{"id":10235,"date":"2018-08-14T01:00:00","date_gmt":"2018-08-13T17:00:00","guid":{"rendered":"http:\/\/swarma.org\/?p=10235"},"modified":"2018-12-05T17:04:10","modified_gmt":"2018-12-05T09:04:10","slug":"%e6%94%be%e5%bc%83julia%ef%bc%9f%e9%95%bf%e6%96%87%e8%a7%a3%e6%9e%90%e5%af%b9julia%e7%9a%84%e8%b4%a8%e7%96%91%e9%94%99%e5%9c%a8%e5%93%aa%e9%87%8c","status":"publish","type":"post","link":"https:\/\/swarma.org\/?p=10235","title":{"rendered":"\u653e\u5f03Julia\uff1f\u957f\u6587\u89e3\u6790\u5bf9Julia\u7684\u8d28\u7591\u9519\u5728\u54ea\u91cc"},"content":{"rendered":"
\n

\"\u96c6\u667a\"<\/p>\n

\n

\u524d\u8a00<\/span><\/strong><\/span><\/p>\n

\n
\n
\n

Julia \u8bed\u8a001.0\u7248\u672c\u8fd1\u671f\u521a\u521a\u6b63\u5f0f\u4e0a\u7ebf\uff0c\u4f5c\u4e3a\u79d1\u5b66\u548c\u6570\u503c\u8ba1\u7b97\u7684\u795e\u5668\uff0cJulia\u5f15\u8d77\u4e86\u4e1a\u5185\u5e7f\u6cdb\u5173\u6ce8\u3002Julia \u8bed\u8a00\u4ee5\u901f\u5ea6\u8457\u79f0\uff0c\u4f46\u57281.0\u7248\u672c\u4e0a\u7ebf\u4e4b\u524d\u5176\u6027\u80fd\u4e5f\u66fe\u53d7\u5230\u8bf8\u591a\u8d28\u7591\uff0c\u672c\u6587\u662f\u5f00\u53d1\u8005Tom Kwong\u5bf9\u8d28\u7591\u58f0\u97f3\u7684\u56de\u5e94\u2014\u2014\u4e3a\u4ec0\u4e48\u201c\u653e\u5f03Julia\u201d\u7684\u6279\u8bc4\u662f\u8f7b\u7387\u7684\u3002\u672c\u6587\u4f7f\u7528Julia0.6\u7248\u672c\u3002<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n

\n<\/section>\n

\n
\n

\u7f16\u8bd1\uff1a\u96c6\u667a\u7ffb\u8bd1\u7ec4
\n<\/span><\/p>\n

\u6765\u6e90\uff1a<\/span><\/p>\n

tk3369.wordpress.com<\/span><\/p>\n

\u539f\u9898\uff1a<\/span><\/p>\n

An Updated Analysis for the \u201cGiving Up on Julia\u201d Blog Post<\/span><\/p>\n<\/blockquote>\n<\/section>\n

\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\u80cc\u666f<\/span><\/strong><\/span><\/strong><\/section>\n<\/section>\n

\u00a0<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u00a0<\/strong><\/section>\n<\/section>\n<\/section>\n

2016\u5e743\u6708\u7684\u4e00\u7bc7\u535a\u5ba2\uff0c<\/span>\u5927\u8086\u5410\u69fd\u4e86\u4e00\u4e0bJulia\u8bed\u8a00\u3002\u4ece\u8fd9\u7bc7\u535a\u5ba2\u7684\u8bc4\u8bba\u548c\u6700\u8fd1\u793e\u533a\u4e0a\u5bf9Julia\u8bed\u8a00\u7684\u8ba8\u8bba\u4e0a\u6765\u770b\uff0c \u90a3\u7bc7\u535a\u5ba2\u786e\u5b9e\u662f\u6709\u70b9\u4e89\u8bae\u7684\u3002<\/span><\/p>\n

\u00a0<\/span><\/p>\n

\n

\u535a\u5ba2\u5730\u5740\uff1a<\/span><\/p>\n

http:\/\/zverovich.net\/2016\/05\/13\/giving-up-on-julia.html\u00a0<\/span><\/p>\n<\/blockquote>\n

\n

\u90a3\u4e48\u8fd9\u7bc7\u535a\u5ba2\u7684\u76ee\u7684\u5c31\u662f\u6765\u53cd\u9a73\u4e00\u4e0b\u90a3\u4e9b\u9519\u8bef\u7684\u89c2\u70b9\uff0c\u7136\u540e\u786e\u5b9a\u4e00\u4e0b\u4ec0\u4e48\u624d\u662f\u6211\u4eec\u9700\u8981\u771f\u6b63\u53bb\u5173\u6ce8\u7684\u3002\u6211\u4f1a\u5c3d\u91cf\u516c\u5e73\u5730\u5217\u4e3e\u4e0d\u540c\u4e4b\u5904\uff0c\u8fd9\u6837\u4f60\u4eec\u5c31\u53ef\u4ee5\u6839\u636e\u4f8b\u5b50\u6765\u5177\u4f53\u5224\u65ad\u3002<\/span><\/p>\n

