{"id":45559,"date":"2023-08-08T11:20:58","date_gmt":"2023-03-06T05:30:33","guid":{"rendered":"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/"},"modified":"2024-04-30T17:32:40","modified_gmt":"2024-04-30T09:32:40","slug":"45559-2","status":"publish","type":"post","link":"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/","title":{"rendered":""},"content":{"rendered":"

(\u3053\u306e\u8a18\u4e8b\u306f\u300cfukuoka.ex Elixir\uff0fPhoenix Advent Calendar 2018\u300d\u306e9\u65e5\u76ee\u3067\u3059)
\n\u6628\u65e5\u306e\u8a18\u4e8b\u306f @mocamocaland \u3055\u3093\u306e\u300cPhoenix\u3092GCP\u306b\u30c7\u30d7\u30ed\u30a4\u3059\u308b\u65b9\u6cd5\u300d\u3067\u3057\u305f\uff0e<\/p>\n

\n

\u3069\u3046\u3082fukuoka.ex\u30ad\u30e3\u30b9\u30c8\u306e@hisaway\u3067\u3059\uff0e<\/p>\n

Elixir\u304b\u3089Rustler(NIF)\u7d4c\u7531\u3067Rust\u306e\u30b3\u30fc\u30c9\u3092\u547c\u3073\u51fa\u3057\u305f\u3089\uff0c\u3069\u308c\u3060\u3051\u30b3\u30fc\u30c9\u304c\u901f\u304f\u306a\u308b\u304b\uff0c\u7dda\u5f62\u56de\u5e30\u3067\u691c\u8a3c\u3057\u307e\u3057\u305f\uff0e @zacky1972 \u3055\u3093\u306e\u300cZEAM\u958b\u767a\u30ed\u30b0\u300d\uff0c\u7279\u306bHastega\u30b7\u30ea\u30fc\u30ba\u306e\u5148\u884c\u63a2\u691c\u968a\u306a\u4f4d\u7f6e\u4ed8\u3051\u3067\u3059. Hastega\u3068\u3044\u3046\u306e\u306fElixir\u304b\u3089SIMD:\u30de\u30eb\u30c1\u30b3\u30a2CPU\/GPU \u4e26\u5217\u51e6\u7406\u3092\u884c\u3046\u6a5f\u80fd\u306e\u7dcf\u79f0\u3067\u3059\uff0e<\/p>\n

\u3088\u308a\u8a73\u3057\u3044Hastega\u306b\u3064\u3044\u3066\u306fZEAM\u958b\u767a\u30ed\u30b0: Elixir \u30de\u30af\u30ed + LLVM \u3067\u8d85\u4e26\u5217\u30d7\u30ed\u30b0\u30e9\u30df\u30f3\u30b0\u51e6\u7406\u7cfb\u3092\u7814\u7a76\u958b\u767a\u4e2d\u3092\u3054\u89a7\u304f\u3060\u3055\u3044\uff0e<\/p>\n

(\u30bf\u30b0\u304cRust\u306a\u306e\u304c\u304d\u306b\u306a\u3063\u3066\u3057\u3087\u3046\u304c\u306a\u3044\u3067\u3059\uff0e\u30bf\u30a4\u30c8\u30eb\u4ee5\u5916\u306fRust\u306b\u7d71\u4e00\u3057\u3066\u307e\u3059)<\/p>\n

\u6982\u8981<\/h1>\n

Elixir\u306e\u901f\u3055<\/h2>\n

Elixir\u306fErlangVM(BEAM)\u4e0a\u3067\u52d5\u304f\u4ed5\u69d8\u3067\u3042\u308b\u305f\u3081\uff0c\u30cd\u30a4\u30c6\u30a3\u30d6\u30b3\u30fc\u30c9\u3067\u306f\u306a\u304f\uff0c\u30d0\u30a4\u30c8\u30b3\u30fc\u30c9\u3092\u51fa\u529b\u3057\u307e\u3059\uff0e
\nPython, Ruby\u3068\u6bd4\u3079\u308b\u3068\u901f\u3044\u3093\u3067\u3059\u304c, \u6bd4\u8f03\u5bfe\u8c61\u306bC\u3084C++\u3092\u6319\u3052\u3066\u3057\u307e\u3046\u3068\u9045\u3044\u3067\u3059\uff0e<\/p>\n

HiPE\u3068\u3044\u3046\u30aa\u30d7\u30b7\u30e7\u30f3\u3092\u6709\u52b9\u306b\u3059\u308b\u3053\u3068\u3067\u30cd\u30a4\u30c6\u30a3\u30d6\u30b3\u30fc\u30c9\u3067\u30b3\u30f3\u30d1\u30a4\u30eb\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\u304c\uff0c\u901f\u5ea6\u3092\u91cd\u8996\u3059\u308b\u969b\u306b\u306f\uff0c\u3084\u306f\u308a\u5143\u304b\u3089\u30b9\u30d4\u30fc\u30c9\u306e\u3042\u308b\u8a00\u8a9e\u3092\u4f7f\u7528\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\uff0e
\n(HiPE\u306fErlang\u306e\u30b3\u30f3\u30d1\u30a4\u30eb\u30aa\u30d7\u30b7\u30e7\u30f3\u3067\u3059\uff0e)
\n|> @sile \u300cHiPE(High Performance Erlang)\u306b\u3064\u3044\u3066\u300d<\/p>\n

\u305d\u3053\u3067Rust\u306e\u529b\u3092\u501f\u308a\u307e\u3059\uff0eElixir\u304b\u3089Rust\u3092\u4f7f\u304a\u3046\u3063\u3066\u8a18\u4e8b\u306f\u5148\u4eba\u306e\u65b9\u304c\u6295\u7a3f\u3057\u3066\u304a\u308a\u307e\u3059\uff0e
\n|> @tatsuya6502 \u3055\u3093 \u300cElixir\u304b\u3089Rust\u306e\u95a2\u6570\u3092\u3064\u304b\u3046 \u2192 \u306f\u3084\u3044\u300d
\n|> @twinbee \u3055\u3093 \u300cElixir\u304b\u3089\u7c21\u5358\u306bRust\u3092\u547c\u3073\u51fa\u305b\u308bRustler #1 \u6e96\u5099\u7de8\u300d<\/p>\n

\u7528\u8a9e<\/h2>\n

NIF<\/h3>\n

Native Implemented Functions\u306e\u7565\u3067\uff0cElixir\u306e\u4e0b\u3067\u52d5\u3044\u3066\u3044\u308bErlangVM(BEAM)\u304b\u3089\u30cd\u30a4\u30c6\u30a3\u30d6\u30b3\u30fc\u30c9\u3092\u4f7f\u3046\u4ed5\u7d44\u307f\u306e\u3053\u3068\u3067\u3059\uff0e<\/p>\n

