5e45\u306b\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30d6\u30ed\u30c3\u30af\u3068\u30b0\u30ea\u30c3\u30c9\u69cb\u9020<\/strong><br>CUDA\u3067\u306f\u3001\u30b9\u30ec\u30c3\u30c9\u306f\u300c\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u300d\u3068\u300c\u30b0\u30ea\u30c3\u30c9\u300d\u306b\u3088\u3063\u3066\u7de8\u6210\u3055\u308c\u307e\u3059\u3002\u5404\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u306f\u884c\u5217\u306e\u4e00\u90e8\u3092\u62c5\u5f53\u3057\u3001\u4e2d\u9593\u8a08\u7b97\u7d50\u679c\u3092\u5171\u6709\u30e1\u30e2\u30ea\u306b\u4fdd\u5b58\u3057\u307e\u3059\u3002\u3053\u306e\u968e\u5c64\u69cb\u9020\u306b\u3088\u308a\u3001\u30e1\u30e2\u30ea\u7ba1\u7406\u3068\u4e26\u5217\u51e6\u7406\u304c\u52b9\u7387\u5316\u3055\u308c\u307e\u3059\u3002\u305f\u3068\u3048\u3070\u3001\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e\u5404\u30b9\u30ec\u30c3\u30c9\u306f\u7d50\u679c\u884c\u5217\u306e1\u3064\u306e\u8981\u7d20\u8a08\u7b97\u3092\u62c5\u5f53\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30e1\u30e2\u30ea\u9045\u5ef6\u306e\u96a0\u853d<\/strong><br>GPU\u306f\u975e\u5e38\u306b\u591a\u304f\u306e\u30b9\u30ec\u30c3\u30c9\u3092\u6301\u3061\u3001\u305d\u308c\u3089\u304c\u5207\u308a\u66ff\u308f\u308b\u3053\u3068\u3067\u30e1\u30e2\u30ea\u9045\u5ef6\u3092\u96a0\u853d\u3057\u307e\u3059\u3002\u4e00\u90e8\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u30e1\u30e2\u30ea\u8aad\u307f\u53d6\u308a\u3067\u30d6\u30ed\u30c3\u30af\u3055\u308c\u3066\u3044\u308b\u9593\u3001\u4ed6\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u8a08\u7b97\u3092\u7d9a\u884c\u3059\u308b\u305f\u3081\u3001\u8a08\u7b97\u30e6\u30cb\u30c3\u30c8\u306e\u5229\u7528\u7387\u3092\u9ad8\u3081\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u547d\u4ee4\u30ec\u30d9\u30eb\u4e26\u5217\u6027\uff08ILP\uff09<\/strong><br>CUDA\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306f\u3001\u5404\u30af\u30ed\u30c3\u30af\u30b5\u30a4\u30af\u30eb\u3067\u8907\u6570\u306e\u547d\u4ee4\u3092\u5b9f\u884c\u3067\u304d\u307e\u3059\u3002\u547d\u4ee4\u30d1\u30a4\u30d7\u30e9\u30a4\u30f3\u3092\u4e26\u5217\u5b9f\u884c\u7d4c\u8def\u306b\u6700\u9069\u5316\u3059\u308b\u3053\u3068\u3067\u3001\u884c\u5217\u7a4d\u306a\u3069\u306e\u8a08\u7b97\u30bf\u30b9\u30af\u3092\u52a0\u901f\u3057\u3001\u540c\u6642\u306b\u3088\u308a\u591a\u304f\u306e\u64cd\u4f5c\u3092\u5b9f\u884c\u3057\u3066\u5168\u4f53\u7684\u306a\u8a08\u7b97\u52b9\u7387\u3092\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u8a08\u7b97\u96c6\u7d04\u578b\u3068\u30e1\u30e2\u30ea\u96c6\u7d04\u578b\u306e\u30d0\u30e9\u30f3\u30b9<\/strong><br>CUDA\u306f\u30b9\u30ec\u30c3\u30c9\u6570\u3001\u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba\u3001\u30e1\u30e2\u30ea\u5272\u308a\u5f53\u3066\u6226\u7565\u3092\u8abf\u6574\u3059\u308b\u3053\u3068\u3067\u3001\u8a08\u7b97\u3068\u30e1\u30e2\u30ea\u30a2\u30af\u30bb\u30b9\u306e\u30d0\u30e9\u30f3\u30b9\u3092\u6700\u9069\u5316\u3057\u3001GPU\u30ea\u30bd\u30fc\u30b9\u3092\u52b9\u7387\u7684\u306b\u6d3b\u7528\u3057\u307e\u3059\u3002<\/li>\n<\/ol>\n\n\n\n<p>\u56f34-1\u306b\u793a\u3059\u3088\u3046\u306b\u3001\u5404\u30b9\u30ec\u30c3\u30c9\u304c\u7d50\u679c\u884c\u5217 $P$ \u306e1\u3064\u306e\u8981\u7d20\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002\u884c\u5217 $M$ \u3068\u884c\u5217 $N$ \u306f\u305d\u308c\u305e\u308c\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\u306b\u683c\u7d0d\u3055\u308c\u307e\u3059\u3002\u5404\u30b9\u30ec\u30c3\u30c9\u306f\u3001\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\u304b\u3089 $M$ \u306e1\u884c\u3068 $N$ \u306e1\u5217\u3092\u8aad\u307f\u53d6\u308a\u3001\u5185\u7a4d\u8a08\u7b97\u3092\u5b9f\u884c\u3057\u3066 $P$ \u306e1\u3064\u306e\u8981\u7d20\u3092\u6c42\u3081\u307e\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized\"><img decoding=\"async\" width=\"762\" height=\"752\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-4-1.png\" alt=\"\" class=\"wp-image-1747\" style=\"width:449px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-4-1.png 762w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-4-1-300x296.png 300w\" sizes=\"(max-width: 762px) 100vw, 762px\" \/><figcaption class=\"wp-element-caption\">\u56f34-1<\/figcaption><\/figure>\n\n\n\n<p>CUDA\u306e\u4e26\u5217\u51e6\u7406\u6280\u8853\u3092\u5229\u7528\u3059\u308b\u3053\u3068\u3067\u3001\u884c\u5217\u7a4d\u306e\u3088\u3046\u306a\u8a08\u7b97\u96c6\u7d04\u578b\u30bf\u30b9\u30af\u3092\u5927\u5e45\u306b\u9ad8\u901f\u5316\u3067\u304d\u3001\u79d1\u5b66\u8a08\u7b97\u3084\u753b\u50cf\u51e6\u7406\u306a\u3069\u306e\u5206\u91ce\u3067\u5e83\u304f\u5229\u7528\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">GPU\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306e\u6982\u8981<\/h4>\n\n\n\n<p>GPU\u306f\u4e3b\u306b\u30b9\u30c8\u30ea\u30fc\u30e0\u30d7\u30ed\u30bb\u30c3\u30b5\u30a2\u30ec\u30a4\u3068\u30e1\u30e2\u30ea\u30b7\u30b9\u30c6\u30e0\u3067\u69cb\u6210\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u305d\u306e\u69cb\u9020\u306f\u56f34-2\u306b\u793a\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized\"><img decoding=\"async\" width=\"803\" height=\"759\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-4-2.png\" alt=\"\" class=\"wp-image-1748\" style=\"width:434px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-4-2.png 803w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-4-2-300x284.png 300w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-4-2-768x726.png 768w\" sizes=\"(max-width: 803px) 100vw, 803px\" \/><figcaption class=\"wp-element-caption\">\u56f34-2<\/figcaption><\/figure>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5404\u30b9\u30c8\u30ea\u30fc\u30e0\u30d7\u30ed\u30bb\u30c3\u30b5\u306f\u3001GPU\u306e\u30b9\u30ec\u30c3\u30c9\uff08Thread\uff09\u3068\u547c\u3070\u308c\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u3059\u3079\u3066\u306e\u30b9\u30ec\u30c3\u30c9\u306f\u300c\u30b0\u30ea\u30c3\u30c9\uff08Grid\uff09 &#8211; \u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\uff08Block\uff09 &#8211; \u30b9\u30ec\u30c3\u30c9\uff08Thread\uff09\u300d\u306e\u968e\u5c64\u69cb\u9020\u3067\u7de8\u6210\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/li>\n\n\n\n<li>GPU\u5185\u306b\u306f\u8907\u6570\u306e\u30b0\u30ea\u30c3\u30c9\u304c\u542b\u307e\u308c\u3001\u5404\u30b0\u30ea\u30c3\u30c9\u306f\u8907\u6570\u306e\u30d6\u30ed\u30c3\u30af\u3067\u69cb\u6210\u3055\u308c\u3001\u5404\u30d6\u30ed\u30c3\u30af\u306f\u8907\u6570\u306e\u30b9\u30ec\u30c3\u30c9\u3092\u6301\u3061\u307e\u3059\u3002\u3053\u306e\u69cb\u9020\u306f\u3001\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u306e\u8981\u4ef6\u306b\u5fdc\u3058\u3066\u30012\u6b21\u5143\u30013\u6b21\u5143\u3001\u3055\u3089\u306b\u306f\u305d\u308c\u4ee5\u4e0a\u306e\u6b21\u5143\u3067\u7de8\u6210\u3067\u304d\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u5404\u30b9\u30ec\u30c3\u30c9\u306b\u306f\u30d7\u30e9\u30a4\u30d9\u30fc\u30c8\u306a\u30ec\u30b8\u30b9\u30bf\u3068\u30ed\u30fc\u30ab\u30eb\u30e1\u30e2\u30ea\uff08Local Memory\uff09\u304c\u3042\u308a\u3001\u540c\u3058\u30d6\u30ed\u30c3\u30af\u5185\u306e\u3059\u3079\u3066\u306e\u30b9\u30ec\u30c3\u30c9\u306f\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e\u5171\u6709\u30e1\u30e2\u30ea\uff08Shared Memory\uff09\u3092\u5171\u6709\u3057\u307e\u3059\u3002\u3059\u3079\u3066\u306e\u30d6\u30ed\u30c3\u30af\u306f\u3001\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\uff08Global Memory\uff09\u3001\u5b9a\u6570\u30e1\u30e2\u30ea\uff08Constant Memory\uff09\u3001\u304a\u3088\u3073\u30c6\u30af\u30b9\u30c1\u30e3\u30e1\u30e2\u30ea\uff08Texture Memory\uff09\u3092\u5171\u6709\u3057\u3001\u3053\u308c\u3089\u306e\u30e1\u30e2\u30ea\u306f\u30c7\u30fc\u30bf\u5c40\u6240\u6027\u3092\u8003\u616e\u3057\u3066\u6700\u9069\u5316\u3055\u308c\u3066\u3044\u307e\u3059\u3002\u8a08\u7b97\u4e2d\u3001\u30db\u30b9\u30c8\uff08Host\uff09\u306f\u3001\u307e\u305a\u51e6\u7406\u5bfe\u8c61\u306e\u30c7\u30fc\u30bf\u3092\u3053\u308c\u3089\u306e\u30e1\u30e2\u30ea\u306b\u30d0\u30c3\u30c1\u8ee2\u9001\u3057\u3001\u8a08\u7b97\u7d42\u4e86\u5f8c\u306b\u7d50\u679c\u3092\u30e1\u30e2\u30ea\u304b\u3089\u8aad\u307f\u53d6\u308a\u307e\u3059\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">CUDA\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u57fa\u672c\u69cb\u6210<\/h4>\n\n\n\n<p>CUDA\u30d7\u30ed\u30b0\u30e9\u30e0\u306f\u4e3b\u306b\u30db\u30b9\u30c8\u30b3\u30fc\u30c9\uff08Host