\n

\u4e0b\u9762\u6240\u6709\u7684\u8868\u73b0\u90fd\u57fa\u4e8e\u6211\u7684Macbook Pro\u7b14\u8bb0\u672c(Intel i5-4258U CPU@2.40 GHz)\u4ee5\u53caJulia 0.6.2\u548cAnaconda\u53d1\u884c\u7248\u7684Python 3.6.3\u3002<\/span><\/p>\n

\n

\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\u5173\u4e8eJulia\u5e95\u5c42\u6027\u80fd\u6d4b\u8bd5\u7684\u62b1\u6028<\/span><\/strong><\/span><\/strong><\/section>\n<\/section>\n

\u00a0<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u4f5c\u8005\u5f15\u7528\u4e86\u51e0\u4e2a\u5173\u4e8e\u6027\u80fd\u6d4b\u8bd5\u7684github issue\u3002
\n<\/span><\/p>\n

\u00a0<\/span><\/p>\n

1.parse_integer benchmark(https:\/\/github.com\/JuliaLang\/julia\/issues\/4662)\u7684C\u8bed\u8a00\u5b9e\u73b0\u4e0d\u662f\u7528\u7528\u60ef\u7528\u6807\u51c6\u7f16\u7801\u7684\uff0c\u6bd4\u5982strlen\u51fd\u6570\u30022013\u5e7410\u6708\u65f6\u6838\u5fc3\u5f00\u53d1\u56e2\u961f\u627f\u8ba4\u4e86\u8fd9\u4e2a\u95ee\u9898\u5e76\u5f88\u5feb\u7684\u5bf9\u5176\u8fdb\u884c\u4e86\u4fee\u590d\uff08\u4e24\u5929\u5185\uff09\u3002<\/span><\/p>\n

\u00a0<\/span><\/p>\n

2.\u4e5f\u6709\u4eba\u63d0\u95ee\u4e86Java\u548cOctave\u5b9e\u73b0\u7684\u4e0d\u540c\u6027\u80fd\u6d4b\u8bd5\u7684\u95ee\u9898\u3002\u5728Github\u4e0a\u7ecf\u8fc7\u4e86\u51e0\u8f6e\u8ba8\u8bba\u4e4b\u540e\uff0c\u6700\u7ec8\u5f52\u7ed3\u4e8e\u5927\u5bb6\u5bf9\u4e8e\u6027\u80fd\u6d4b\u8bd5\u7684\u4e0d\u540c\u7406\u5ff5\u3002\u6211\u4eec\u5230\u5e95\u662f\u60f3\u8981\u901a\u8fc7\u6027\u80fd\u6d4b\u8bd5\u6765\u7406\u89e3\u5bf9\u4e8e\u76f8\u540c\u7b97\u6cd5\u7684\u4e0d\u540c\u8bed\u8a00\u95f4\u5b9e\u73b0\u7684\u6027\u80fd\u8fd8\u662f\u60f3\u77e5\u9053\u5728\u8fd9\u4e9b\u8bed\u8a00\u4e4b\u95f4\u7528\u6700\u4f18\u5316\u7684\u4ee3\u7801\u6765\u8fdb\u884c\u6bd4\u8f83\u5462\uff1fJulia\u7684\u6027\u80fd\u6d4b\u8bd5\u89e3\u91ca\u5230\uff1a<\/span><\/p>\n

\u00a0<\/span><\/p>\n

\n

\u9700\u8981\u4e3b\u8981\u5230\u7684\u662f\uff0c\u6027\u80fd\u6d4b\u8bd5\u4ee3\u7801\u5e76\u4e0d\u662f\u7528\u7edd\u5bf9\u6700\u5927\u6027\u80fd\u5199\u7684\uff08\u6700\u5feb\u9012\u5f52\u8ba1\u7b97\u6590\u6ce2\u90a3\u5951\u6570\u5217\u7b2c\u4e8c\u5341\u4e2a\u5143\u7d20\u7684\u4ee3\u7801\u662f6765\uff09\u3002\u76f8\u53cd\uff0c\u7f16\u5199\u8fd9\u4e9b\u6027\u80fd\u6d4b\u8bd5\u662f\u4e3a\u4e86\u6d4b\u8bd5\u6bcf\u79cd\u8bed\u8a00\u4e2d\u5b9e\u73b0\u76f8\u540c\u7b97\u6cd5\u548c\u4ee3\u7801\u6a21\u5f0f\u4e2d\u7684\u6027\u80fd\u3002<\/span><\/p>\n<\/blockquote>\n