Rustler<\/h3>\n

NIF\u7d4c\u7531\u3067Rust\u306e\u95a2\u6570\u3092\u547c\u3073\u3060\u3059Elixir\u306e\u30e2\u30b8\u30e5\u30fc\u30eb\u3067\u3059\uff0eRust\u3067\u8a18\u8ff0\u3057\u305f\u95a2\u6570\u3068Elixir\u306e\u95a2\u6570\u3092\u5bfe\u5fdc\u3065\u3051\u3066\u8a18\u8ff0\u3059\u308b\u3053\u3068\u3067\u697d\u306b\u30b3\u30fc\u30c7\u30a3\u30f3\u30b0\u3059\u308b\u3053\u3068\u304c\u3067\u304d\u307e\u3059\uff0e<\/p>\n

Rustler_export_nifs!<\/span> {<\/span>\r\n  \"Elixir.LinearRegressorNif\"<\/span>,<\/span> \/\/ Elixir\u306e\u30e2\u30b8\u30e5\u30fc\u30eb\u540d\u3092\u6307\u5b9a<\/span>\r\n  [<\/span>\r\n    \/\/(\"Elixir's func-name, number of arguments, Rust's func)<\/span>\r\n    (<\/span>\"_fit\"<\/span>,<\/span> 5<\/span>,<\/span> nif_fit<\/span>),<\/span> \r\n  ],<\/span>\r\n  None<\/span>\r\n}<\/span>\r\n<\/code><\/pre>\n
SIMD<\/h3>\n
Single Instruction Multiple Data\u306e\u7565\u3067\u3059\uff0e\u305f\u304f\u3055\u3093\u306e\u30c7\u30fc\u30bf\u306b\u5bfe\u3057\u3066\uff0c\uff11\u3064\u3060\u3051\u51e6\u7406\u3092\u4e0e\u3048\u308b\u3088\u3046\u306a\u547d\u4ee4\u3092\u3044\u3044\u307e\u3059\uff0e\u300c\u3053\u306e\u30d9\u30af\u30c8\u30eb\u306e\u5168\u8981\u7d20\u30922\u500d\u3057\u3066\u3088\u300d\u307f\u305f\u3044\u306a\u611f\u3058\u3067\u3059\uff0e
\nRust\u306f\u30c7\u30d5\u30a9\u30eb\u30c8\u3067SIMD\u547d\u4ee4\u306b\u30b3\u30f3\u30d1\u30a4\u30eb\u3057\u3066\u304f\u308c\u307e\u3059\uff0e<\/p>\n
\u984c\u6750\u306b\u3064\u3044\u3066<\/h2>\n
\u9ad8\u901f\u5316\u3059\u308b\u5bfe\u8c61\u304c\u7dda\u5f62\u56de\u5e30\u306a\u306e\u306f\uff0cElixir\u3067\u6a5f\u68b0\u5b66\u7fd2\u3057\u3088\u3046\u3068\u3044\u3046\u8a71\u306b\u306a\u3063\u305f\u3068\u304d\uff0c\u53e4\u5178\u7684\u306a\u6a5f\u68b0\u5b66\u7fd2\u3067\u53d6\u308a\u7d44\u307f\u3084\u3059\u3044\u3068\u3057\u3066\uff0c @piacere_ex \u3055\u3093\u304cElixir\u3067\u66f8\u304b\u308c\u305f\u7dda\u5f62\u56de\u5e30\u306e\u30b3\u30fc\u30c9\u3092\u63d0\u4f9b\u3057\u3066\u304f\u3060\u3055\u3063\u305f\u3053\u3068\u304c\u7406\u7531\u3067\u3059\uff0e<\/p>\n
\u9ad8\u901f\u5316\u306e\u65b9\u91dd<\/h1>\n
\u4ee5\u4e0b\u306e\uff12\u3064\u306e\u65b9\u6cd5\u3067\u9ad8\u901f\u5316\u3092\u884c\u3044\u307e\u3057\u305f\uff0e\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u306e\u6700\u9069\u5316\u306f\u3057\u307e\u305b\u3093\uff0e
\n1. Rustler\u7d4c\u7531\u3067\u30cd\u30a4\u30c6\u30a3\u30d6\u30b3\u30fc\u30c9\u306e\u547c\u3073\u51fa\u3057\u53ca\u3073\uff0cNIF\u306e\u975e\u540c\u671f\u547c\u3073\u51fa\u3057
\n2. Rust \u306e Rayon\u3092\u4f7f\u7528\u3057\u305fCPU\u30de\u30eb\u30c1\u30b3\u30a2\u4e26\u5217\u5b9f\u884c<\/p>\n
Rayon\u306f\u9023\u7d9a\u3057\u3066\u8a08\u7b97\u3092\u3059\u308b\u30b3\u30fc\u30c9\u3092\u7c21\u5358\u306b\u4e26\u5217\u8a08\u7b97\u4ed5\u69d8\u306b\u3067\u304d\u308b\u30c7\u30fc\u30bf\u51e6\u7406\u30e9\u30a4\u30d6\u30e9\u30ea\u3067\u3059\uff0e
\n|> Crate Rayon<\/p>\n
\u5099\u8003<\/h2>\n
\u4eca\u56de\u306f\u7dda\u5f62\u56de\u5e30\u3092\u9ad8\u901f\u5316\u3055\u305b\u308b\u306e\u304c\u76ee\u7684\u3067\u306f\u306a\u304f\uff0cElixir\u306e\u30cd\u30a4\u30c6\u30a3\u30d6\u30b3\u30fc\u30c9\u5316\u306b\u3088\u308b\u901f\u5ea6\u5411\u4e0a\u3092\u8a55\u4fa1\u3059\u308b\u3053\u3068\u304c\u76ee\u7684\u3067\u3059\uff0e
\n\u3082\u3057\u5287\u7684\u306a\u9ad8\u901f\u5316\u3092\u56f3\u308b\u969b\u306f\uff0c\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u304b\u3089\u7d44\u307f\u76f4\u3057\u3066\u304f\u3060\u3055\u3044\u306d\uff0e<\/p>\n
\u52d5\u4f5c\u74b0\u5883<\/h1>\n
PC<\/h2>\n
    \n
  • \n
      MacBook Air(Early 2015)<\/ul>\n<\/li>\n<\/ul>\n
      Processor: 1.6 GHz Intel Core i5
      \nMemory: 8 GB 1600 MHz DDR3
      \nGraphics: Intel HD Graphics 6000 1536MB<\/p>\n
      iMac Pro(2017)<\/p>\n
      Processor: 2.3 GHz Intel Xeon W (\u30d5\u309a\u30ed\u30bb\u30c3\u30b5\u6570 1\uff0c\u7269\u7406\u30b3\u30a2 18\uff0c\u8ad6\u7406\u30b3\u30a2 36)
      \nMemory: 32 GB 2666 MHz DDR4
      \nGraphics: Radeon Pro Vega 64 16368MB<\/p>\n
      Mac Pro (Mid 2010)<\/p>\n
      Processor:
      \n2 x 3.46 GHz 6-Core Intel Xeon (\u30d7\u30ed\u30bb\u30c3\u30b5\u6570 2\uff0c\u7269\u7406\u30b3\u30a2\u6570 6 x 2\uff0c\u8ad6\u7406\u30b3\u30a2\u6570 12 x 2)
      \nMemory: 16 GB 1066 MHz DDR3
      \nGraphics: ATI Radeon HD 5770 1024MB
      \nOS : Sierra 10.12.6<\/p>\n
      GPGPU\u30b5\u30fc\u30d0\u30fc \u30e6\u30cb\u30c3\u30c8\u30b3\u30e0 UCGPU-E2630V4- 32GB<\/p>\n
      Processor: 2.20GHz Intel Xeon E5-2630 v4 (\u30d7\u30ed\u30bb\u30c3\u30b5\u6570 2 \u7269\u7406\u30b3\u30a2\u6570 10\u30b3\u30a2 x 2\uff0c\u8ad6\u7406\u30b3\u30a2\u6570 20\u30b3\u30a2 x 2)