Code\uff09\u3068\u30ab\u30fc\u30cd\u30eb\u95a2\u6570\u3067\u69cb\u6210\u3055\u308c\u307e\u3059\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u30db\u30b9\u30c8\u30b3\u30fc\u30c9<\/strong>\u306f\u3001CPU\uff08Host\uff09\u4e0a\u3067\u5b9f\u884c\u3055\u308c\u308bC\/C++\u30d7\u30ed\u30b0\u30e9\u30e0\u3067\u3001\u30e1\u30e2\u30ea\u306e\u78ba\u4fdd\u3001\u30c7\u30fc\u30bf\u306e\u524d\u51e6\u7406\u3001\u30c7\u30fc\u30bf\u8ee2\u9001\u3001CUDA\u30b9\u30ec\u30c3\u30c9\u6570\u306e\u5272\u308a\u5f53\u3066\u3001\u30ab\u30fc\u30cd\u30eb\u95a2\u6570\u306e\u547c\u3073\u51fa\u3057\u306a\u3069\u3092\u884c\u3044\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30ab\u30fc\u30cd\u30eb\u95a2\u6570<\/strong>\u306f\u3001GPU\uff08Device\uff09\u306e\u30cf\u30fc\u30c9\u30a6\u30a7\u30a2\u30b9\u30ec\u30c3\u30c9\uff08\u30b3\u30a2\uff09\u4e0a\u3067\u5b9f\u884c\u3055\u308c\u308b\u30b3\u30fc\u30c9\u3067\u3042\u308a\u3001\u305d\u306e\u305f\u3081\u30ab\u30fc\u30cd\u30eb\u95a2\u6570\u3068\u547c\u3070\u308c\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u4ee5\u4e0b\u306b\u3001\u884c\u5217\u52a0\u7b97\u306eCUDA\u30b5\u30f3\u30d7\u30eb\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u793a\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>\/\/ Device\u30ab\u30fc\u30cd\u30eb\u95a2\u6570\uff08GPU\u306e\u30cf\u30fc\u30c9\u30a6\u30a7\u30a2\u30b9\u30ec\u30c3\u30c9\u4e0a\u3067\u5b9f\u884c\uff09\n__global__ void matrixAdd(float A&#91;Ny]&#91;Nx], float B&#91;Ny]&#91;Nx], float C&#91;Ny]&#91;Nx]) \n{\n    \/\/ \u884c\u5217\u8981\u7d20\u306e\u884c\u756a\u53f7\u3068\u5217\u756a\u53f7\u3092\u53d6\u5f97\n    int row = blockIdx.y * blockDim.y + threadIdx.y;\n    int col = blockIdx.x * blockDim.x + threadIdx.x;\n\n    C&#91;row]&#91;col] = A&#91;row]&#91;col] + B&#91;row]&#91;col];\n}\n\n\/\/ Host\u30b3\u30fc\u30c9\uff08CPU\u4e0a\u3067\u5b9f\u884c\uff09\nint main()\n{\n    ......\n\n    const int Nx = 12; \n    const int Ny = 6;\n\n    \/\/ 1\u3064\u306eBlock\u306b4\u00d73\u306eThread\u3092\u5b9a\u7fa9\n    dim3 threadsPerBlock(4, 3); \n    \/\/ 1\u3064\u306eGrid\u306b3\u00d72\u306eBlock\u3092\u5b9a\u7fa9\n    dim3 numBlocks(Nx\/threadsPerBlock.x, Ny\/threadsPerBlock.y); \n    \/\/ \u30ab\u30fc\u30cd\u30eb\u95a2\u6570\u3092\u547c\u3073\u51fa\u3057\uff086\u30d6\u30ed\u30c3\u30af\u00d712\u30b9\u30ec\u30c3\u30c9=72\u30b9\u30ec\u30c3\u30c9\u4e26\u5217\u5b9f\u884c\uff09\n    matrixAdd&lt;&lt;&lt;numBlocks, threadsPerBlock&gt;&gt;&gt;(A, B, C);\n\n    ......\n}<\/code><\/pre>\n\n\n\n<p><strong>\u30b3\u30fc\u30c9\u306e\u8aac\u660e<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>2\u884c\u76ee<\/strong>\uff1a<code>__global__<\/code>\u30ad\u30fc\u30ef\u30fc\u30c9\u3092\u4f7f\u7528\u3057\u3066\u30ab\u30fc\u30cd\u30eb\u95a2\u6570\u3092\u5ba3\u8a00\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>5\uff5e6\u884c\u76ee<\/strong>\uff1aBlock\u3068Thread\u306e\u30b5\u30a4\u30ba\u5c5e\u6027\u3092\u4f7f\u7528\u3057\u3066\u884c\u5217\u8981\u7d20\u306e\u884c\u756a\u53f7\u3068\u5217\u756a\u53f7\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002\n<ul class=\"wp-block-list\">\n<li><code>blockIdx<\/code>\u306f\u73fe\u5728\u306e\u30ab\u30fc\u30cd\u30eb\u95a2\u6570\u304c\u5b9f\u884c\u3055\u308c\u308b\u30b9\u30ec\u30c3\u30c9\u304c\u5c5e\u3059\u308bBlock\u306e\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u3092\u8868\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><code>blockIdx.x<\/code>\u3068<code>blockIdx.y<\/code>\u306f\u3001\u305d\u308c\u305e\u308cBlock\u306ex\u8ef8\u65b9\u5411\u3068y\u8ef8\u65b9\u5411\u306e\u5ea7\u6a19\u3092\u793a\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><code>blockDim<\/code>\u306f\u73fe\u5728\u306eBlock\u306e\u30b5\u30a4\u30ba\u3092\u793a\u3057\u3001<code>blockDim.x<\/code>\u3068<code>blockDim.y<\/code>\u306f\u3001\u305d\u308c\u305e\u308cBlock\u306e\u5217\u6570\u3068\u884c\u6570\u3092\u8868\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><code>threadIdx<\/code>\u3082\u540c\u69d8\u3067\u3059\u3002<\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>14\uff5e18\u884c\u76ee<\/strong>\uff1a\u5909\u6570\u306e\u5b9a\u7fa9\u3001\u30e1\u30e2\u30ea\u306e\u78ba\u4fdd\u3001\u30c7\u30fc\u30bf\u306e\u524d\u51e6\u7406\u3001\u30c7\u30fc\u30bf\u8ee2\u9001\u306a\u3069\u3092\u884c\u3044\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>20\uff5e22\u884c\u76ee<\/strong>\uff1aGrid\u3068Block\u306e\u30b5\u30a4\u30ba\u3092\u5b9a\u7fa9\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>24\u884c\u76ee<\/strong>\uff1a\u5b9a\u7fa9\u3057\u305fGrid\u3068Block\u306e\u30d1\u30e9\u30e1\u30fc\u30bf\u3092\u4f7f\u7528\u3057\u3066\u30ab\u30fc\u30cd\u30eb\u95a2\u6570\u3092\u547c\u3073\u51fa\u3057\u3001\u884c\u5217\u52a0\u7b97\u3092\u5b9f\u884c\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u5185\u5bb9<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\">GPU\u306e\u884c\u5217\u4e57\u7b97<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\">\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9\u3092\u57fa\u306bCUDA\u306b\u3088\u308b\u884c\u5217\u7a4d\u306e\u5b9f\u88c5\u3092\u5b8c\u6210\u3055\u305b\u308b\u3002<\/p>\n\n\n\n<p>GPU\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u30e2\u30c7\u30eb\u306b\u6163\u308c\u308b\u3002<br>CPU\u306b\u3088\u308b\u884c\u5217\u7a4d\u306e\u5b9f\u88c5\u306e\u8003\u3048\u65b9\u3092\u5b66\u3076\u3002<br>CUDA\u30d5\u30a1\u30a4\u30eb\u306e\u30b3\u30f3\u30d1\u30a4\u30eb\u3068\u5b9f\u884c\u30d7\u30ed\u30bb\u30b9\u3092\u7fd2\u5f97\u3059\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>GPU\u306b\u3088\u308b\u884c\u5217\u7a4d\u306e\u7d50\u679c\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u7570\u306a\u308b\u30b5\u30a4\u30ba\u306e\u884c\u5217\u7a4d\u3092\u30c6\u30b9\u30c8\u3057\u3001\u5b9f\u9a13\u7d50\u679c\u3092\u89b3\u5bdf\u3059\u308b\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u5b9f\u9a13\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3068\u89e3\u51cd<\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u5b9f\u884c\u3059\u308b\u524d\u306b\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u304c\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u306b\u6b63\u5e38\u306b\u63a5\u7d9a\u3055\u308c\u3066\u3044\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br><code>lab4-5.tar.gz<\/code> \u3092<a href=\"https:\/\/www.yanagichiaki.jp\/index.php\/2024\/12\/09\/computer-architecture-labs-guidebook-hitsz\/#index_id6\">\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9<\/a>\u3057\u3001\u30e6\u30fc\u30b6\u30fc\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u30b3\u30d4\u30fc\u3057\u3066\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<p><code>tar -zxvf lab4-5.tar.gz<\/code><\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>CUDA\u884c\u5217\u7a4d\u30b3\u30fc\u30c9\u306e\u5b9f\u88c5<\/strong><\/div><div class=\"swell-block-step__body\">\n<ul class=\"wp-block-list\">\n<li><code>lab4-5<\/code> \u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u5185\u306e\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9 <code>matrix_mul.cu<\/code> \u3092\u958b\u304d\u307e\u3059\u3002<\/li>\n\n\n\n<li><code>MatrixMulKernel<\/code> \u95a2\u6570\u306e\u4e0b\u306bCUDA\u884c\u5217\u7a4d\u306e\u30b3\u30fc\u30c9\u3092\u5b9f\u88c5\u3057\u307e\u3059\u3002\n<ul class=\"wp-block-list\">\n<li>\u7d50\u679c\u884c\u5217\u306e\u4f4d\u7f6e\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u3092\u8a08\u7b97<\/li>\n\n\n\n<li>\u5404\u30b9\u30ec\u30c3\u30c9\u304c\u884c\u5217 $C$ \u306e1\u3064\u306e\u8981\u7d20\u306e\u5024\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002\u7d50\u679c\u884c\u5217 $C$ \u306e\u5404\u8981\u7d20\u3092\u8a08\u7b97\u3059\u308b\u30eb\u30fc\u30d7\u3092\u5b9f\u88c5\u3057\u306a\u3055\u3044\u3002<\/li>\n\n\n\n<li>\u8a08\u7b97\u7d50\u679c\u3092\u5bfe\u5fdc\u3059\u308b\u4f4d\u7f6e\u306e\u7d50\u679c\u884c\u5217\u306b\u5272\u308a\u5f53\u3066\u3002<\/li>\n\n\n\n<li>\u95a2\u6570\u547c\u3073\u51fa\u3057\u306e\u78ba\u8a8d: <code>main<\/code>\u95a2\u6570 \u3067 <code>MatrixMulKernel<\/code> \u95a2\u6570\u304c\u547c\u3073\u51fa\u3055\u308c\u3066\u3044\u308b\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>bash compile.sh\n.