\u00a0<\/span><\/p>\n

\u5173\u4e8e\u4f18\u5316<\/strong><\/span><\/p>\n

\u00a0<\/strong><\/span><\/p>\n

\u5728\u4efb\u610f\u53ef\u62d3\u5c55\u7684\u8bed\u8a00\u4e2d\u4eba\u4eec\u90fd\u53ef\u4ee5\u4f18\u5316\u4efb\u4f55\u4e1c\u897f\u5e76\u53d6\u5f97\u826f\u597d\u7684\u6027\u80fd\u3002\u5728Python\u7684\u4f8b\u5b50\u4e2d\uff0c\u6211\u4eec\u53ef\u4ee5\u5e94\u7528Cython\uff0cNumba\uff0cNumpy\u548c\u5176\u4ed6\u7684\u6280\u5de7\u3002\u8fd9\u662fStephan Karpinski\u6240\u63cf\u8ff0\u7684\u5178\u578b\u7684\u53cc\u8bed\u8a00\u95ee\u9898\u3002<\/span><\/p>\n

\u00a0<\/span><\/p>\n

\u5373\uff1a<\/span><\/p>\n

1.\u8981\u4f18\u5316\u4e00\u4e2a\u4e8b\u7269\u6709\u591a\u56f0\u96be\uff0c\u8981\u4ed8\u51fa\u591a\u5927\u7684\u52aa\u529b\u3002<\/span><\/p>\n

2.\u4f18\u5316\u662f\u5426\u4f1a\u5728\u4ee3\u7801\u4e2d\u5f15\u5165\u989d\u5916\u7684\u590d\u6742\u6027\u5e76\u963b\u788d\u751f\u4ea7\u529b\u3002<\/span><\/p>\n

\n

\n
\n
\n
\n
\n
\n
\n
\n

\u00a0Hello World\u7684\u6027\u80fd<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u4f5c\u8005\u5bf9\u4e8e\u4e00\u4e2a\u7b80\u5355\u7684Hello World\u7a0b\u5e8f\u7684\u8fd0\u884c\u65f6\u95f4\u6709\u70b9\u4e0d\u6ee1\u3002\u5305\u542b\u542f\u52a8\u65f6\u95f4\u5f88\u660e\u663e\u662f\u5f88\u5947\u8469\u4e14\u7247\u9762\u7684\u3002
\n<\/span><\/p>\n

\u00a0<\/span><\/p>\n

\u95ee\u9898\u5c31\u662f\uff0c\u8fd9\u6709\u5173\u7cfb\u5417\uff1f\u4e5f\u8bb8\u6709\uff0c\u4e5f\u8bb8\u6ca1\u6709\u3002<\/span><\/p>\n

\n

1.\u5982\u679c\u4f60\u6709\u4e00\u5806\u53ea\u9700\u8981\u4e00\u79d2\u949f\u5c31\u80fd\u8fd0\u884c\u7684\u77ed\u65f6\u95f4\u5e94\u7528\u7a0b\u5e8f\uff0c\u90a3\u4e48\u8f83\u957f\u7684\u542f\u52a8\u65f6\u95f4\u53ef\u80fd\u5bf9\u4f60\u4ea7\u751f\u5f88\u5927\u7684\u5f71\u54cd\u3002<\/span><\/span><\/p>\n

\u00a0<\/span><\/span><\/p>\n

2.\u5982\u679c\u4f60\u6709\u4e00\u4e2a\u5927\u6982\u8981\u8017\u65f630\u5206\u949f\u7684\u5927\u578b\u8ba1\u7b97\u9879\u76ee\uff0c\u90a3\u4e480.5\u79d2\u7684\u542f\u52a8\u65f6\u95f4\u5c31\u6ca1\u4ec0\u4e48\u610f\u4e49\u4e86\u3002\u5bf9\u4e8e\u5927\u591a\u6570\u5546\u4e1a\u5e94\u7528\u7a0b\u5e8f\u6765\u8bf4\uff0c\u8fd9\u624d\u662f\u6bd4\u8f83\u5e38\u89c1\u7684\u60c5\u51b5\uff0c\u66f4\u522b\u63d0\u6570\u503c\u8fd0\u7b97\u548c\u79d1\u5b66\u8ba1\u7b97\u793e\u533a\u4e86\u3002<\/span><\/p>\n

\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\n
\u535a\u5ba2<\/strong><\/span><\/strong><\/span><\/strong><\/strong><\/p>\n