      \nMemory: 32 GB 2400 MHz DDR4
      \nGraphics: NVIDIA GeForce GTX 1080 Ti
      \nOS : Linux Ubuntu 16.04<\/p>\n
      \u8a00\u8a9e<\/p>\n
      Elixir 1.7.3(Mac Pro\u306f 1.7.4\u3067\u3057\u305f)
      \nErlang\/OTP 21 [erts-10.1] [source] [64-bit] [smp:40:40] [ds:40:40:10] [async-threads:1] [hipe] [sharing-preserving]
      \nrustc 1.28.0
      \nRustler 0.18.0<\/p>\n
      \u6027\u80fd\u8a55\u4fa1<\/h1>\n
      inline: Elixir\u306e\u307f\uff0c\u4e00\u90e8\u6a5f\u80fd\u3092\u30a4\u30f3\u30e9\u30a4\u30f3\u5c55\u958b\u3057\u3066\u6700\u9069\u5316\u3057\u305f\u30b3\u30fc\u30c9<\/p>\n
      Rust: Rustler\u3092\u4f7f\u3063\u3066\uff0cElixir\u304b\u3089Rust\u306e\u30b3\u30fc\u30c9\u3092\u547c\u3073\u51fa\u3057<\/p>\n
      Rayon: Rust\u306e\u30b3\u30fc\u30c9\u306b\u52a0\u3048\u3066\u4e00\u90e8\u3092\u4e26\u5217\u51e6\u7406<\/p>\n
      inline\u3092\u57fa\u6e96\u3068\u3057\u3066\u500d\u7387\u3092\u6e2c\u5b9a\u3057\u3066\u3044\u307e\u3059\uff0e<\/p>\n
      MacBook Air(Early 2015)<\/h2>\n
      \n
      \ninlineRustRayon\u5b9f\u884c\u6642\u9593[s]7.2404214.8608357.394763\u500d\u7387(1.00)1.480.97<\/div>\n<\/div>\n
      Mac Pro(Mid 2010)<\/h2>\n
      \n
      \ninlineRustRayon\u5b9f\u884c\u6642\u9593[s]7.0667383.646148.535108\u500d\u7387(1.00)1.930.82<\/div>\n<\/div>\n
      iMac Pro(2017)<\/h2>\n
      \n
      \ninlineRustRayon\u5b9f\u884c\u6642\u9593[s]4.1594933.011757.962244\u500d\u7387(1.00)1.380.52<\/div>\n<\/div>\n
      GPGPU\u30b5\u30fc\u30d0\u30fc<\/h2>\n
      \n
      \ninlineRustRayon\u5b9f\u884c\u6642\u9593[s]6.6439133.1136066.690754\u500d\u7387(1.00)2.130.99<\/div>\n<\/div>\n
      \u8a55\u4fa1\u7dcf\u62ec<\/h2>\n
        \n
      • \n
          \u7d14\u7c8b\u306b\u30cd\u30a4\u30c6\u30a3\u30d6\u30b3\u30fc\u30c9\u306b\u843d\u3068\u3059\u3068\uff0c1.38 ~ 2.13 \u500d\u307b\u3069\u901f\u304f\u306a\u308a\u307e\u3057\u305f\uff0e<\/ul>\n<\/li>\n<\/ul>\n
           <\/p>\n
            \n
          • \n
              \u4e26\u5217\u51e6\u7406\u306b\u3059\u308b\u3068\uff0c0.52 ~ 0.99\u500d\u3068\u9045\u304f\u306a\u308a\u307e\u3057\u305f<\/ul>\n<\/li>\n<\/ul>\n
               <\/p>\n
                iMac Pro\u3060\u3051inline\u304c\u7570\u5e38\u306b\u901f\u3044(\u539f\u56e0\u306f\u672a\u8abf\u67fb)<\/ul>\n
                \u5b9f\u88c5<\/h1>\n
                \u7dda\u5f62\u56de\u5e30\u306e\u30e1\u30a4\u30f3\u3067\u3042\u308bfitting\u3092\u884c\u3046\u95a2\u6570\u3067\u3059\uff0e
                \n\u5927\u307e\u304b\u306a\u6d41\u308c\u306f\uff0c<\/p>\n
                  \n
                1. \n
                    NIF\u7d4c\u7531\u3067\u8d77\u52d5\u3055\u308c\u305f\u30b9\u30ec\u30c3\u30c9\u304b\u3089\u5225\u30b9\u30ec\u30c3\u30c9\u3092\u7acb\u3066\u308b\u6e96\u5099\u3092\u3059\u308b<\/ol>\n<\/li>\n<\/ol>\n
                     <\/p>\n
                      \n
                    1. \n
                        \u5143\u306e\u30b9\u30ec\u30c3\u30c9\u306f\u5236\u7d04\u304c\u3042\u308b\u306e\u3067\u3055\u3063\u3055\u3068\u7d42\u4e86\u3059\u308b<\/ol>\n<\/li>\n<\/ol>\n
                         <\/p>\n
                          \u5225\u30b9\u30ec\u30c3\u30c9\u306e\u65b9\u3067\u30e1\u30a4\u30f3\u306e\u8a08\u7b97\u3092\u3059\u308b<\/ol>\n
                          \u3068\u3044\u3063\u305f\u611f\u3058\u3067\u3059\uff0e<\/p>\n
                          \u3059\u3050\u304b\u306f\u308f\u304b\u308a\u307e\u305b\u3093\u304c\uff0c\u5f8c\u307b\u3069\u30ea\u30dd\u30b8\u30c8\u30ea\u3092\u516c\u958b\u3059\u308b\u4e88\u5b9a\u3067\u3059\uff01<\/p>\n
                          fn<\/span> nif_fit<\/span><<\/span>'a>(env: Env<'<\/span>a<\/span>><\/span>,<\/span> args:<\/span> &<\/span>[<\/span>Term<\/span><<\/span>'a>])-> NifResult<Term<'<\/span>a<\/span>>><\/span> {<\/span>\r\n  let<\/span> pid<\/span> =<\/span> env<\/span>.<\/span>pid<\/span>();<\/span>\r\n  let<\/span> mut<\/span> my_env<\/span> =<\/span> OwnedEnv<\/span>::<\/span>new<\/span>();<\/span>\r\n\r\n  let<\/span> saved_list<\/span> =<\/span> my_env<\/span>.<\/span>run<\/span>(<\/span>|<\/span>env<\/span>|<\/span> -><\/span> NifResult<\/span><<\/span>SavedTerm<\/span>><\/span> {<\/span>\r\n    let<\/span> _x<\/span> =<\/span> args<\/span>[<\/span>0<\/span>]<\/span>.<\/span>in_env<\/span>(<\/span>env<\/span>);<\/span>\r\n    let<\/span> _y<\/span> =<\/span> args<\/span>[<\/span>1<\/span>]<\/span>.<\/span>in_env<\/span>(<\/span>env<\/span>);<\/span>\r\n    let<\/span> theta<\/span> =<\/span> args<\/span>[<\/span>2<\/span>]<\/span>.<\/span>in_env<\/span>(<\/span>env<\/span>);<\/span>\r\n    let<\/span> alpha<\/span> =<\/span> args<\/span>[<\/span>3<\/span>]<\/span>.