\/a.out 1 1000<\/code><\/pre>\n\n\n\n<p>Total Errors = 0\u306e\u5834\u5408\u3001\u30ab\u30fc\u30cd\u30eb\u304c\u6b63\u78ba\u3002<\/p>\n\n\n\n<p><strong>GPU\u3067\u306e\u884c\u5217\u7a4d\u306e\u8a08\u7b97\u6642\u9593<\/strong> \u3068 <strong>\u8a08\u7b97\u7d50\u679c\u306e\u6b63\u78ba\u6027<\/strong> \u3092\u78ba\u8a8d\u3057\u307e\u3059\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong>\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u5909\u66f4\u3068\u7d50\u679c\u306e\u6bd4\u8f03<\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<ul class=\"wp-block-list\">\n<li><strong>\u884c\u5217\u306e\u30b5\u30a4\u30ba<\/strong>\u3092\u5909\u66f4\u3057\u3066\u30c6\u30b9\u30c8\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba <code>TILE_SIZE<\/code><\/strong> \u3092\u5909\u66f4\u3057\u3066\u30c6\u30b9\u30c8\u3057\u307e\u3059\uff08\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u3082\u6b63\u65b9\u5f62\uff09\u3002<\/li>\n\n\n\n<li><strong>\u7570\u306a\u308b\u30d1\u30e9\u30e1\u30fc\u30bf\u8a2d\u5b9a\u4e0b\u3067\u8a08\u7b97\u7d50\u679c\u3068\u6027\u80fd\u3092\u6bd4\u8f03<\/strong> \u3057\u3001\u6700\u9069\u306a\u8a2d\u5b9a\u3092\u8003\u5bdf\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading\">\u6b21\u306e\u5b9f\u9a13\u306e\u4e88\u7fd2<\/h3>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u8cc7\u6599\u3092\u8aad\u307f\u3001CUDA\u306b\u304a\u3051\u308b\u5171\u6709\u30e1\u30e2\u30ea\u3092\u4f7f\u7528\u3057\u305f\u884c\u5217\u7a4d\u306e\u6700\u9069\u5316\u306e\u57fa\u672c\u539f\u7406\u3092\u7406\u89e3\u3057\u3066\u304f\u3060\u3055\u3044\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/docs.nvidia.com\/cuda\/cuda-c-programming-guide\/\">CUDA C++ Programming Guide<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/docs.nvidia.com\/cuda\/cuda-c-best-practices-guide\/\">CUDA C++ Best Practices Guide<\/a><\/li>\n<\/ul>\n\n\n\n<h2 class=\"wp-block-heading\">Lab5 CUDA\uff1aGPU\u306e\u884c\u5217\u4e57\u7b97\u6700\u9069\u5316<\/h2>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u76ee\u7684<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>CUDA\u306e\u5171\u6709\u30e1\u30e2\u30ea\u3092\u5229\u7528\u3057\u305f\u6700\u9069\u5316\u6280\u8853<\/strong>\u3092\u7fd2\u5f97\u3057\u3001\u884c\u5217\u7a4d\u306e\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u65b9\u6cd5\u3092\u7406\u89e3\u3059\u308b\u3002<\/li>\n\n\n\n<li><strong>GPU\u30d7\u30ed\u30b0\u30e9\u30e0\u306e\u6027\u80fd\u30dc\u30c8\u30eb\u30cd\u30c3\u30af\u3092\u5206\u6790\u3059\u308b\u65b9\u6cd5<\/strong>\u3092\u7fd2\u5f97\u3059\u308b\u3002<\/li>\n\n\n\n<li><strong>\u5171\u6709\u30e1\u30e2\u30ea\u3092\u6d3b\u7528\u3057\u3066\u30c7\u30fc\u30bf\u30a2\u30af\u30bb\u30b9\u7d4c\u8def\u3092\u6700\u9069\u5316\u3059\u308b\u624b\u6cd5<\/strong>\u3092\u5b66\u3076\u3002<\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u524d\u306e\u6e96\u5099<\/h3>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u4ee5\u4e0b\u306e\u8cc7\u6599\u3092\u8aad\u307f\u3001CUDA\u306b\u304a\u3051\u308b\u5171\u6709\u30e1\u30e2\u30ea\u3092\u4f7f\u7528\u3057\u305f\u884c\u5217\u7a4d\u306e\u6700\u9069\u5316\u306e\u57fa\u672c\u539f\u7406\u3092\u7406\u89e3\u3057\u3066\u304f\u3060\u3055\u3044\uff1a\n<ul class=\"wp-block-list\">\n<li><a href=\"https:\/\/docs.nvidia.com\/cuda\/cuda-c-programming-guide\/\">CUDA C++ Programming Guide<\/a><\/li>\n\n\n\n<li><a href=\"https:\/\/docs.nvidia.com\/cuda\/cuda-c-best-practices-guide\/\">CUDA C++ Best Practices Guide<\/a><\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>Lab\u74b0\u5883<\/strong>\uff1aCUDA Toolkit\u3001NVIDIA\u30c9\u30e9\u30a4\u30d0\u30fc\u3001\u304a\u3088\u3073\u95a2\u9023\u958b\u767a\u30c4\u30fc\u30eb\uff08<code>nvcc<\/code>\u3001<code>cuda-gdb<\/code> \u306a\u3069\uff09\u304c\u30a4\u30f3\u30b9\u30c8\u30fc\u30eb\u3055\u308c\u3066\u3044\u308b\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u3002<\/li>\n<\/ul>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u539f\u7406<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\">CUDA\u5171\u6709\u30e1\u30e2\u30ea\u6700\u9069\u5316<\/h4>\n\n\n\n<p><strong>\u5171\u6709\u30e1\u30e2\u30ea\uff08Shared Memory\uff09<\/strong>\u306f\u3001CUDA\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u304a\u3044\u3066\u3001\u5404\u30b9\u30c8\u30ea\u30fc\u30df\u30f3\u30b0\u30de\u30eb\u30c1\u30d7\u30ed\u30bb\u30c3\u30b5\uff08Streaming Multiprocessor, SM\uff09\u5185\u306b\u914d\u7f6e\u3055\u308c\u305f\u9ad8\u901f\u30ad\u30e3\u30c3\u30b7\u30e5\u3067\u3042\u308a\u3001\u540c\u3058\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e\u30b9\u30ec\u30c3\u30c9\u9593\u3067\u30c7\u30fc\u30bf\u3092\u5171\u6709\u3059\u308b\u305f\u3081\u306b\u4f7f\u7528\u3055\u308c\u307e\u3059\u3002\u5171\u6709\u30e1\u30e2\u30ea\u306f\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\u3068\u6bd4\u8f03\u3057\u3066\u30a2\u30af\u30bb\u30b9\u901f\u5ea6\u304c\u975e\u5e38\u306b\u901f\u3044\u305f\u3081\u3001\u983b\u7e41\u306b\u540c\u3058\u30c7\u30fc\u30bf\u306b\u30a2\u30af\u30bb\u30b9\u3059\u308b\u5fc5\u8981\u304c\u3042\u308b\u5834\u5408\u306b\u3001\u6027\u80fd\u3092\u5927\u5e45\u306b\u5411\u4e0a\u3055\u305b\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002\u5171\u6709\u30e1\u30e2\u30ea\u3092\u6d3b\u7528\u3057\u3066\u518d\u5229\u7528\u3059\u308b\u30c7\u30fc\u30bf\u3092\u4e8b\u524d\u306b\u30ed\u30fc\u30c9\u3059\u308b\u3053\u3068\u3067\u3001\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\u3078\u306e\u30a2\u30af\u30bb\u30b9\u56de\u6570\u3092\u6e1b\u3089\u3057\u3001\u9045\u5ef6\u3092\u4f4e\u6e1b\u3057\u3001\u5e2f\u57df\u5e45\u306e\u5229\u7528\u52b9\u7387\u3092\u5411\u4e0a\u3055\u305b\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u5171\u6709\u30e1\u30e2\u30ea\u306e\u9ad8\u901f\u5316\u539f\u7406\u306f\u3001\u6b21\u306e\u91cd\u8981\u306a\u30dd\u30a4\u30f3\u30c8\u306b\u96c6\u7d04\u3055\u308c\u307e\u3059\uff1a<\/p>\n\n\n\n<p><strong>1. \u9ad8\u901f\u306a\u30a2\u30af\u30bb\u30b9\u901f\u5ea6<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5171\u6709\u30e1\u30e2\u30ea\u306fCUDA\u306e\u5404\u30b9\u30c8\u30ea\u30fc\u30df\u30f3\u30b0\u30de\u30eb\u30c1\u30d7\u30ed\u30bb\u30c3\u30b5\u4e0a\u306b\u914d\u7f6e\u3055\u308c\u3066\u304a\u308a\u3001\u30a2\u30af\u30bb\u30b9\u9045\u5ef6\u304c\u975e\u5e38\u306b\u4f4e\u304f\u3001\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\u3088\u308a\u3082\u9065\u304b\u306b\u9ad8\u901f\u3067\u3059\u3002<\/li>\n\n\n\n<li>\u983b\u7e41\u306b\u30a2\u30af\u30bb\u30b9\u3059\u308b\u30c7\u30fc\u30bf\u3092\u683c\u7d0d\u3059\u308b\u306e\u306b\u9069\u3057\u3066\u304a\u308a\u3001\u8a08\u7b97\u51e6\u7406\u306e\u52b9\u7387\u3092\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>2. \u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u3067\u306e\u5171\u6709<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5171\u6709\u30e1\u30e2\u30ea\u306f\u3001\u540c\u3058\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e\u3059\u3079\u3066\u306e\u30b9\u30ec\u30c3\u30c9\u306b\u3088\u3063\u3066\u30a2\u30af\u30bb\u30b9\u304a\u3088\u3073\u5171\u6709\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u8907\u6570\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u5354\u8abf\u3057\u3066\u4f5c\u696d\u3057\u3001\u540c\u3058\u30c7\u30fc\u30bf\u3092\u7e70\u308a\u8fd4\u3057\u8aad\u307f\u53d6\u308b\u5fc5\u8981\u304c\u306a\u304f\u306a\u308b\u305f\u3081\u3001\u30e1\u30e2\u30ea\u30a2\u30af\u30bb\u30b9\u306b\u304b\u304b\u308b\u6642\u9593\u3092\u524a\u6e1b\u3067\u304d\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u56f35-1\u306b\u793a\u3059\u3088\u3046\u306b\u3001\u5171\u6709\u30e1\u30e2\u30ea\u306f\u5404\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306b\u914d\u7f6e\u3055\u308c\u3066\u304a\u308a\u3001\u540c\u3058\u30d6\u30ed\u30c3\u30af\u5185\u306e\u3059\u3079\u3066\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u9ad8\u901f\u306b\u30c7\u30fc\u30bf\u3092\u5171\u6709\u3067\u304d\u307e\u3059\u3002\u3053\u308c\u306b\u3088\u308a\u3001\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\u3078\u306e\u30a2\u30af\u30bb\u30b9\u56de\u6570\u304c\u6e1b\u5c11\u3057\u3001\u9045\u5ef6\u304c\u8efd\u6e1b\u3055\u308c\u3001\u30e1\u30e2\u30ea\u5e2f\u57df\u5e45\u306e\u5229\u7528\u52b9\u7387\u304c\u5411\u4e0a\u3057\u307e\u3059\u3002\u5171\u6709\u30e1\u30e2\u30ea\u306f\u7279\u306b\u3001\u8907\u6570\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u983b\u7e41\u306b\u540c\u3058\u30c7\u30fc\u30bf\u3092\u5fc5\u8981\u3068\u3059\u308b\u8a08\u7b97\u51e6\u7406\u306b\u304a\u3044\u3066\u3001\u975e\u5e38\u306b\u52b9\u679c\u7684\u3067\u3059\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image