\u300a\u5982\u4f55\u8ba9Python\u8ddfJulia\u4e00\u6837\u5feb\u300b<\/strong><\/span><\/p>\n<\/section>\n<\/section>\n

\u00a0<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\u4f5c\u8005\u6307\u51fa\u5728\u535a\u5ba2”\u5982\u4f55\u8ba9Python\u8ddfJulia\u4e00\u6837\u5feb”\u4e2d\u6709\u5f88\u591a\u65b9\u5f0f\u53ef\u4ee5Python\u66f4\u5feb\u3002\u8fd9\u4e2a\u5f15\u7528\u611f\u89c9\u5c31\u50cf\u662f\u5728\u8bf4\u201c\u563f\uff0c\u6211\u4eecPython\u5f00\u53d1\u8005\u5df2\u7ecf\u77e5\u9053\u5982\u4f55\u8ba9\u7a0b\u5e8f\u8dd1\u7684\u66f4\u5feb\u4e86\uff0c\u800c\u4e14\u6ca1\u7528\u7406\u7531\u5207\u6362\u5230\u5176\u4ed6\u8bed\u8a00\u3002\u201d<\/span><\/p>\n

\n

\n
\n
\n
\n
\n

\u00a0\u6590\u6ce2\u90a3\u5951\u6027\u80fd\u6d4b\u8bd5<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n

\n

\u6211\u7684\u6d4b\u8bd5\u7ed3\u679c\u663e\u793a\uff0cCython-Typed\u7684\u901f\u5ea6\u548cJulia\u76f8\u5f53\u3002<\/span><\/p>\n

\u00a0<\/span><\/p>\n

\u53ea\u662f\u6709\u51e0\u4e2a\u6ce8\u610f\u4e8b\u9879<\/span><\/p>\n

    \n
  • \n

    \u535a\u5ba2\u7684\u6027\u80fd\u6d4b\u8bd5\u7ed3\u679c\u662fJulia 80us\u548cCython-Typed 24us\u3002\u6211\u4e0d\u80fd\u590d\u73b0\u8fd9\u6837\u7684\u8868\u73b0\u3002\u6211\u53ea\u80fd\u89e3\u91ca\u8bf4\uff0c\u65f6\u4ee3\u53d8\u4e86\uff0cJulia\u7684\u6027\u80fd\u4e5f\u63d0\u9ad8\u4e86\u3002<\/span><\/p>\n<\/li>\n

  • \n

    \u6590\u6ce2\u90a3\u5951\u51fd\u6570\u4f7f\u7528\u4e86LRU\u7f13\u5b58\u505a\u4e86\u4f18\u5316\u3002\u4ece\u6027\u80fd\u6d4b\u8bd5\u89d2\u5ea6\u6765\u770b\uff0c\u8fd9\u4e0d\u662f\u4e00\u4e2a\u516c\u5e73\u7684\u6bd4\u8f83\uff0c\u56e0\u4e3a\u7b97\u6cd5\u672c\u8eab\u6539\u53d8\u4e86\u3002\u6240\u4ee5\u8bf4\u6211\u4eec\u53ef\u4ee5\u5ffd\u7565\u8fd9\u4e9b\u7ed3\u679c\u3002<\/span><\/p>\n<\/li>\n

  • \n

    NUmba\u4f7f\u7528LRU-enhanced\u7684\u8fdb\u4e00\u6b65\u4f18\u5316\u4e86\u4ee3\u7801\u3002\u51fa\u4e8e\u540c\u6837\u7684\u539f\u56e0\u8003\u8651\uff0c\u6211\u4eec\u4e5f\u5c06\u5ffd\u7565\u8fd9\u4e9b\u7ed3\u679c\u3002<\/span><\/p>\n