<\/span>in_env<\/span>(<\/span>env<\/span>);<\/span>\r\n    let<\/span> iterations<\/span> =<\/span> args<\/span>[<\/span>4<\/span>]<\/span>.<\/span>in_env<\/span>(<\/span>env<\/span>);<\/span>\r\n    Ok<\/span>(<\/span>my_env<\/span>.<\/span>save<\/span>(<\/span>make_tuple<\/span>(<\/span>env<\/span>,<\/span> &<\/span>[<\/span>_x<\/span>,<\/span> _y<\/span>,<\/span> theta<\/span>,<\/span> alpha<\/span>,<\/span> iterations<\/span>])))<\/span>\r\n  })<\/span>?;<\/span>\r\n\r\n  std<\/span>::<\/span>thread<\/span>::<\/span>spawn<\/span>(<\/span>move<\/span> ||<\/span>  {<\/span>\r\n    my_env<\/span>.<\/span>send_and_clear<\/span>(<\/span>&<\/span>pid<\/span>,<\/span> |<\/span>env<\/span>|<\/span> {<\/span>\r\n      let<\/span> result:<\/span> NifResult<\/span><<\/span>Term<\/span>><\/span> =<\/span> (<\/span>||<\/span> {<\/span>\r\n        let<\/span> tuple<\/span> =<\/span> saved_list<\/span>\r\n        .<\/span>load<\/span>(<\/span>env<\/span>)<\/span>.<\/span>decode<\/span>::<\/span><<\/span>(<\/span>\r\n          Term<\/span>,<\/span> \r\n          Term<\/span>,<\/span>\r\n          Term<\/span>,<\/span>\r\n          Num<\/span>,<\/span>\r\n          i64<\/span>)<\/span>><\/span>()<\/span>?;<\/span> \r\n\r\n        let<\/span> x:<\/span> Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span> =<\/span> try!<\/span>(<\/span>tuple<\/span>.<\/span>0<\/span>.<\/span>decode<\/span>());<\/span>\r\n        let<\/span> y:<\/span> Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span> =<\/span> try!<\/span>(<\/span>tuple<\/span>.<\/span>1<\/span>.<\/span>decode<\/span>());<\/span>\r\n        let<\/span> theta:<\/span> Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span> =<\/span> try!<\/span>(<\/span>tuple<\/span>.<\/span>2<\/span>.<\/span>decode<\/span>());<\/span>\r\n        let<\/span> alpha:<\/span> Num<\/span> =<\/span> tuple<\/span>.<\/span>3<\/span>;<\/span>\r\n        let<\/span> iterations:<\/span> i64<\/span> =<\/span> tuple<\/span>.<\/span>4<\/span>;<\/span>\r\n\r\n        let<\/span> tx<\/span> =<\/span> transpose<\/span>(<\/span>&<\/span>x<\/span>);<\/span>\r\n        let<\/span> m<\/span> =<\/span> y<\/span>.<\/span>len<\/span>()<\/span> as<\/span> Num<\/span>;<\/span>\r\n        let<\/span> (<\/span>row<\/span>,<\/span> col<\/span>)<\/span> =<\/span> (<\/span>theta<\/span>.<\/span>len<\/span>(),<\/span> theta<\/span>[<\/span>0<\/span>]<\/span>.<\/span>len<\/span>());<\/span>\r\n        let<\/span> tmp<\/span> =<\/span> alpha<\/span>\/<\/span>m<\/span>;<\/span>\r\n        let<\/span> a<\/span> :Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span> =<\/span> vec!<\/span>[<\/span>vec!<\/span>[<\/span>tmp<\/span>;<\/span> col<\/span>];<\/span> row<\/span>];<\/span> \r\n\r\n        let<\/span> ans<\/span> =<\/span> (<\/span>0<\/span>..<\/span>iterations<\/span>)<\/span>\r\n          .<\/span>fold<\/span>(<\/span> theta<\/span>,<\/span> |<\/span>theta<\/span>,<\/span> _iteration<\/span>|<\/span>{<\/span>\r\n           sub2d<\/span>(<\/span>&<\/span>theta<\/span>,<\/span> &<\/span>emult2d<\/span>(<\/span>&<\/span>mult<\/span>(<\/span> &<\/span>tx<\/span>,<\/span> &<\/span>sub2d<\/span>(<\/span> &<\/span>mult<\/span>(<\/span> &<\/span>x<\/span>,<\/span> &<\/span>theta<\/span> ),<\/span> &<\/span>y<\/span> )<\/span> ),<\/span> &<\/span>a<\/span>))<\/span>\r\n          });<\/span>\r\n\r\n        Ok<\/span>(<\/span>ans<\/span>.<\/span>encode<\/span>(<\/span>env<\/span>))<\/span>\r\n      })();<\/span>\r\n      match<\/span> result<\/span> {<\/span>\r\n          Err<\/span>(<\/span>_err<\/span>)<\/span> =><\/span> env<\/span>.<\/span>error_tuple<\/span>(<\/span>\"test failed\"<\/span>.<\/span>encode<\/span>(<\/span>env<\/span>)),<\/span>\r\n          Ok<\/span>(<\/span>term<\/span>)<\/span> =><\/span> term<\/span>\r\n      }<\/span>  \r\n    });<\/span>\r\n  });<\/span>\r\n  Ok<\/span>(<\/span>atoms<\/span>::<\/span>ok<\/span>()<\/span>.<\/span>to_term<\/span>(<\/span>env<\/span>))<\/span>\r\n}<\/span>\r\n<\/code><\/pre>\n
                          \u89e3\u8aac<\/h1>\n
                          Elixir\u3068Rust\u306e\u901a\u4fe1<\/h2>\n
                          Elixir\u304b\u3089Rust\u306b\u5f15\u6570\u3092\u6e21\u3059\u306e\u306f1\u624b\u9593\u304b\u304b\u308a\u307e\u3059\uff0eElixir\u304b\u3089\u6e21\u3055\u308c\u308b\u30c7\u30fc\u30bf\u306fErlangVM\u304c\u8aad\u3080\u305f\u3081\u306e\u30d0\u30a4\u30c8\u30b3\u30fc\u30c9\u306a\u306e\u3067\uff0c\u30b3\u30f3\u30d1\u30a4\u30eb\u6642\u306b\u578b\u63a8\u8ad6\u304c\u3067\u304d\u307e\u305b\u3093\uff0e\u3061\u3083\u3093\u3068\u6307\u5b9a\u3057\u3066\u3042\u3052\u307e\u3057\u3087\u3046\uff0e<\/p>\n