size-full is-resized\"><img decoding=\"async\" width=\"718\" height=\"634\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-1.png\" alt=\"\" class=\"wp-image-1749\" style=\"width:482px;height:auto\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-1.png 718w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-1-300x265.png 300w\" sizes=\"(max-width: 718px) 100vw, 718px\" \/><figcaption class=\"wp-element-caption\">\u56f35-1<\/figcaption><\/figure>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u5185\u5bb9<\/h3>\n\n\n\n<h4 class=\"wp-block-heading\">\u5171\u6709\u30e1\u30e2\u30ea\u3067\u306eGPU\u884c\u5217\u4e57\u7b97\u6700\u9069\u5316<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\">\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9\u3092\u57fa\u306bCUDA\u306b\u3088\u308b\u884c\u5217\u7a4d\u306e\u5b9f\u88c5\u3092\u5b8c\u6210\u3055\u305b\u308b\u3002<\/p>\n\n\n\n<p>\u5171\u6709\u30e1\u30e2\u30ea\u3092\u4f7f\u7528\u3057\u3066\u30b0\u30ed\u30fc\u30d0\u30eb\u30e1\u30e2\u30ea\u3078\u306e\u30a2\u30af\u30bb\u30b9\u3092\u6e1b\u3089\u3057\u3001\u30c7\u30fc\u30bf\u30a2\u30af\u30bb\u30b9\u52b9\u7387\u3092\u5411\u4e0a\u3055\u305b\u308b\u65b9\u6cd5\u3092\u5b66\u3076\u3002<br>\u5171\u6709\u30e1\u30e2\u30ea\u30b5\u30a4\u30ba\u306b\u57fa\u3065\u3044\u3066\u884c\u5217\u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba\u3092\u6700\u9069\u5316\u3057\u3001\u8a08\u7b97\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u65b9\u6cd5\u3092\u7406\u89e3\u3059\u308b\u3002<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>CUDA\u306e\u5171\u6709\u30e1\u30e2\u30ea\u30e1\u30ab\u30cb\u30ba\u30e0\u3092\u4f7f\u7528\u3057\u3066\u884c\u5217\u7a4d\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u6700\u9069\u5316\u3057\u3001\u6700\u9069\u5316\u524d\u5f8c\u306e\u8a08\u7b97\u6642\u9593\u3068\u52b9\u7387\u3092\u6bd4\u8f03\u3059\u308b\u3002<\/li>\n\n\n\n<li>BLOCKSIZE\u306e\u30b5\u30a4\u30ba\u3092\u8abf\u6574\u3057\u3001\u7570\u306a\u308bBLOCKSIZE\u304c\u8a08\u7b97\u52b9\u7387\u306b\u4e0e\u3048\u308b\u5f71\u97ff\u3092\u6bd4\u8f03\u3059\u308b\u3002<\/li>\n<\/ul>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u30b5\u30f3\u30d7\u30eb\u30b3\u30fc\u30c9\u3092\u5c0e\u5165<\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u4ee5\u4e0b\u306e\u30b3\u30fc\u30c9\u3092<a href=\"https:\/\/www.yanagichiaki.jp\/index.php\/2024\/12\/09\/computer-architecture-labs-guidebook-hitsz\/#index_id55\">Lab4\u304b\u3089<\/a>\u306e<code>matrix_mul.cu<\/code>\u306e<code>MatrixMulSharedMemKernel<\/code>\u95a2\u6570\u306b\u8cbc\u308a\u4ed8\u3051\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>  \/\/ Block index\n  int bx = blockIdx.x;\n  int by = blockIdx.y;\n\n  \/\/ Thread index\n  int tx = threadIdx.x;\n  int ty = threadIdx.y;\n\n  \/\/ Index of the first sub-matrix of A processed by the block\n  int aBegin = wA * BLOCK_SIZE * by;\n\n  \/\/ Index of the last sub-matrix of A processed by the block\n  int aEnd   = aBegin + wA - 1;\n\n  \/\/ Step size used to iterate through the sub-matrices of A\n  int aStep  = BLOCK_SIZE;\n\n  \/\/ Index of the first sub-matrix of B processed by the block\n  int bBegin = BLOCK_SIZE * bx;\n\n  \/\/ Step size used to iterate through the sub-matrices of B\n  int bStep  = BLOCK_SIZE * wB;\n\n  \/\/ Csub is used to store the element of the block sub-matrix\n  \/\/ that is computed by the thread\n  float Csub = 0;\n\n  \/\/ Loop over all the sub-matrices of A and B\n  \/\/ required to compute the block sub-matrix\n  for (int a = aBegin, b = bBegin;\n       a &lt; aEnd;\n       a += aStep, b += bStep) {\n    \/\/ Declaration of the shared memory array As used to\n    \/\/ store the sub-matrix of A\n    __shared__ float As&#91;BLOCK_SIZE]&#91;BLOCK_SIZE];\n\n    \/\/ Declaration of the shared memory array Bs used to\n    \/\/ store the sub-matrix of B\n    __shared__ float Bs&#91;BLOCK_SIZE]&#91;BLOCK_SIZE];\n\n    \/\/ Load the matrices from device memory\n    \/\/ to shared memory; each **thread** loads\n    \/\/ one element of each matrix\n    \/\/ --- TO DO :Load the elements of the sub-matrix of A into As ---\n    \/\/ ---        Load the elements of the sub-matrix of B into Bs ---\n    \/\/ NOTE: Ensure that the thread indices do not exceed the matrix dimensions to avoid out-of-bounds access.\n    \/\/       Use boundary checks to load valid elements into shared memory, and set invalid elements to 0.0f\n\n\n\n\n    \/\/ Synchronize to make sure the matrices are loaded\n    __syncthreads();\n\n    \/\/ Multiply the two matrices together;\n    \/\/ each thread computes one element\n    \/\/ of the block sub-matrix\n#pragma unroll\n    \/\/ --- TO DO :Implement the matrix multiplication using the sub-matrices As and Bs ---\n\n\n\n\n    \/\/ Synchronize to make sure that the preceding\n    \/\/ computation is done before loading two new\n    \/\/ sub-matrices of A and B in the next iteration\n    __syncthreads();\n  }\n\n  \/\/ Write the block sub-matrix to device memory;\n  \/\/ each thread writes one element\n  int c = wB * BLOCK_SIZE * by + BLOCK_SIZE * bx;\n  \/\/ --- TO DO :Store the computed Csub result into matrix C ---\n  \/\/ NOTE: Ensure that the thread indices \"c\" do not exceed the matrix dimensions to avoid out-of-bounds access.\n  \/\/       Use boundary checks to write valid elements to the output matrix C.<\/code><\/pre>\n\n\n\n<p><strong>\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9\u306e\u5b9f\u88c5\u6982\u8981<\/strong><\/p>\n\n\n\n<p>\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u30b3\u30fc\u30c9\u3067\u306f\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u30b9\u30ec\u30c3\u30c9\u3068\u884c\u5217\u8981\u7d20\u306e\u5bfe\u5fdc\u95a2\u4fc2\u3092\u5b9a\u7fa9\u3057\u3066\u3044\u307e\u3059\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5404\u30b9\u30ec\u30c3\u30c9\u304c\u884c\u5217 $C$ \u306e1\u3064\u306e\u8981\u7d20\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u5404\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u304c\u884c\u5217\u306e1\u3064\u306e\u30d6\u30ed\u30c3\u30af\u3092\u62c5\u5f53\u3057\u3001\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u30d6\u30ed\u30c3\u30af\u5185\u306e\u5404\u8981\u7d20\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<figure class=\"wp-block-image size-full\"><img decoding=\"async\" width=\"1774\" height=\"901\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-2.png\" alt=\"\" class=\"wp-image-1763\" srcset=\"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-2.png 1774w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-2-300x152.png 300w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-2-1024x520.png 1024w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-2-768x390.png 768w, https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALab-5-2-1536x780.png 1536w\" sizes=\"(max-width: 1774px) 100vw, 1774px\" \/><\/figure>\n\n\n\n<p class=\"is-style-big_kakko_box\">\u53f3\u56f3\u306f\u3001\u5de6\u56f3\u306e\u7b2c2\u884c\u6700\u5f8c\u306e\u5217\u306b\u3042\u308b\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u306b\u5bfe\u5fdc\u3059\u308b\u5206\u5272\u884c\u5217\u3092\u8868\u3057\u307e\u3059\u3002<br>\u4f8b\u3048\u3070\u3001$C'[2][1]$ \u306f\u305d\u306e\u5206\u5272\u884c\u5217\u306e3\u884c\u76ee2\u5217\u76ee\u306e\u8981\u7d20\u3092\u793a\u3057\u307e\u3059\u3002\u4ed6\u306e\u8981\u7d20\u3082\u540c\u69d8\u306b\u5bfe\u5fdc\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u8a08\u7b97\u306e\u969b\u3001\u5404\u30b9\u30ec\u30c3\u30c9\u306f\u884c\u5217 $A$ \u304a\u3088\u3073 $B$ \u304b\u3089\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e\u7279\u5b9a\u4f4d\u7f6e\u306b\u5bfe\u5fdc\u3059\u308b\u8981\u7d20\u30921\u3064\u305a\u3064\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<br>\u4f8b\u3048\u3070\u3001\u56f3\u306e\u30b9\u30ec\u30c3\u30c9 <strong>Thread(1,2)<\/strong> \u306f\u3001\u6700\u521d\u306e\u30eb\u30fc\u30d7\u3067\u306f\u56f3\u306e <code>aBegin<\/code> \u304c\u6307\u3057\u793a\u3059\u5206\u5272\u884c\u5217\u304b\u3089 $A'[2][1]$\u3001<code>bBegin<\/code> \u304c\u6307\u3057\u793a\u3059\u5206\u5272\u884c\u5217\u304b\u3089 $B'[2][1]$ \u3092\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<br>2\u56de\u76ee\u306e\u30eb\u30fc\u30d7\u3067\u306f\u3001<code>aBegin + aStep<\/code> \u304c\u6307\u3057\u793a\u3059\u5206\u5272\u884c\u5217\u304b\u3089 $A'[2][1]$\u3001<code>bBegin + bStep<\/code> \u304c\u6307\u3057\u793a\u3059\u5206\u5272\u884c\u5217\u304b\u3089 $B'[2][1]$ \u3092\u30ed\u30fc\u30c9\u3057\u3001\u3053\u306e\u51e6\u7406\u3092\u7e70\u308a\u8fd4\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u30b9\u30ec\u30c3\u30c9 <strong>Thread(i,j)<\/strong> \u306f\u5404\u30eb\u30fc\u30d7\u3067 $A'[j][i]$ \u3068 $B'[j][i]$ \u306e2\u3064\u306e\u8981\u7d20\u3092\u30ed\u30fc\u30c9\u3059\u308b\u3060\u3051\u3067\u3059\u304c\u3001\u8a08\u7b97\u3067\u5fc5\u8981\u306a $A'[j][k]$ \u304a\u3088\u3073 $B'[k][i]$ \u306e\u3059\u3079\u3066\u306e\u8981\u7d20$0 \\leq k &lt; \\text{BLOCK\\_SIZE}$\u306f\u3001\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e\u4ed6\u306e\u30b9\u30ec\u30c3\u30c9\u306b\u3088\u3063\u3066\u4e26\u5217\u7684\u306b\u5171\u6709\u30e1\u30e2\u30ea\u306b\u30ed\u30fc\u30c9\u3055\u308c\u307e\u3059\u3002\u3053\u306e\u305f\u3081\u3001<strong>Thread(i,j)<\/strong> \u306f\u6b63\u78ba\u306b $C'[j][i]$ \u3092\u8a08\u7b97\u3059\u308b\u3053\u3068\u304c\u53ef\u80fd\u3067\u3059\u3002<\/p>\n\n\n\n<p>\u5b9f\u88c5\u3092\u7c21\u7565\u5316\u3059\u308b\u305f\u3081\u3001\u884c\u5217\u8981\u7d20\u306e\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u3092\u300c\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u300d\u3068\u300c\u30d6\u30ed\u30c3\u30af\u5185\u30b9\u30ec\u30c3\u30c9\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u300d\u306e2\u6bb5\u968e\u306b\u5206\u3051\u3066\u8003\u3048\u307e\u3059\u3002<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong><br>\u3053\u308c\u306f\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5de6\u4e0a\u306e\u30b9\u30ec\u30c3\u30c9\u306e\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u3067\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30d6\u30ed\u30c3\u30af\u5185\u30b9\u30ec\u30c3\u30c9\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong><br>\u30d6\u30ed\u30c3\u30af\u5185\u306e\u30b9\u30ec\u30c3\u30c9\u304c\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u304b\u3089\u3069\u308c\u3060\u3051\u305a\u308c\u3066\u3044\u308b\u304b\u3092\u793a\u3057\u307e\u3059\u3002<\/li>\n<\/ol>\n\n\n\n<p>\u56f3\u304b\u3089\u5206\u304b\u308b\u3088\u3046\u306b\u3001$C'[2][1]$ \u306f $(X,1)$ \u756a\u76ee\u306e\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u5185\u306e $(1,2)$ \u756a\u76ee\u306e\u30b9\u30ec\u30c3\u30c9\u306b\u3088\u3063\u3066\u8a08\u7b97\u3055\u308c\u307e\u3059\u3002<br>\u884c\u5217 $C$ \u306e\u5217\u6570 $N$ \u3068\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u306e\u8fba\u9577 $\\text{BLOCK\\_SIZE}$ \u3092\u7d44\u307f\u5408\u308f\u305b\u308b\u3053\u3068\u3067\u4ee5\u4e0b\u304c\u5f97\u3089\u308c\u307e\u3059\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af(X,1)\u306e2\u6b21\u5143\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong>\uff1a\\[[1\\times\\text{BLOCK\\_SIZE}][X\\times\\text{BLOCK\\_SIZE}]\\]<\/li>\n\n\n\n<li><strong>\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af(X,1)\u306e1\u6b21\u5143\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong>\uff1a\\[(1\\times N + X)\\times \\text{BLOCK\\_SIZE}\\]<\/li>\n<\/ul>\n\n\n\n<p>\u307e\u305f\u3001\u30d6\u30ed\u30c3\u30af\u5185\u30b9\u30ec\u30c3\u30c9\u306e\u5ea7\u6a19 $(1,2)$ \u304b\u3089\u4ee5\u4e0b\u3092\u5c0e\u304d\u51fa\u305b\u307e\u3059\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u30d6\u30ed\u30c3\u30af\u5185\u30b9\u30ec\u30c3\u30c9\u306e2\u6b21\u5143\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong>\uff1a$[2][1]$<\/li>\n\n\n\n<li><strong>\u30d6\u30ed\u30c3\u30af\u5185\u30b9\u30ec\u30c3\u30c9\u306e1\u6b21\u5143\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong>\uff1a$2\\times N + 1$<\/li>\n<\/ul>\n\n\n\n<p>\u3053\u308c\u3089\u3092\u52a0\u7b97\u3059\u308b\u3068\u3001$C'[2][1]$ \u304c\u884c\u5217 $C$ \u306b\u5bfe\u5fdc\u3059\u308b<strong>\u5168\u4f53\u7684\u306a2\u6b21\u5143\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong>\u306f\uff1a\\[[1*\\text{BLOCK\\_SIZE} + 2][X*\\text{BLOCK\\_SIZE} + 1]\\]<\/p>\n\n\n\n<p>\u5168\u4f53\u7684\u306a<strong>1\u6b21\u5143\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<\/strong>\u306f\uff1a\\[(1\\times N + X)\\times\\text{BLOCK\\_SIZE} + 2\\times N + 1\\]<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>CUDA\u5171\u6709\u30e1\u30e2\u30ea\u884c\u5217\u7a4d\u30b3\u30fc\u30c9\u306e\u5b9f\u88c5<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><strong>\u5171\u6709\u30e1\u30e2\u30ea\u3078\u306e\u30c7\u30fc\u30bf\u8aad\u307f\u53d6\u308a<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u5404\u30b9\u30ec\u30c3\u30c9\u304c\u884c\u5217 $A$ \u3068 $B$ \u306e\u8981\u7d20\u3092\u305d\u308c\u305e\u308c\u5171\u6709\u30e1\u30e2\u30ea<code>As[ty][tx]<\/code>\u3001<code>Bs[ty][tx]<\/code>\u306b\u30ed\u30fc\u30c9\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u5883\u754c\u6761\u4ef6\u3092\u78ba\u8a8d\u3057\u3001\u7bc4\u56f2\u5916\u30a2\u30af\u30bb\u30b9\u3092\u9632\u6b62\u3059\u308b\u305f\u3081\u3001\u6709\u52b9\u306a\u8981\u7d20\u306e\u307f\u3092\u8aad\u307f\u53d6\u308a\u3001\u7121\u52b9\u306a\u8981\u7d20\u306b\u306f<code>0.0f<\/code>\u3092\u8a2d\u5b9a\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>\u884c\u5217 $C$ \u306e\u8a08\u7b97<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><code>for<\/code>\u30eb\u30fc\u30d7\u3092\u4f7f\u7528\u3057\u3066\u5171\u6709\u30e1\u30e2\u30ea<code>As<\/code>\u3068<code>Bs<\/code>\u3092\u4f7f\u3044\u3001\u5bfe\u5fdc\u3059\u308b\u8981\u7d20\u3092\u639b\u3051\u7b97\u3057\u3001\u7d2f\u7a4d\u3057\u3066<code>Csub<\/code>\u306b\u4fdd\u5b58\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>\u8a08\u7b97\u7d50\u679c\u306e\u66f8\u304d\u623b\u3057<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li>\u30d6\u30ed\u30c3\u30af\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<code>(bx, by)<\/code>\u3068\u30b9\u30ec\u30c3\u30c9\u30a4\u30f3\u30c7\u30c3\u30af\u30b9<code>(tx, ty)<\/code>\u3092\u4f7f\u7528\u3057\u3066\u3001\u884c\u5217 $C$ \u306e\u30b0\u30ed\u30fc\u30d0\u30eb\u4f4d\u7f6e\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u66f8\u304d\u623b\u3057\u524d\u306b\u30b0\u30ed\u30fc\u30d0\u30eb\u30a4\u30f3\u30c7\u30c3\u30af\u30b9\u304c\u884c\u5217 $C$ \u306e\u6709\u52b9\u7bc4\u56f2\u5185\u3067\u3042\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n\n\n\n<p><strong>main\u95a2\u6570\u306e\u5909\u66f4<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><code>for<\/code>\u30eb\u30fc\u30d7\u3092\u5909\u66f4\u3057\u3001<code>MatrixMulSharedMemKernel<\/code>\u95a2\u6570\u3092\u547c\u3073\u51fa\u3059\u3088\u3046\u306b\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u30ab\u30fc\u30cd\u30eb\u306e\u6b63\u78ba\u6027\u3092\u691c\u8a3c<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>bash compile.sh\n.