    \u00a0<\/span><\/li>\n<\/ul>\n

    Python<\/strong><\/p>\n

    In [43]: def fib(n):<\/span><\/p>\n

    …: \u00a0 \u00a0 if n<2:<\/span><\/p>\n

    …: \u00a0 \u00a0 \u00a0 \u00a0 return n<\/span><\/p>\n

    …: \u00a0 \u00a0 return fib(n-1)+fib(n-2)<\/span><\/p>\n

    In [44]: %timeit fib(20)<\/span><\/p>\n

    <\/pre>\n
    3.27 ms \u00b1 48.3 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 100 loops each)<\/span><\/p>\n
    \u00a0<\/strong><\/p>\n
    Cython<\/strong><\/p>\n
    In [48]: %%cython<\/span><\/p>\n
        ...: def fib_cython(n):<\/pre>\n
        ...:     if n<2:<\/pre>\n
        ...:         return n<\/pre>\n
        ...:     return fib_cython(n-1)+fib_cython(n-2)<\/pre>\n
    \nIn [51]: %timeit fib_cython(20)<\/pre>\n
     1.48 ms \u00b1 329 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 1000 loops each)<\/pre>\n
    \u00a0<\/strong><\/p>\n
    Cython Typed<\/strong><\/p>\n
    In [71]: %%cython<\/span><\/pre>\n
     \u00a0 \u00a0...: cpdef long fib_cython_type(long n):<\/span><\/pre>\n
     \u00a0 \u00a0...: \u00a0 \u00a0 if n<2:<\/span><\/pre>\n
     \u00a0 \u00a0...: \u00a0 \u00a0 \u00a0 \u00a0 return n<\/span><\/pre>\n
     \u00a0 \u00a0...: \u00a0 \u00a0 return fib_cython_type(n-1)+fib_cython_type(n-2)<\/span><\/pre>\n
     In [72]: %timeit fib_cython_type(20)<\/span><\/pre>\n
     47.8 \u00b5s \u00b1 182 ns per loop (mean \u00b1 std. dev. of 7 runs, 10000 loops each<\/span><\/pre>\n
    \u00a0<\/strong><\/p>\n
    Julia<\/strong><\/p>\n
    julia> fib(n) = n < 2 ? n : fib(n-1) + fib(n-2) \u00a0<\/span><\/pre>\n
    fib (generic function with 1 method) \u00a0 \u00a0\n\n<\/span><\/pre>\n
    julia> @btime fib(20)<\/span><\/pre>\n
     48.075 \u03bcs (0 allocations: 0 bytes)<\/span><\/pre>\n
    \n
    \n
    \n
    \n
    \n
    \n
    \u00a0\u5feb\u901f\u6392\u5e8f<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \n
    \u4f5c\u8005\u5c06\u5feb\u901f\u6392\u5e8f\u51fd\u6570\u8f6c\u5316\u4e3aCython\uff0c\u6700\u540e\u5f97\u51fa\u7ed3\u8bbaNumpy\u7684\u6392\u5e8f\u65f6\u95f4\u662f\u6700\u5feb\u7684\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    1.\u6027\u80fd\u6d4b\u8bd5\u7b97\u6cd5 – Julia(356us) vs Cython (1030) \u3002<\/span><\/p>\n
    2.\u57fa\u6392\u51fd\u6570 – Julia(233us) vs Numpy(292us)\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    Python<\/strong><\/p>\n
    In [83]: def qsort1(a, lo, hi):<\/span><\/pre>\n
        ...: \u00a0 \u00a0 i = lo<\/span><\/pre>\n
        ...: \u00a0 \u00a0 j = hi<\/span><\/pre>\n
        ...: \u00a0 \u00a0 while i < hi:<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 pivot = a[(lo+hi) \/\/ 2]<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 while i <= j:<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 while a[i] < pivot:<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 i += 1<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 while a[j] > pivot:<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 j -= 1<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 if i <= j:<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 a[i], a[j] = a[j], a[i]<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 i += 1<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 j -= 1<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 if lo < j:<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 qsort1(a, lo, j)<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 lo = i<\/span><\/pre>\n
        ...: \u00a0 \u00a0 \u00a0 \u00a0 j = hi<\/span><\/pre>\n
        ...: \u00a0 \u00a0 return a<\/span><\/pre>\n
     In [84]: %timeit qsort1(lst, 0, len(lst)-1)<\/span><\/pre>\n
     13.7 ms \u00b1 140 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 100 loops each)<\/span><\/pre>\n
    \n
    Cython<\/strong><\/p>\n
      \u00a0 In [100]: %%cython<\/span><\/pre>\n
             ...: cdef double[:] qsort2(double[:] a, long lo, long hi):<\/span><\/pre>\n
             ...: \u00a0 \u00a0 cdef: <\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 long i, j<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 double pivot<\/span><\/pre>\n
             ...: \u00a0 \u00a0 i = lo<\/span><\/pre>\n
             ...: \u00a0 \u00a0 j = hi<\/span><\/pre>\n