                          |>\u300cElixir + Rustler\u3067\u30d9\u30af\u30c8\u30eb\u6f14\u7b97\u3092\u9ad8\u901f\u5316\u3057\u3088\u3046\u301cRustler\u521d\u7d1a\u8005\u7de8 1 \u301c\u300d<\/p>\n
                          \u307e\u305fElixir\u3067\u5f15\u6570\u3092\u6e21\u3059\u969b\u306b\u306f\uff0c\u3042\u3089\u304b\u3058\u3081\u30ea\u30b9\u30c8\u306e\u4e2d\u8eab\u3092\u3059\u3079\u3066\u6d6e\u52d5\u5c0f\u6570\u70b9\u578b\u306b\u3057\u307e\u3057\u3087\u3046\uff0e<\/p>\n
                          \r\n# Rust\u5074\u306e\u95a2\u6570\u3068\u5bfe\u5fdc\u3065\u3051<\/span>\r\ndef<\/span> _fit<\/span>(<\/span>_x<\/span>,<\/span> _y<\/span>,<\/span> _theta<\/span>,<\/span> _alpha<\/span>,<\/span> _iteration<\/span>),<\/span> do<\/span>:<\/span> exit<\/span>(<\/span>:nif_not_loaded<\/span>)<\/span>\r\n\r\ndef<\/span> to_float<\/span>(<\/span>num<\/span>)<\/span> when<\/span> is_integer<\/span>(<\/span>num<\/span>),<\/span> do<\/span>:<\/span> num<\/span> \/<\/span>1<\/span>\r\ndef<\/span> to_float<\/span>(<\/span>r<\/span>)<\/span> when<\/span> is_list<\/span>(<\/span>r<\/span>)<\/span> do<\/span>\r\n    r<\/span>\r\n    |><\/span>Enum<\/span>.<\/span>map<\/span>(<\/span> &<\/span>to_float<\/span>(<\/span>&1<\/span>)<\/span> )<\/span>\r\nend<\/span>\r\n\r\ndef<\/span> fit<\/span>(<\/span> x<\/span>,<\/span> y<\/span>,<\/span> theta<\/span>,<\/span> alpha<\/span>,<\/span> iterations<\/span> )<\/span> do<\/span>\r\n   _fit<\/span>(<\/span>\r\n      x<\/span> |><\/span> to_float<\/span>,<\/span> \r\n      y<\/span> |><\/span> to_float<\/span>,<\/span> \r\n      theta<\/span> |><\/span> to_float<\/span>,<\/span> \r\n      alpha<\/span> |><\/span> to_float<\/span>,<\/span>\r\n      iterations<\/span>)<\/span>\r\n    receive<\/span> do<\/span>\r\n      l<\/span> -><\/span> l<\/span>\r\n    end<\/span>\r\n  end<\/span>\r\n<\/code><\/pre>\n
                          Rust\u306e\u65b9\u304b\u3089\u7d50\u679c\u3092\u53d7\u3051\u53d6\u308b\u306b\u306freceive\u3092\u4f7f\u3044\u307e\u3059\uff0e<\/p>\n
                          NIF\u306e\u5236\u7d04<\/h2>\n
                          NIF\u306fElixir\u3068\u306e\u901a\u4fe1\uff0c\u5185\u90e8\u3067\u306e\u30c7\u30fc\u30bf\u51e6\u7406\u30921ms\u4ee5\u4e0b\u306b\u6291\u3048\u306a\u3044\u3068\uff0c\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u304c\u5927\u5e45\u306b\u4f4e\u4e0b\u3057\u307e\u3059\uff0e
                          \n\u305d\u306e\u305f\u3081\uff0cRust\u5074\u3067\u30b9\u30ec\u30c3\u30c9\u3092\u7acb\u3066\u3066\uff0c\u305d\u3053\u304b\u3089\u8a08\u7b97\u7d50\u679c\u3092\u53d7\u4fe1\u3059\u308b\u69cb\u9020\u306b\u306a\u3063\u3066\u3044\u307e\u3059\uff0e<\/p>\n
                          \u53c2\u7167\u3084\u6240\u6709\u6a29\u306b\u3064\u3044\u3066<\/h2>\n
                          Rust\u3067\u306f\u30c7\u30fc\u30bf\u306b\u3064\u304d\uff0c1\u3064\u306e\u5909\u6570\u304c\u6240\u6709\u6a29\u3092\u3082\u3063\u3066\u3044\u307e\u3059\uff0e\u305d\u306e\u305f\u3081\uff0c\u57fa\u672c\u7684\u306b\u8aad\u307f\u53d6\u308a\u5c02\u7528\u306e\u53c2\u7167\u3067\u95a2\u6570\u306b\u6e21\u3059\u3068\u901f\u304f\u306a\u308a\u307e\u3059\u3057\uff0c\u305d\u306e\u5f8c\u306e\u30b3\u30fc\u30c9\u304c\u66f8\u304d\u3084\u3059\u3044\u3067\u3059\uff0e<\/p>\n
                          \u3082\u3057\u53c2\u7167\u3092\u6e21\u3055\u306a\u3044\u5834\u5408\u306f\uff0c\u95a2\u6570\u306b\u8a2d\u5b9a\u3057\u305f\u5f15\u6570\u306b\u5143\u306e\u5909\u6570\u306e\u6240\u6709\u6a29\u304c\u79fb\u3063\u3066\u3057\u307e\u3046\u305f\u3081\u306b\u305d\u306e\u95a2\u6570\u304c\u51e6\u7406\u3092\u7d42\u3048\u305f\u5f8c\uff0c\u518d\u5229\u7528\u3067\u304d\u306a\u304f\u306a\u3063\u3066\u3057\u307e\u3044\u307e\u3059\uff0e\u4ee5\u524d\u306f\u3053\u306e\u5bfe\u51e6\u306b\u4e8b\u524d\u306bclone()\u3057\u3066\u95a2\u6570\u306b\u6e21\u3059\u3068\u3044\u3046\u65b9\u6cd5\u3092\u3068\u3063\u3066\u3044\u307e\u3057\u305f\uff0e
                          \n\u4eca\u601d\u3048\u3070\uff0c\u30d7\u30ed\u30b0\u30e9\u30de\u30fc\u306b\u53c2\u7167\u6e21\u3057\u3092\u5f37\u5236\u3055\u305b\u3089\u308c\u308b\u4ed5\u69d8\u306b\u306a\u3063\u3066\u3044\u308b\u3088\u3046\u306b\u601d\u3048\u307e\u3059\u306d\uff0e<\/p>\n
                          \u4e26\u5217\u51e6\u7406\u90e8<\/h2>\n
                          \u4e0b\u8a18\u30b3\u30fc\u30c9\u304c\u30e1\u30a4\u30f3\u306e\u51e6\u7406\u90e8\u5206\u3067\u3059\uff0e<\/p>\n