\/a.out 1 1000<\/code><\/pre>\n\n\n\n<p>Total Errors = 0\u306e\u5834\u5408\u3001\u30ab\u30fc\u30cd\u30eb\u304c\u6b63\u78ba\u3002<\/p>\n\n\n\n<p>\u4e00\u90e8\u306eGPU\u306e\u8a08\u7b97\u7cbe\u5ea6\u304c\u4f4e\u3044\u306e\u5834\u5408\u3001\u3064\u307e\u308a\u6570\u5024\u8aa4\u5dee\u304c<code>0.5<\/code>\u4ee5\u5185\u3067\u3042\u308c\u3070\u3001\u5207\u4e0a\u3001\u5207\u4e0b\u306e\u30ed\u30b8\u30c3\u30af\u3092\u8ffd\u52a0\u3057\u3066\u88dc\u6b63\u53ef\u80fd\u3067\u3059\u3002<\/p>\n\n\n\n<p><strong>GPU\u3067\u306e\u884c\u5217\u7a4d\u306e\u8a08\u7b97\u6642\u9593<\/strong> \u3068 <strong>\u8a08\u7b97\u7d50\u679c\u306e\u6b63\u78ba\u6027<\/strong> \u3092\u78ba\u8a8d\u3057\u307e\u3059\u3002Lab4\u306e\u8a08\u7b97\u6027\u80fd\u3068\u6bd4\u8f03\u3057\u3066\u4e0b\u3055\u3044\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong>\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u5909\u66f4\u3068\u7d50\u679c\u306e\u6bd4\u8f03<\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<ul class=\"wp-block-list\">\n<li><strong>\u884c\u5217\u306e\u30b5\u30a4\u30ba<\/strong>\u3092\u5909\u66f4\u3057\u3066\u30c6\u30b9\u30c8\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba <code>TILE_SIZE<\/code><\/strong> \u3092\u5909\u66f4\u3057\u3066\u30c6\u30b9\u30c8\u3057\u307e\u3059\uff08\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u3082\u6b63\u65b9\u5f62\uff09\u3002<\/li>\n\n\n\n<li><strong>\u7570\u306a\u308b\u30d1\u30e9\u30e1\u30fc\u30bf\u8a2d\u5b9a\u4e0b\u3067\u8a08\u7b97\u7d50\u679c\u3068\u6027\u80fd\u3092\u6bd4\u8f03<\/strong> \u3057\u3001\u6700\u9069\u306a\u8a2d\u5b9a\u3092\u8003\u5bdf\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n<\/div>\n\n\n\n<h4 class=\"wp-block-heading\">CUBLAS\u3067\u306eGPU\u884c\u5217\u4e57\u7b97\u6700\u9069\u5316<\/h4>\n\n\n\n<p class=\"is-style-icon_pen\">\u5b9f\u88c5\u3057\u305f\u884c\u5217\u7a4d\u3068CUDA\u306e\u7d44\u307f\u8fbc\u307f\u884c\u5217\u7a4d\u6f14\u7b97\u5b50 <code>cublasSgemm<\/code> \u306e\u5b9f\u884c\u7d50\u679c\u3092\u6bd4\u8f03\u3059\u308b\u3002<\/p>\n\n\n\n<p><code>cublasSgemm<\/code>\u306fCUDA\u306ecuBLAS\u30e9\u30a4\u30d6\u30e9\u30ea\u306e\u884c\u5217\u7a4d\u6f14\u7b97\u95a2\u6570\u3067\u3001\u4ee5\u4e0b\u306e\u3088\u3046\u306b\u884c\u5217\u8a08\u7b97\u3092\u884c\u3044\u307e\u3059\uff1a\\[C = \\alpha \\cdot A \\cdot B + \\beta \\cdot C\\]<\/p>\n\n\n\n<p>\u8a73\u7d30\u306a\u4f7f\u7528\u65b9\u6cd5\u306f\u4ee5\u4e0b\u3092\u53c2\u7167\u3057\u3066\u304f\u3060\u3055\u3044\uff1a<a href=\"https:\/\/docs.nvidia.com\/cuda\/cublas\/index.html?#cublas-t-gemm\">CUDA cuBLAS Documentation &#8211; GEMM<\/a><\/p>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u30de\u30af\u30ed\u3092\u8d77\u7528<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><code>matrix_mul.cu<\/code>\u5148\u982d\u306e<code>#define USE_CUBLAS<\/code>\u3092\u8d77\u7528\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>cublasSgemm\u3092\u547c\u3073\u51fa\u3059<\/strong><\/div><div class=\"swell-block-step__body\">\n<p><strong>main\u95a2\u6570\u306e\u5909\u66f4<\/strong><\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><code>for<\/code>\u30eb\u30fc\u30d7\u3092\u5909\u66f4\u3057\u3001<code>cublasSgemm<\/code>\u95a2\u6570\u3092\u547c\u3073\u51fa\u3059\u3088\u3046\u306b\u3057\u307e\u3059\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong>\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u8d77\u52d5<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<pre class=\"wp-block-code\"><code>bash compile.sh\n.\/a.out 0 1000<\/code><\/pre>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong>\u30d1\u30e9\u30e1\u30fc\u30bf\u306e\u5909\u66f4\u3068\u7d50\u679c\u306e\u6bd4\u8f03<\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<ul class=\"wp-block-list\">\n<li><strong>\u884c\u5217\u306e\u30b5\u30a4\u30ba<\/strong>\u3092\u5909\u66f4\u3057\u3066\u30c6\u30b9\u30c8\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li><strong>\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u30b5\u30a4\u30ba <code>TILE_SIZE<\/code><\/strong> \u3092\u5909\u66f4\u3057\u3066\u30c6\u30b9\u30c8\u3057\u307e\u3059\uff08\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u3082\u6b63\u65b9\u5f62\uff09\u3002<\/li>\n<\/ul>\n<\/div><\/div>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading\">\u30ec\u30dd\u30fc\u30c8\u306e\u8981\u4ef6<\/h3>\n\n\n\n<p>Lab4\uff5eLab5\u3092\u5b8c\u4e86\u3057\u3001\u4ee5\u4e0b\u306e\u8981\u4ef6\u306b\u5f93\u3063\u3066\u30ec\u30dd\u30fc\u30c8\u3092\u4f5c\u6210\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br>\u30ec\u30dd\u30fc\u30c8\u306e\u5f62\u5f0f\u306b\u306f\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u306f\u3042\u308a\u307e\u305b\u3093\u304c\u3001\u6b21\u306e\u5185\u5bb9\u3092\u542b\u3081\u308b\u3088\u3046\u306b\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>Labs\u306e\u30d7\u30ed\u30bb\u30b9<\/strong><\/p>\n\n\n\n<p>\u56f3\u8868\u3092\u7d44\u307f\u5408\u308f\u305b\u3001\u6587\u7ae0\u5f62\u5f0f\u3067\u8868\u73fe\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>\u30c6\u30b9\u30c8\u7d50\u679c\u3068\u539f\u7406\u306e\u5206\u6790<\/strong><\/p>\n\n\n\n<p>\u56f3\u8868\u3001\u6587\u7ae0\u3001\u30bd\u30fc\u30b9\u30b3\u30fc\u30c9\u3092\u7528\u3044\u3066\u4ee5\u4e0b\u3092\u6bd4\u8f03\u5206\u6790\u3057\u3066\u304f\u3060\u3055\u3044\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u5165\u529b\u884c\u5217\u306e\u30b5\u30a4\u30ba<\/strong>\u304a\u3088\u3073<strong>\u30b9\u30ec\u30c3\u30c9\u30d6\u30ed\u30c3\u30af\u306e\u30b5\u30a4\u30ba<\/strong>\u304cCUDA\u884c\u5217\u7a4d\u8a08\u7b97\u7d50\u679c\u306b\u4e0e\u3048\u308b\u5f71\u97ff\u3092\u5206\u6790\u3001\u305d\u306e\u539f\u7406\u3092\u8aac\u660e\u3057\u306a\u3055\u3044\u3002<\/li>\n\n\n\n<li><strong>\u5171\u6709\u30e1\u30e2\u30ea\u6700\u9069\u5316\u524d\u5f8c\u306e\u5b9f\u884c\u7d50\u679c<\/strong>\u3092\u6bd4\u8f03\u3057\u3001\u95a2\u9023\u3059\u308b\u539f\u7406\u3092\u8a73\u3057\u304f\u8aac\u660e\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/li>\n<\/ul>\n\n\n\n<p>\u30ec\u30dd\u30fc\u30c8\u3092<strong>PDF\u5f62\u5f0f<\/strong>\u3067\u3001\u5bbf\u984c\u30b7\u30b9\u30c6\u30e0\u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u30bd\u30fc\u30b9\u30d5\u30a1\u30a4\u30eb\uff081\u3064\u3001\u307e\u305f\u306f2\u3064\uff09\u3092<code>.zip<\/code>\u5f62\u5f0f\u306b\u5727\u7e2e\u3057\u3001\u5bbf\u984c\u30b7\u30b9\u30c6\u30e0\u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>1\u3064\u306e\u30d5\u30a1\u30a4\u30eb\u306b\u5b9f\u9a134\u3068\u5b9f\u9a135\u306e\u30b3\u30fc\u30c9\u304c\u542b\u307e\u308c\u3066\u3044\u308b\u5834\u5408<\/strong>\u306f\u305d\u306e1\u3064\u306e\u30d5\u30a1\u30a4\u30eb\u3060\u3051\u3092\u63d0\u51fa\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br><strong>\u5225\u3005\u306e\u30d5\u30a1\u30a4\u30eb\u3067\u5b9f\u88c5\u3055\u308c\u3066\u3044\u308b\u5834\u5408<\/strong>\u306f\u3001\uff12\u3064\u306e\u30d5\u30a1\u30a4\u30eb\u3092\u63d0\u51fa\u3057\u3066\u3067\u3082\u3044\u3044\u3002<\/p>\n\n\n\n<div class=\"swell-block-tab is-style-default\" data-width-pc=\"auto\" data-width-sp=\"auto\"><ul class=\"c-tabList\" role=\"tablist\"><li class=\"c-tabList__item\" role=\"presentation\"><button role=\"tab\" class=\"c-tabList__button\" aria-selected=\"true\" aria-controls=\"tab-b2d3d5e4-0\" data-onclick=\"tabControl\">\uff11\u3064\u30d5\u30a1\u30a4\u30eb<\/button><\/li><li class=\"c-tabList__item\" role=\"presentation\"><button role=\"tab\" class=\"c-tabList__button\" aria-selected=\"false\" aria-controls=\"tab-b2d3d5e4-1\" data-onclick=\"tabControl\">2\u3064\u30d5\u30a1\u30a4\u30eb<\/button><\/li><\/ul><div class=\"c-tabBody\">\n<div id=\"tab-b2d3d5e4-0\" class=\"c-tabBody__item\" aria-hidden=\"false\">\n<ul class=\"wp-block-list\">\n<li><strong>Lab4~5<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li><code>matrix_mul.cu<\/code><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/div>\n\n\n\n<div id=\"tab-b2d3d5e4-1\" class=\"c-tabBody__item\" aria-hidden=\"true\">\n<ul class=\"wp-block-list\">\n<li><strong>Lab4<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li><code>matrix_mul4.cu<\/code><\/li>\n<\/ul>\n<\/li>\n\n\n\n<li><strong>Lab5<\/strong>\uff1a\n<ul class=\"wp-block-list\">\n<li><code>matrix_mul5.cu<\/code><\/li>\n<\/ul>\n<\/li>\n<\/ul>\n<\/div>\n<\/div><\/div>\n\n\n\n<h2 class=\"wp-block-heading\">Lab6 LLaMA\u306e\u6700\u9069\u5316<\/h2>\n\n\n\n<p class=\"is-style-big_icon_hatena\">Open Project<\/p>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u76ee\u7684<\/h3>\n\n\n\n<ol