             ...: \u00a0 \u00a0 while i < hi:<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 pivot = a[(lo+hi) \/\/ 2]<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 while i <= j:<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 while a[i] < pivot:<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 i += 1<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 while a[j] > pivot:<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 j -= 1<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 if i <= j:<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 a[i], a[j] = a[j], a[i]<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 i += 1<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 j -= 1<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 if lo < j:<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 qsort2(a, lo, j)<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 lo = i<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 j = hi<\/span><\/pre>\n
             ...: \u00a0 \u00a0 return a<\/span><\/pre>\n
             ...: <\/span><\/pre>\n
             ...: def qsort2_py(a, b, c):<\/span><\/pre>\n
             ...: \u00a0 \u00a0 return qsort2(a, b, c)\n \u00a0 \u00a0 \u00a0 \u00a0 \n<\/span><\/pre>\n
     In [105]: %timeit qsort2_py(np.random.rand(5000), 0, 4999)<\/span><\/pre>\n
     1.03 ms \u00b1 110 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 1000 loops each)<\/span><\/pre>\n
    \u00a0<\/strong><\/p>\n
    Numpy<\/strong><\/p>\n
    In [61]: %timeit np.sort(lst)<\/span><\/pre>\n
     292 \u00b5s \u00b1 4.9 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 1000 loops each)<\/span><\/pre>\n
    \u00a0<\/strong><\/p>\n
    Julia Microbenchmark<\/strong><\/p>\n
        julia> function qsort!(a,lo,hi)<\/span><\/pre>\n
                   i, j = lo, hi<\/span><\/pre>\n
                   while i < hi<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 pivot = a[(lo+hi)>>>1]<\/span><\/pre>\n
                       while i <= j<\/span><\/pre>\n
                           while a[i] < pivot; i += 1; end \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0while a[j] > pivot; j -= 1; end<\/span><\/pre>\n
                           if i <= j<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 a[i], a[j] = a[j], a[i]<\/span><\/pre>\n
                               i, j = i+1, j-1<\/span><\/pre>\n
                           end<\/span><\/pre>\n
                       end<\/span><\/pre>\n
                       if lo < j; qsort!(a,lo,j); end<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 lo, j = i, hi<\/span><\/pre>\n
                   end<\/span><\/pre>\n
                   return a<\/span><\/pre>\n
               end<\/span><\/pre>\n
     \u00a0 \u00a0qsort! (generic function with 1 method)\n\n<\/span><\/pre>\n
     \u00a0 \u00a0julia> sortperf(n) = qsort!(rand(n), 1, n)<\/span><\/pre>\n
     \u00a0 \u00a0sortperf (generic function with 1 method)\n\n<\/span><\/pre>\n
     \u00a0 \u00a0julia> @btime sortperf(5000)<\/span><\/pre>\n
         355.957 \u03bcs (2 allocations: 39.14 KiB)<\/span><\/pre>\n
    \n
    Julia Base.sort<\/strong><\/p>\n
    julia> @btime sort(rand(5000); alg=QuickSort)<\/span><\/pre>\n
    233.293 \u03bcs (11 allocations: 78.66 KiB)<\/span><\/pre>\n
    \u00a0<\/span><\/strong><\/p>\n
    \n
    \n
    \n
    \n
    \n
    \u00a0\u66fc\u5fb7\u5e03\u7f57\u7279\u96c6<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \n
    \u6211\u7684\u6d4b\u8bd5\u7ed3\u679c\u663e\u793a Numba(159us) vs Julia(72us)\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    Numba<\/strong><\/p>\n
        In [106]: @jit<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: def mandel_numba(z):<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 maxiter = 80<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 c = z<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 for n in range(maxiter):<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 if abs(z) > 2:<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 return n<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 z = z*z + c<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 return maxiter<\/span><\/pre>\n
        In [107]: @jit<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: def mandelperf_numba_mesh():<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 width = 26<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 height = 21<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 r1 = np.linspace(-2.0, 0.5, width)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 r2 = np.linspace(-1.0, 1.0, height)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 mandel_set = np.empty((width,height), dtype=int)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 for i in range(width):<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 for j in range(height):<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 mandel_set[i,j] = mandel_numba(r1[i] + 1j*r2[j])<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 return mandel_set<\/span><\/pre>\n