                          let<\/span> ans<\/span> =<\/span> (<\/span>0<\/span>..<\/span>iterations<\/span>)<\/span>\r\n  .fold<\/span>(<\/span> theta<\/span>,<\/span> |<\/span>theta<\/span>,<\/span> _<\/span>iteration<\/span>|{<\/span>\r\n    sub2d<\/span>(<\/span>&<\/span>theta<\/span>,<\/span> &<\/span>emult2d<\/span>(<\/span>&<\/span>mult<\/span>(<\/span> &<\/span>tx<\/span>,<\/span> &<\/span>sub2d<\/span>(<\/span> &<\/span>mult<\/span>(<\/span> &<\/span>x<\/span>,<\/span> &<\/span>theta<\/span> ),<\/span> &<\/span>y<\/span> )<\/span> ),<\/span> &<\/span>a<\/span>))<\/span>\r\n  });<\/span>\r\n\r\n<\/code><\/pre>\n
                          \u30de\u30eb\u30c1\u30b3\u30a2\u3067\u4e26\u5217\u5b9f\u884c\u3059\u308b\u969b\u306b\u306fiter\u3092par_iter\u306b\u5909\u66f4\u3057\u307e\u3059\uff0e\u307e\u305f\uff0c\u4e00\u756a\u5916\u5074\u306e\u51e6\u7406\u3092\u4e26\u5217\u5316\u3059\u308b\u3068\u52b9\u679c\u304c\u9ad8\u3044\u3067\u3059\uff0e\u4eca\u56de\u306e\u5834\u5408\u3060\u3068\uff0c\u884c\u5217\u306e\u5404\u8981\u7d20\u3067\u6e1b\u7b97\u3092\u884c\u3046sub2d\u95a2\u6570\u306b\u5909\u66f4\u3092\u52a0\u3048\u307e\u3059\uff0e<\/p>\n
                          \u5909\u66f4\u524d<\/h3>\n
                          pub<\/span> fn<\/span> sub2d<\/span>(<\/span>x<\/span>:<\/span> &<\/span>Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span>,<\/span> y<\/span>:<\/span> &<\/span>Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span>)<\/span> -><\/span> Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span>{<\/span>\r\n  x<\/span>.iter<\/span>()<\/span>.zip<\/span>(<\/span>y<\/span>.iter<\/span>())<\/span>\r\n  .map<\/span>(|<\/span>t<\/span>|<\/span> sub<\/span>(<\/span>&<\/span>t<\/span>.0<\/span>.to_vec<\/span>(),<\/span> &<\/span>t<\/span>.1<\/span>.to_vec<\/span>()))<\/span>\r\n  .collect<\/span>()<\/span>\r\n}<\/span>\r\n<\/code><\/pre>\n
                          \u5909\u66f4\u5f8c<\/h3>\n
                          pub<\/span> fn<\/span> sub2d<\/span>(<\/span>x<\/span>:<\/span> &<\/span>Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span>,<\/span> y<\/span>:<\/span> &<\/span>Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span>)<\/span> -><\/span> Vec<\/span><<\/span>Vec<\/span><<\/span>Num<\/span>>><\/span>{<\/span>\r\n  x<\/span>.par_iter<\/span>()<\/span>.zip<\/span>(<\/span>y<\/span>.par_iter<\/span>())<\/span>\r\n  .map<\/span>(|<\/span>t<\/span>|<\/span> sub<\/span>(<\/span>&<\/span>t<\/span>.0<\/span>.to_vec<\/span>(),<\/span> &<\/span>t<\/span>.1<\/span>.to_vec<\/span>()))<\/span>\r\n  .collect<\/span>()<\/span>\r\n}<\/span>\r\n<\/code><\/pre>\n
                          Rayon\u306fiter->par_iter\u3068\u5909\u3048\u308b\u3060\u3051\u3067\u4e26\u5217\u51e6\u7406\u304c\u3067\u304d\u308b\u4fbf\u5229\u306a\u30e9\u30a4\u30d6\u30e9\u30ea\u3067\u3059\u304c\uff0czip()\u3068\u4f75\u7528\u3059\u308b\u3068\u3081\u3063\u3061\u3083\u9045\u3044\u307f\u305f\u3044\u306a\u3053\u3068\u3092\u3069\u3053\u304b\u3067\u898b\u305f\u899a\u3048\u304c\u3042\u308b\u306e\u3067\uff0c\u8fd1\u3044\u3046\u3061\u306b\u8abf\u67fb\u3057\u3088\u3046\u3068\u601d\u3063\u3066\u307e\u3059\uff0e<\/p>\n
                          \u4eca\u5f8c\u3078\u306e\u7e4b\u304e<\/h1>\n
                          \u672c\u7a3f\u3067\u306f\uff0cElixir\u306e\u30cd\u30a4\u30c6\u30a3\u30d6\u30b3\u30fc\u30c9\u5316\uff0cCPU\u30de\u30eb\u30c1\u30b3\u30a2\u3092\u751f\u304b\u3057\u305f\u4e26\u5217\u51e6\u7406\u3092\u30b3\u30fc\u30c9\u306b\u7d44\u307f\u8fbc\u3093\u3060\u5834\u5408\u306e\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u3092\u6e2c\u5b9a\u3057\u307e\u3057\u305f\uff0e<\/p>\n
                          \u4e0a\u8a18\u306e\u5909\u66f4\u3060\u3051\u3067\u306f\uff0c\u30d1\u30d5\u30a9\u30fc\u30de\u30f3\u30b9\u306f\u5287\u7684\u306b\u306f\u4f38\u3073\u307e\u305b\u3093\u3067\u3057\u305f\u306d\uff0e\u305d\u306e\u8981\u56e0\u306e1\u3064\u306b\u51e6\u7406\u81ea\u4f53\u304c\u305d\u3093\u306a\u306b\u91cd\u304f\u306a\u3044\u3068\u3044\u3046\u306e\u304c\u3042\u308a\u307e\u3059\u304c\uff0c\u624b\u52d5\u3067\u6700\u9069\u5316\u30fb\u9ad8\u901f\u5316\u3057\u3088\u3046\u3068\u3057\u3066\u307f\u308b\u3068\uff0c\u3068\u3066\u3082\u624b\u9593\u304c\u304b\u304b\u308b\u4e0a\u306b\u7d50\u679c\u304c\u3088\u308d\u3057\u304f\u306a\u3044\u3068\u3044\u3046\u3042\u307e\u308a\u304a\u3082\u3093\u306a\u3044\u7d50\u679c\u3067\u3057\u305f\uff0e<\/p>\n