class=\"wp-block-list\">\n<li><strong>\u5b66\u3093\u3060\u884c\u5217\u7a4d\u6700\u9069\u5316\u624b\u6cd5\u3092\u7dcf\u5408\u7684\u306b\u6d3b\u7528\u3057\u3001\u4e2d\u898f\u6a21\u30e2\u30c7\u30eb\u306e\u63a8\u8ad6\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u65b9\u6cd5\u3092\u7fd2\u5f97\u3059\u308b\u3002<\/strong><\/li>\n\n\n\n<li><strong>\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u95a2\u9023\u3059\u308b\u539f\u7406\u3068\u6280\u8853\u304c\u5b9f\u969b\u306e\u30a2\u30d7\u30ea\u30b1\u30fc\u30b7\u30e7\u30f3\u3067\u3069\u306e\u3088\u3046\u306b\u6d3b\u7528\u3055\u308c\u308b\u304b\u3092\u7406\u89e3\u3059\u308b\u3002<\/strong><\/li>\n<\/ol>\n\n\n\n<h3 class=\"wp-block-heading\">Lab\u5185\u5bb9<\/h3>\n\n\n\n<p>Llama2\u306f\u3001Transformer\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3\u306b\u57fa\u3065\u304f\u8efd\u91cf\u30aa\u30fc\u30d7\u30f3\u30bd\u30fc\u30b9\u306e\u81ea\u7136\u8a00\u8a9e\u51e6\u7406\u30e2\u30c7\u30eb\u3067\u3059\u3002\u63a8\u8ad6\u3068\u91cf\u5b50\u5316\u306f\u5358\u4e00\u306eC\u30d7\u30ed\u30b0\u30e9\u30e0\u3067\u5b9f\u88c5\u3055\u308c\u3001\u8c4a\u5bcc\u306a\u30b3\u30e1\u30f3\u30c8\u304c\u542b\u307e\u308c\u3066\u304a\u308a\u3001\u79fb\u690d\u3001\u5c55\u958b\u3001\u6700\u9069\u5316\u304c\u5bb9\u6613\u3067\u3059\u3002<\/p>\n\n\n\n<p>\u3053\u306e\u5b9f\u9a13\u306f\u81ea\u7531\u5ea6\u306e\u9ad8\u3044\u30aa\u30fc\u30d7\u30f3\u5f62\u5f0f\u306e\u5185\u5bb9\u3067\u3059\u3002\u3053\u308c\u307e\u3067\u306e\u5b9f\u9a13\u3067\u5b66\u3093\u3060\u884c\u5217\u7a4d\u6700\u9069\u5316\u624b\u6cd5\u304b\u3089\u81ea\u7531\u306b\u9078\u629e\u304a\u3088\u3073\u7d44\u307f\u5408\u308f\u305b\u3092\u884c\u3044\u3001\u6700\u9069\u5316\u3055\u308c\u305f\u884c\u5217\u7a4d\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092Llama2\u30e2\u30c7\u30eb\u306b\u7d71\u5408\u3059\u308b\u3053\u3068\u3067\u3001\u63a8\u8ad6\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u5177\u4f53\u7684\u306a\u30bf\u30b9\u30af\u306f\u4ee5\u4e0b\u306e\u901a\u308a\u3067\u3059\uff1a<\/p>\n\n\n\n<ol class=\"wp-block-list\">\n<li>Llama2\u306e\u63a8\u8ad6\u30d7\u30ed\u30b0\u30e9\u30e0\u3092\u8aad\u307f\u3001\u884c\u5217\u7a4d\u306e\u5b9f\u88c5\u65b9\u6cd5\u3092\u7406\u89e3\u3057\u3001\u5206\u6790\u3084\u30c7\u30d0\u30c3\u30b0\u306b\u3088\u3063\u3066\u884c\u5217\u7a4d\u6f14\u7b97\u306e\u30c7\u30fc\u30bf\u898f\u6a21\u3092\u628a\u63e1\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u6700\u9069\u5316\u624b\u6cd5\u30921\u3064\u9078\u629e\u3059\u308b\u3001\u307e\u305f\u306f\u8907\u6570\u306e\u6700\u9069\u5316\u624b\u6cd5\u3092\u7d44\u307f\u5408\u308f\u305b\u3066\u3001Llama2\u306e\u884c\u5217\u7a4d\u6027\u80fd\u3092\u5411\u4e0a\u3055\u305b\u308b\u3002<\/li>\n\n\n\n<li>\u7570\u306a\u308b\u30b5\u30a4\u30ba\u306eLlama2\u30e2\u30c7\u30eb\u3092\u5b9f\u884c\u3057\u3001\u6700\u9069\u5316\u524d\u5f8c\u306e\u63a8\u8ad6\u6027\u80fd\u3092\u6bd4\u8f03\u3057\u3001\u5408\u7406\u7684\u306b\u5206\u6790\u3059\u308b\u3002<\/li>\n\n\n\n<li>\u6700\u9069\u5316\u5f8c\u306e\u63a8\u8ad6\u901f\u5ea6\u304c\u6700\u9069\u5316\u524d\u3068\u6bd4\u8f03\u3057\u3066\u5b89\u5b9a\u3057\u305f\u6027\u80fd\u5411\u4e0a\u3092\u793a\u3057\u305f\u5834\u5408\u3001\u8ffd\u52a0\u5f97\u70b9\u304c\u5f97\u3089\u308c\u307e\u3059\uff08\u5f97\u70b9\u306f\u6700\u9069\u5316\u52b9\u679c\u306b\u57fa\u3065\u3044\u3066\u5224\u65ad\uff09\u3002<\/li>\n<\/ol>\n\n\n\n<div class=\"swell-block-step\" data-num-style=\"circle\">\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong>\u5b9f\u9a13\u30d1\u30c3\u30b1\u30fc\u30b8\u306e\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3068\u89e3\u51cd<\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u5b9f\u884c\u3059\u308b\u524d\u306b\u3001\u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u304c\u30a4\u30f3\u30bf\u30fc\u30cd\u30c3\u30c8\u306b\u6b63\u5e38\u306b\u63a5\u7d9a\u3055\u308c\u3066\u3044\u308b\u3053\u3068\u3092\u78ba\u8a8d\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br><code>llama2.c.tar.gz<\/code> \u3001<code>stories15M.bin<\/code>\u3001<code>stories42M.bin<\/code>\u3001<code>stories110M.bin<\/code>\u3092<a href=\"https:\/\/github.com\/karpathy\/llama2.c\">GitHub<\/a>\u3068HuggingFace(<a href=\"https:\/\/huggingface.co\/karpathy\/tinyllamas\/resolve\/main\/stories15M.bin\">stories15M.bin<\/a>, <a href=\"https:\/\/huggingface.co\/karpathy\/tinyllamas\/resolve\/main\/stories42M.bin\">stories42M.bin<\/a>, <a href=\"https:\/\/huggingface.co\/karpathy\/tinyllamas\/resolve\/main\/stories110M.bin\">stories110M.bin<\/a>)\u304b\u3089\u53c8\u306f<a href=\"http:\/\/10.249.10.96:3011\/lab6\/\">\u30ad\u30e3\u30f3\u30d1\u30b9\u30cd\u30c3\u30c8<\/a>\u304b\u3089\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3057\u3001\u30e6\u30fc\u30b6\u30fc\u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u30b3\u30d4\u30fc\u3057\u3066\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<p><code>tar -zxvf llama2.c.tar.gz<\/code><br>\u30c0\u30a6\u30f3\u30ed\u30fc\u30c9\u3057\u305f <code>.bin<\/code>\u3092\u89e3\u51cd\u5f8c\u306e <code>llama2.c\/<\/code> \u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u914d\u7f6e\u3057\u307e\u3059\u3002<br><code>mv stories15M.bin\u3000llama2.c<\/code><br><code>mv stories42M.bin\u3000llama2.c<\/code><br><code>mv stories110M.bin\u3000llama2.c<\/code><\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>Llama2\u30e2\u30c7\u30ebCPU Baseline\u3092\u5b9f\u884c\u3059\u308b<\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u30bf\u30fc\u30df\u30ca\u30eb\u3092\u958b\u304d\u3001<code>llama2.c\/<\/code> \u30c7\u30a3\u30ec\u30af\u30c8\u30ea\u306b\u79fb\u52d5\u3057\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3067\u30b3\u30fc\u30c9\u3092\u30b3\u30f3\u30d1\u30a4\u30eb\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>make run<\/code><\/pre>\n\n\n\n<p>\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3092\u5b9f\u884c\u3057\u3066Llama2\u30e2\u30c7\u30eb\u3092\u52d5\u4f5c\u3055\u305b\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>.\/run stories15M.bin<\/code><\/pre>\n\n\n\n<p>\u6b21\u306b\u3001\u7570\u306a\u308b\u30b5\u30a4\u30ba\u306e\u30e2\u30c7\u30eb\u3092\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<pre class=\"wp-block-code\"><code>.\/run stories42M.bin\n.\/run stories110M.bin<\/code><\/pre>\n\n\n\n<p>\u30e2\u30c7\u30eb\u30b5\u30a4\u30ba\u306b\u5fdc\u3058\u305f\u5b9f\u884c\u6642\u9593\u306e\u9055\u3044\u3092\u89b3\u5bdf\u3057\u306a\u3055\u3044\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong><strong><strong><strong><strong>\u884c\u5217\u7a4d\u306e\u5b9f\u88c5\u3092\u78ba\u8a8d\u3059\u308b<\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>Llama2\u306e\u63a8\u8ad6\u51e6\u7406\u306f <code>run.c<\/code>\u306b\u5b9f\u88c5\u3055\u308c\u3066\u3044\u307e\u3059\u3002<\/p>\n\n\n\n<p>\u884c\u5217\u7a4d\u95a2\u6570\u306f <code><a href=\"https:\/\/github.com\/karpathy\/llama2.c\/blob\/master\/run.c#L217\">matmul<\/a><\/code> \u306b\u3042\u308a\u3002\u30b3\u30e1\u30f3\u30c8\u306b\u3088\u308b\u3068\u3001<code>matmul<\/code> \u95a2\u6570\u306f $d \\times n$ \u306e\u884c\u5217 $W$ \u3068\u6b21\u5143 $n$ \u306e\u5217\u30d9\u30af\u30c8\u30eb $x$ \u306e\u7a4d\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002\u307e\u305f\u3001\u63a8\u8ad6\u51e6\u7406\u306e\u6027\u80fd\u30dc\u30c8\u30eb\u30cd\u30c3\u30af\u306f\u3053\u306e\u884c\u5217\u7a4d\u95a2\u6570\u306b\u3042\u308b\u3053\u3068\u304c\u793a\u3055\u308c\u3066\u3044\u307e\u3059\uff08\u8208\u5473\u304c\u3042\u308b\u5834\u5408\u306f\u3001\u30c7\u30d0\u30c3\u30b0\u3067\u78ba\u8a8d\u53ef\u80fd\u3067\u3059\uff09\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong><strong><strong><strong>\u884c\u5217\u7a4d\u306e\u30c7\u30fc\u30bf\u898f\u6a21\u3092\u78ba\u8a8d\u3059\u308b<\/strong><\/strong><\/strong><\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u304a\u597d\u304d\u306a\u65b9\u6cd5\u3067\uff08\u30b3\u30fc\u30c9\u5185\u306b <code>printf<\/code> \u6587\u3092\u8ffd\u52a0\u3059\u308b\u3001\u307e\u305f\u306f GDB \u306a\u3069\u306e\u30c7\u30d0\u30c3\u30b0\u30c4\u30fc\u30eb\u3092\u4f7f\u7528\u3057\u3066\uff09<code>matmul<\/code> \u95a2\u6570\u306e\u5165\u529b\u884c\u5217\u306e\u30b5\u30a4\u30ba\u3092\u78ba\u8a8d\u3057\u306a\u3055\u3044\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u597d\u304d\u306a\u6700\u9069\u5316\u65b9\u6cd5\u3092\u9078\u3093\u3067<code>matmul<\/code>\u95a2\u6570\u3092\u6700\u9069\u5316\u3059\u308b<\/strong><\/div><div