        In [109]: %timeit mandelperf_numba_mesh()<\/span><\/pre>\n
     \u00a0 \u00a0:1: NumbaWarning: Function \"mandelperf_numba_mesh\" failed type inference: Invalid usage of Function() with parameters ((int64 x 2), dtype=Function(<class 'int'>))<\/span><\/pre>\n
     \u00a0 \u00a0 * parameterized<\/span><\/pre>\n
        File \"\", line 7<\/span><\/pre>\n
     \u00a0 \u00a0[1] During: resolving callee type: Function()<\/span><\/pre>\n
     \u00a0 \u00a0[2] During: typing of call at \u00a0(7)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0@jit<\/span><\/pre>\n
     \u00a0 \u00a0:1: NumbaWarning: Function \"mandelperf_numba_mesh\" failed type inference: cannot determine Numba type of <class 'numba.dispatcher.LiftedLoop'><\/span><\/pre>\n
        File \"\", line 8<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0@jit<\/span><\/pre>\n
     \u00a0 \u00a0:1: NumbaWarning: Function \"mandelperf_numba_mesh\" was compiled in object mode without forceobj=True, but has lifted loops.<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0@jit<\/span><\/pre>\n
        159 \u00b5s \u00b1 7.55 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 10000 loops each) \u00a0 <\/span><\/pre>\n
    \n
    Julia<\/strong><\/p>\n
     \u00a0 \u00a0julia> function mandel(z)<\/span><\/pre>\n
                   c = z<\/span><\/pre>\n
                   maxiter = 80<\/span><\/pre>\n
                   for n = 1:maxiter<\/span><\/pre>\n
                       if myabs2(z) > 4<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 return n-1<\/span><\/pre>\n
                       end<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 z = z^2 + c<\/span><\/pre>\n
                   end<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 return maxiter<\/span><\/pre>\n
               end<\/span><\/pre>\n
     \u00a0 \u00a0mandel (generic function with 1 method)<\/span><\/pre>\n
        julia> mandelperf() = [ mandel(complex(r,i)) for i=-1.:.1:1., r=-2.0:.1:0.5 ]<\/span><\/pre>\n
     \u00a0 \u00a0mandelperf (generic function with 1 method)<\/span><\/pre>\n
        julia> @btime mandelperf()<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a072.942 \u03bcs (1 allocation: 4.44 KiB)<\/span><\/pre>\n
    \n
    \n
    \n
    \n
    \n
    \u00a0<\/strong><\/span><\/p>\n
    \u00a0\u6574\u578b\u8f6c\u5316<\/strong><\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \u00a0<\/span><\/p>\n
    \u6ce8\u610f\uff0c\u6211\u5220\u9664\u4e86\u58f0\u660e\u8bed\u53e5\uff0c\u56e0\u4e3a\u8fd9\u4e2a\u57fa\u51c6\u6d4b\u8bd5\u4f3c\u4e4e\u6ca1\u6709\u5fc5\u8981\u4f7f\u7528\u58f0\u660e\u8bed\u53e5\u3002\u6211\u7684\u6d4b\u8bd5\u7ed3\u679c\u663e\u793a:<\/span><\/p>\n
    \n
    1.\u539f\u59cb \u2013 Numpy (2340 \u03bcs) vs. Julia (176 \u03bcs)\u3002<\/span><\/p>\n
    2.\u5916\u5faa\u73af\u968f\u673a\u6570\u751f\u6210 \u2013 Cython (378 \u00b5s) vs Julia (176 \u03bcs)\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
       In [112]: def parse_int1_numpy():<\/span><\/pre>\n
             ...: \u00a0 \u00a0 for i in range(1,1000):<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 n = np.random.randint(0,2**31-1)<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 s = hex(n)<\/span><\/pre>\n
             ...: \u00a0 \u00a0 \u00a0 \u00a0 m = int(s,16)\n \u00a0 \u00a0 \u00a0 \u00a0 \n<\/span><\/pre>\n
        In [113]: %timeit parse_int1_numpy()<\/span><\/pre>\n
     \u00a0 \u00a02.34 ms \u00b1 80 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 100 loops each) <\/span><\/pre>\n
    \u00a0<\/strong><\/p>\n
    Julia<\/strong><\/p>\n
        julia> function<\/span> parseintperf(t)<\/pre>\n
                   local<\/span> n, m<\/pre>\n
                   for<\/span> i=1<\/span>:t<\/pre>\n
                       n = rand(UInt32<\/span>)<\/pre>\n
                       s = hex(n)<\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 m = UInt32(parse(Int64,s,16))<\/span><\/p>\n
    end<\/span><\/p>\n
    end<\/span><\/p>\n
     \u00a0 \u00a0parseintperf (generic function with 1 method)<\/p>\n
    <\/span><\/p>\n
     \u00a0 \u00a0julia> @btime parseintperf(1000)<\/span><\/p>\n