                          \u305d\u306e\u305f\u3081\uff0cHastega\u3092\u6709\u52b9\u306b\u4f7f\u3046\u305f\u3081\u306b\u306f\u666e\u901a\u306b\u52d5\u304b\u3059\u5834\u5408\uff0cCPU\u30de\u30eb\u30c1\u30b3\u30a2\uff0cGPU\u30de\u30eb\u30c1\u30b3\u30a2\u306e\u3069\u308c\u304c\u901f\u304f\u306a\u308b\u306e\u304b\u5224\u65ad\u3059\u308b\u6a5f\u80fd\u304c\u5fc5\u8981\u306b\u306a\u308a\u307e\u3059\u306d\uff0e\u4e26\u5217\u51e6\u7406\u3055\u305b\u308b\u65b9\u6cd5\u3068\u3057\u3066\u306f\uff0cElixir\u306e\u30d7\u30ed\u30bb\u30b9\u3082\u3042\u3052\u3089\u308c\u308b\u306e\u3067\u30d7\u30ed\u30bb\u30b9\u9593\u306e\u901a\u4fe1\u65b9\u6cd5\u3092\u6539\u826f\u3059\u308b(Flow\u306e\u6539\u826f?)\u3068\u304b\u3082\u3059\u308b\u304b\u3082\u3057\u308c\u307e\u305b\u3093\uff0e<\/p>\n
                          \u7814\u7a76\u8a08\u753b<\/h2>\n
                          \u4eca\u5f8c\u306e\u53d6\u308a\u7d44\u307f\u3068\u3057\u3066\uff0c<\/p>\n
                            \n
                          1. \n
                              \u5927\u304d\u3081\u306e\u7dda\u5f62\u7684\u306a\u64ec\u4f3c\u30e2\u30c7\u30eb\u3092\u751f\u6210\u3059\u308b\u6a5f\u80fd\u3092\u4f5c\u308b(\u73fe\u5b9f\u7684\u306a\u30e2\u30c7\u30eb\u3068\u3057\u3066\u5b58\u5728\u3059\u308b\u898f\u6a21\u304b\u306f\u4e00\u65e6\u7f6e\u3044\u3068\u304f)<\/ol>\n<\/li>\n<\/ol>\n
                               <\/p>\n
                                \n
                              1. \n
                                  SIMD & CPU\u30de\u30eb\u30c1\u30b3\u30a2\u306b\u304a\u3051\u308b\u6027\u80fd\u8a55\u4fa1\u3092\u518d\u5ea6\u884c\u3046\uff0e<\/ol>\n<\/li>\n<\/ol>\n
                                   <\/p>\n
                                    Rust\u306eOpenCL\u30e9\u30c3\u30d1\u30fc\u3067\u3042\u308bCrate ocl\u3092\u4f7f\u3063\u305fGPGPU\u306e\u30b3\u30fc\u30c9\u3092\u8a55\u4fa1\u3092\u884c\u3044\u307e\u3059\uff0e<\/ol>\n
                                    \u51fa\u6765\u4e0a\u304c\u308a\u6b21\u7b2c\u8a18\u4e8b\u306b\u3057\u307e\u3059\uff0e<\/p>\n
                                    \u6700\u5f8c\u306b<\/h1>\n
                                    \u7814\u7a76\u306e\u5168\u4f53\u69cb\u60f3\u306b\u3064\u3044\u3066\u77e5\u308a\u305f\u3044\u65b9\u306f\u3053\u3061\u3089\u3092\u5408\u308f\u305b\u3066\u304a\u8aad\u307f\u304f\u3060\u3055\u3044\uff0e<\/p>\n
                                    \u300cZEAM\u958b\u767a\u30ed\u30b02018\u5e74\u7dcf\u96c6\u7de8\u305d\u306e1: Elixir \u7814\u7a76\u69cb\u60f3\u306b\u3064\u3044\u3066\u3075\u308a\u304b\u3048\u308b(\u524d\u7de8)\u300d<\/p>\n
                                    2018\u5e7412\u670815\u65e5\u306b\u516c\u958b\u4e88\u5b9a:
                                    \n\u300cfukuoka.ex Elixir\uff0fPhoenix Advent Calendar 2018\u300d15\u65e5\u76ee
                                    \n\u300cZEAM\u958b\u767a\u30ed\u30b02018\u5e74\u7dcf\u96c6\u7de8\u305d\u306e2: Elixir \u7814\u7a76\u69cb\u60f3\u306b\u3064\u3044\u3066\u3075\u308a\u304b\u3048\u308b(\u5f8c\u7de8)\u300d<\/p>\n
                                    \u8ffd\u8a18<\/h1>\n
                                    \u30ea\u30dd\u30b8\u30c8\u30ea\u3092\u516c\u958b\u3057\u307e\u3057\u305f
                                    \nhttps:\/\/github.com\/zeam-vm\/linear_regressor_cli<\/p>\n
                                    \u4eca\u9031\u4e2d(~2018\/12\/16\u307e\u3067)\u306b\u30b3\u30e1\u30f3\u30c8\u3068\u3054\u6307\u6458\u3044\u305f\u3060\u3044\u305f\u5185\u5bb9\u3068\u4e0a\u8a18\u3067\u793a\u3057\u305f\u7814\u7a76\u8a08\u753b\u3092\u5b9f\u65bd\uff0c\u305d\u308c\u3068\u30c9\u30ad\u30e5\u30e1\u30f3\u30c8\u3092\u6574\u5099\u3059\u308b\u4e88\u5b9a\u3067\u3059\uff0e<\/p>\n
                                    \n
                                    \u660e\u65e5\u306e\u300cfukuoka.ex Elixir\uff0fPhoenix Advent Calendar 2018\u300d10\u65e5\u76ee\u306e\u8a18\u4e8b\u306f\uff0c @curry_on_a_rice \u3055\u3093\u306e\u300cElixir\u3067\u30cb\u30e5\u30fc\u30e9\u30eb\u30cd\u30c3\u30c8\u3092\u5b9f\u88c5\u3057\u3088\u3046\u3068\u3057\u305f\u8a71\u300d\u300d\u3067\u3059\u3002\u3053\u3061\u3089\u3082\u304a\u697d\u3057\u307f\u306b\uff01<\/p>\n","protected":false},"excerpt":{"rendered":"