class=\"swell-block-step__body\">\n<p>\u3053\u308c\u307e\u3067\u306e\u5b9f\u9a13\u3067\u5b9f\u88c5\u3057\u305f\u884c\u5217\u7a4d\u6700\u9069\u5316\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3092\u5909\u66f4\u3057\u3001<code>run.c<\/code> \u53c8\u306f<code>run.cu<\/code>\u306b\u7d71\u5408\u3057\u307e\u3059\u3002<br>\u30b3\u30fc\u30c9\u3092\u5909\u66f4\u3057\u305f\u5f8c\u3001\u4ee5\u4e0b\u306e\u30b3\u30de\u30f3\u30c9\u3067\u30b3\u30f3\u30d1\u30a4\u30eb\u3057\u3066\u5b9f\u884c\u3057\u307e\u3059\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>CUDA\u6700\u9069\u5316\u3092\u4f7f\u7528\u3059\u308b\u5834\u5408<\/strong>\uff1a<code>nvcc -O3 -o run run.cu -lm<\/code><\/li>\n\n\n\n<li><strong>cuBLAS\u30e9\u30a4\u30d6\u30e9\u30ea\u3092\u4f7f\u7528\u3059\u308b\u5834\u5408<\/strong>\uff1a<code>nvcc -O3 -o run run.cu -lm -L\/usr\/local\/lib64 -lcublas<\/code><\/li>\n\n\n\n<li><strong>OpenMP\u3092\u4f7f\u7528\u3059\u308b\u5834\u5408<\/strong>\uff1a<code>gcc -O3 -o run run.c -lm -fopenmp<\/code><\/li>\n\n\n\n<li><strong>AVX\u306a\u3069\u4ed6\u306eCPU\u5411\u3051\u306e\u6700\u9069\u5316\u624b\u6cd5\u3092\u4f7f\u7528\u3059\u308b\u5834\u5408<\/strong>\uff1a<code>make run<\/code><\/li>\n<\/ul>\n\n\n\n<p class=\"is-style-icon_info\">\u6700\u9069\u5316\u65b9\u6cd5\uff08\u4e3b\u306bGPU\u5411\u3051\u306e\u6700\u9069\u5316\uff09\u306b\u3088\u3063\u3066\u3001\u4e0a\u8a18\u306e\u547d\u4ee4\u306f\u3046\u307e\u304f\u52d5\u4f5c\u3067\u304d\u306a\u3044\u306e\u53ef\u80fd\u6027\u304c\u3042\u308b\u3088\u3046\u3067\u3059\u3002\u30b3\u30f3\u30d1\u30a4\u30eb\u5931\u6557\u304c\u767a\u751f\u3057\u305f\u5834\u5408\u3001\u30cd\u30c3\u30c8\u3067\u89e3\u6c7a\u7b56\u3092\u63a2\u3057\u306a\u3055\u3044\u3002<br><br>\u4f8b\u3048\u3070<code>nvcc<\/code>\u3067\u3001\u30d5\u30e9\u30b0<code>-std=c++17<\/code>\u3001 <code>-arch=compute_86<\/code>\u3092\u6dfb\u4ed8\u3001\u30ea\u30f3\u30ab\u30fc\u30d1\u30b9\u3092<code>-L\/usr\/local\/cuda\/lib64<\/code>\u3078\u8a02\u6b63\u306a\u3069\u306e\u5834\u5408\u304c\u3042\u308b\u3002<\/p>\n<\/div><\/div>\n\n\n\n<div class=\"swell-block-step__item\"><div class=\"swell-block-step__number u-bg-main\"><span class=\"__label\">STEP<\/span><\/div><div class=\"swell-block-step__title u-fz-l\"><strong>\u6700\u9069\u5316\u524d\u5f8c\u306e\u63a8\u8ad6\u6027\u80fd\u3092\u6bd4\u8f03\u30fb\u5206\u6790\u3059\u308b<\/strong><\/div><div class=\"swell-block-step__body\">\n<ol class=\"wp-block-list\">\n<li>\u5404 <code>.bin<\/code> \u30d5\u30a1\u30a4\u30eb\u3092\u4f7f\u7528\u3057\u3066\u6700\u9069\u5316\u524d\u5f8c\u306eLlama2\u30e2\u30c7\u30eb\u3092\u305d\u308c\u305e\u308c3\uff5e5\u56de\u5b9f\u884c\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u63a8\u8ad6\u901f\u5ea6\u306e\u5e73\u5747\u5024\u3092\u8a18\u9332\u3057\u3001\u6700\u9069\u5316\u524d\u306e\u7d50\u679c\u3068\u6bd4\u8f03\u3057\u307e\u3059\u3002<\/li>\n\n\n\n<li>\u7d50\u679c\u3092\u5206\u6790\u3057\u3001\u3055\u3089\u306b\u6539\u5584\u306e\u53ef\u80fd\u6027\u3092\u691c\u8a0e\u3057\u307e\u3059\u3002<\/li>\n<\/ol>\n\n\n\n<p>\u4ee5\u4e0b\u306fCUDA\u6700\u9069\u5316\u3092\u4f7f\u7528\u3057\u305f\u4f8b\u3067\u3059\uff1a<\/p>\n<\/div><\/div>\n<\/div>\n\n\n\n<h3 class=\"wp-block-heading\">\u30ec\u30dd\u30fc\u30c8\u306e\u8981\u4ef6<\/h3>\n\n\n\n<p>Lab6\u3092\u5b8c\u4e86\u3057\u3001\u4ee5\u4e0b\u306e\u8981\u4ef6\u306b\u5f93\u3063\u3066\u30ec\u30dd\u30fc\u30c8\u3092\u4f5c\u6210\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br>\u30ec\u30dd\u30fc\u30c8\u306e\u5f62\u5f0f\u306b\u306f\u30c6\u30f3\u30d7\u30ec\u30fc\u30c8\u306f\u3042\u308a\u307e\u305b\u3093\u304c\u3001\u6b21\u306e\u5185\u5bb9\u3092\u542b\u3081\u308b\u3088\u3046\u306b\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>Lab\u306e\u30d7\u30ed\u30bb\u30b9<\/strong><\/p>\n\n\n\n<p>\u56f3\u8868\u3092\u7d44\u307f\u5408\u308f\u305b\u3001\u6587\u7ae0\u5f62\u5f0f\u3067\u8868\u73fe\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p><strong>\u30c6\u30b9\u30c8\u7d50\u679c\u3068\u539f\u7406\u306e\u5206\u6790<\/strong><\/p>\n\n\n\n<p>\u56f3\u8868\u3001\u6587\u7ae0\u3001\u30bd\u30fc\u30b9\u30b3\u30fc\u30c9\u3092\u7528\u3044\u3066\u4ee5\u4e0b\u3092\u6bd4\u8f03\u5206\u6790\u3057\u3066\u304f\u3060\u3055\u3044\u3002<br><strong>\u63a1\u7528\u3057\u305f\u6700\u9069\u5316\u624b\u6cd5\u306e\u539f\u7406<\/strong>\u306b\u3064\u3044\u3066\u3082\u8a73\u3057\u304f\u8aac\u660e\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p>\u30ec\u30dd\u30fc\u30c8\u3092<strong>PDF\u5f62\u5f0f<\/strong>\u3067\u3001\u5bbf\u984c\u30b7\u30b9\u30c6\u30e0\u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p>Lab\u4e2d\u306b\u5909\u66f4\u3057\u305f\u30d5\u30a1\u30a4\u30eb\u3092 <code>.zip<\/code> \u5f62\u5f0f\u306b\u5727\u7e2e\u3057\u3001\u5bbf\u984c\u30b7\u30b9\u30c6\u30e0\u306b\u30a2\u30c3\u30d7\u30ed\u30fc\u30c9\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n\n\n\n<p class=\"has-text-align-center is-style-kakko_box\">\u4ee5\u4e0a\u3067\u3059<\/p>\n\n\n\n<p class=\"is-style-icon_good\">\u53c2\u8003\u3092\u7528\u3044\u3066\u89e3\u7b54\u306f\u3001\u3053\u3061\u3089\u3078\u3002\u21e9<\/p>\n\n\n<div class=\"swell-block-postLink\">\t\t\t<div class=\"p-blogCard -external\" data-type=\"type3\" data-onclick=\"clickLink\">\n\t\t\t\t<div class=\"p-blogCard__inner\">\n\t\t\t\t\t<span class=\"p-blogCard__caption\">\u7d42\u308f\u3089\u306a\u3044\u79cb\u98a8<\/span>\n\t\t\t\t\t<div class=\"p-blogCard__thumb c-postThumb\"><figure class=\"c-postThumb__figure\"><img decoding=\"async\" src=\"https:\/\/www.yanagichiaki.jp\/wp-content\/uploads\/2024\/06\/logo.png\" alt=\"\" class=\"c-postThumb__img u-obf-cover\" width=\"320\" height=\"180\"><\/figure><\/div>\t\t\t\t\t<div class=\"p-blogCard__body\">\n\t\t\t\t\t\t<a class=\"p-blogCard__title\" href=\"https:\/\/www.yanagichiaki.jp\/index.php\/2024\/11\/14\/computer-architecture-labs\/\" target=\"_blank\" rel=\"noopener noreferrer\">Computer Architecture Labs | \u7d42\u308f\u3089\u306a\u3044\u79cb\u98a8<\/a>\n\t\t\t\t\t\t<span class=\"p-blogCard__excerpt\">\u30cf\u30eb\u30d3\u30f3\u5de5\u696d\u5927\u5b66\uff08\u6df1\u5733\uff09\u2022 2024 \u2022 \u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30fb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3 Lab\u2022 \u306b\u304a\u3051\u308b\u89e3\u6c7a\u7b56 \u2022 HITSZ \u8ba1\u7b97\u673a\u4f53\u7cfb\u7ed3\u6784\u5b9e\u9a8c 2024 \u5fa1\u8cea\u554f\u304c\u5fa1\u5ea7\u3044\u307e\u3057\u305f\u3089\u3001\u3053\u306e\u30da\u30fc\u30b8\u306e\u4e0b\u90e8\u306b\u3042\u308b&#8230;<\/span>\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t<\/div>\n\t\t<\/div>\n\n\n\n","protected":false},"excerpt":{"rendered":"<p>\u30cf\u30eb\u30d3\u30f3\u5de5\u696d\u5927\u5b66\uff08\u6df1\u5733\uff09\u2022 2024 \u2022 \u30b3\u30f3\u30d4\u30e5\u30fc\u30bf\u30fb\u30a2\u30fc\u30ad\u30c6\u30af\u30c1\u30e3 Lab Guidebook 2024<br \/>\nHITSZ \u8ba1\u7b97\u673a\u4f53\u7cfb\u7ed3\u6784\u5b9e\u9a8c\u6307\u5bfc\u4e66 2024<\/p>\n","protected":false},"author":1,"featured_media":1758,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"swell_btn_cv_data":"","_jetpack_memberships_contains_paid_content":false,"footnotes":""},"categories":[90,9],"tags":[],"class_list":["post-1688","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-90","category-computerscience"],"jetpack_sharing_enabled":true,"jetpack_featured_media_url":"https:\/\/yanagichiaki.jp\/wp-content\/uploads\/2024\/12\/CALabGuideBookCover.png","_links":{"self":[{"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/posts\/1688","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/comments?post=1688"}],"version-history":[{"count":41,"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/posts\/1688\/revisions"}],"predecessor-version":[{"id":1896,"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/posts\/1688\/revisions\/1896"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/media\/1758"}],"wp:attachment":[{"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/media?parent=1688"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/categories?post=1688"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/yanagichiaki.jp\/index.php\/wp-json\/wp\/v2\/tags?post=1688"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}