                                    (\u3053\u306e\u8a18\u4e8b\u306f\u300cfukuoka.ex Elixir\uff0fPhoenix Advent Calendar 2018\u300d\u306e […]<\/p>\n","protected":false},"author":11,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-45559","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"yoast_head":"\n- Blog - Silicon Cloud<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:description\" content=\"(\u3053\u306e\u8a18\u4e8b\u306f\u300cfukuoka.ex Elixir\uff0fPhoenix Advent Calendar 2018\u300d\u306e […]\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/\" \/>\n<meta property=\"og:site_name\" content=\"Blog - Silicon Cloud\" \/>\n<meta property=\"article:published_time\" content=\"2023-03-06T05:30:33+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-04-30T09:32:40+00:00\" \/>\n<meta name=\"author\" content=\"\u65b0, \u97f5\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"\u65b0, \u97f5\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"4 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/\",\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/\",\"name\":\"- Blog - Silicon Cloud\",\"isPartOf\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#website\"},\"datePublished\":\"2023-03-06T05:30:33+00:00\",\"dateModified\":\"2024-04-30T09:32:40+00:00\",\"author\":{\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/4ba4019495123db3038fd0809e6959c9\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/\"]}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#website\",\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/\",\"name\":\"Blog - Silicon Cloud\",\"description\":\"\",\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/4ba4019495123db3038fd0809e6959c9\",\"name\":\"\u65b0, \u97f5\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/secure.gravatar.com\/avatar\/d484b6c6e4ae82e8a9efea989e1d2af46d9b6ef128101e63b18f559fca0ae627?s=96&d=mm&r=g\",\"contentUrl\":\"https:\/\/secure.gravatar.com\/avatar\/d484b6c6e4ae82e8a9efea989e1d2af46d9b6ef128101e63b18f559fca0ae627?s=96&d=mm&r=g\",\"caption\":\"\u65b0, \u97f5\"},\"url\":\"https:\/\/www.silicloud.com\/zh\/blog\/author\/yunxin\/\"},{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/#local-main-organization-logo\",\"url\":\"\",\"contentUrl\":\"\",\"caption\":\"Blog - Silicon Cloud\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"- Blog - Silicon Cloud","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/","og_locale":"zh_CN","og_type":"article","og_description":"(\u3053\u306e\u8a18\u4e8b\u306f\u300cfukuoka.ex Elixir\uff0fPhoenix Advent Calendar 2018\u300d\u306e […]","og_url":"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/","og_site_name":"Blog - Silicon Cloud","article_published_time":"2023-03-06T05:30:33+00:00","article_modified_time":"2024-04-30T09:32:40+00:00","author":"\u65b0, \u97f5","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"\u65b0, \u97f5","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"4 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/","url":"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/","name":"- Blog - Silicon Cloud","isPartOf":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/#website"},"datePublished":"2023-03-06T05:30:33+00:00","dateModified":"2024-04-30T09:32:40+00:00","author":{"@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/4ba4019495123db3038fd0809e6959c9"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/"]}]},{"@type":"WebSite","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#website","url":"https:\/\/www.silicloud.com\/zh\/blog\/","name":"Blog - Silicon Cloud","description":"","inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/4ba4019495123db3038fd0809e6959c9","name":"\u65b0, \u97f5","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.silicloud.com\/zh\/blog\/#\/schema\/person\/image\/","url":"https:\/\/secure.gravatar.com\/avatar\/d484b6c6e4ae82e8a9efea989e1d2af46d9b6ef128101e63b18f559fca0ae627?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/d484b6c6e4ae82e8a9efea989e1d2af46d9b6ef128101e63b18f559fca0ae627?s=96&d=mm&r=g","caption":"\u65b0, \u97f5"},"url":"https:\/\/www.silicloud.com\/zh\/blog\/author\/yunxin\/"},{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.silicloud.com\/zh\/blog\/45559-2\/#local-main-organization-logo","url":"","contentUrl":"","caption":"Blog - Silicon Cloud"}]}},"_links":{"self":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/45559","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/users\/11"}],"replies":[{"embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/comments?post=45559"}],"version-history":[{"count":2,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/45559\/revisions"}],"predecessor-version":[{"id":93299,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/posts\/45559\/revisions\/93299"}],"wp:attachment":[{"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/media?parent=45559"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/categories?post=45559"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.silicloud.com\/zh\/blog\/wp-json\/wp\/v2\/tags?post=45559"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}