    176.061 \u03bcs (2000 allocations: 93.75 KiB) \u00a0 \u00a0<\/span><\/p>\n
    \u00a0<\/strong><\/p>\n
    Cython(\u5916\u5faa\u73af)<\/strong><\/p>\n
        In [116]: %%cython<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: import numpy as np<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: import cython<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: <\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: @cython.boundscheck(False)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: @cython.wraparound(False)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: cpdef parse_int_vec_cython():<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 cdef:<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 long i,m<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 long[:] n<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 n = np.random.randint(0,2**31-1,1000)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 for i in range(1,1000):<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: \u00a0 \u00a0 \u00a0 \u00a0 m = int(hex(n[i]),16)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 ...: <\/span><\/pre>\n
     \u00a0 \u00a0In [118]: %timeit parse_int_vec_cython()<\/span><\/pre>\n
        378 \u00b5s \u00b1 5.29 \u00b5s per loop (mean \u00b1 std. dev. of 7 runs, 1000 loops each)<\/span><\/pre>\n
    \u00a0<\/strong><\/p>\n
    Julia(\u5916\u5faa\u73af)<\/strong><\/p>\n
      \u00a0 julia> function parseintperf2(t)<\/span><\/pre>\n
                   n = rand(UInt32, t)<\/span><\/pre>\n
                   for i=1:t<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 s = hex(n[i])<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 \u00a0 m = UInt32(parse(Int64,s,16))<\/span><\/pre>\n
                   end<\/span><\/pre>\n
               end<\/span><\/pre>\n
     \u00a0 \u00a0parseintperf2 (generic function with 1 method)\n\n<\/span><\/pre>\n
        julia> @btime parseintperf2(1000)<\/span><\/pre>\n
     \u00a0 \u00a0 \u00a0176.053 \u03bcs (2003 allocations: 97.97 KiB)<\/span><\/pre>\n
    \u00a0<\/span><\/p>\n
    \u5478\uff0c\u8ba9\u6211\u4eec\u56de\u5230\u6700\u521d\u7684\u535a\u5ba2\u5e16\u5b50\u3002<\/span><\/p>\n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \u8bed\u8a00<\/span><\/strong><\/span><\/strong><\/section>\n<\/section>\n
    \u00a0<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \u4f5c\u8005\u5bf9Julia\u8bed\u8a00\u672c\u8eab\u5c31\u6709\u4e00\u4e9b\u6279\u8bc4\uff1a
    \n<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \n
    \u4e0d\u5e73\u8861(Unbalanced)\u7684end<\/span><\/p>\n<\/blockquote>\n
    \u6211\u4e2a\u4eba\u8ba4\u4e3a\u8fd9\u5e76\u4e0d\u662f\u4ec0\u4e48\u95ee\u9898\uff0c\u800c\u662f\u4e2a\u4eba\u559c\u6b22\u3002\u5927\u591a\u6570\u8bed\u8a00\u90fd\u6709\u4e00\u4e2a\u5757\u7684\u7ed3\u5c3e\u6807\u8bb0\u3002\u4f46\u662f\u6211\u4e0d\u660e\u767d\u4e0d\u5e73\u8861\u662f\u4ec0\u4e48\u610f\u601d\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \n
    \u4f7f\u7528::<\/span><\/p>\n<\/blockquote>\n
    \u8fd9\u4e0d\u662f\u95ee\u9898\uff0c\u662f\u4e2a\u4eba\u559c\u597d\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \n
    \u975e\u6b63\u7edf\u7684\u64cd\u4f5c\u7b26\uff0c\u6807\u70b9\uff0c\u548c\u591a\u884c\u6ce8\u91ca\u8bed\u6cd5<\/span><\/p>\n<\/blockquote>\n
    Julia\u7684\u6807\u70b9\u4f3c\u4e4e\u4e0e\u8bb8\u591a\u7f16\u7a0b\u8bed\u8a00\u5e76\u6ca1\u6709\u4ec0\u4e48\u592a\u5927\u7684\u4e0d\u540c\u3002\u8fd9\u611f\u89c9\u53c8\u662f\u4e00\u4e2a\u4e2a\u4eba\u559c\u597d\u95ee\u9898\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \n
    1\u57fa\u6570\u7684\u7d22\u5f15\uff08one-based-indexing\uff09<\/span><\/p>\n<\/blockquote>\n
    \u8fd9\u662f\u6211\u6700\u559c\u6b22\u7684\u3002\u4e24\u4e2a\u9635\u8425\u90fd\u6709\u4f18\u70b9\u3002\u8fd9\u771f\u7684\u53d6\u51b3\u4e8e\u4f60\u7684\u4f7f\u7528\u548c\u5177\u4f53\u60c5\u5883\u3002\u6b63\u5982\u5927\u5bb6\u5728\u8bba\u575b\u4e0a\u6307\u51fa\u7684\u90a3\u6837\uff0c\u6709\u4e00\u4e9b\u5305\uff08\u6bd4\u5982OffserArray\u548cPemutedDimsArray\uff09\u4f7f\u75280\u4f5c\u4e3a\u57fa\u6570\u4f1a\u66f4\u52a0\u81ea\u7136\u3002\u633a\u597d\u7684\uff0c\u81f3\u5c11\u6211\u4eec\u6709\u9009\u62e9\u6743\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \n
    \u6807\u51c6\u6587\u6863\u7cfb\u7edf\u4f7f\u7528Markdown<\/span><\/p>\n<\/blockquote>\n
    \u51e0\u4e2a\u6708\u524d\u6211\u5f00\u59cb\u4f7f\u7528Julia\u7f16\u7a0b\uff0c\u53d1\u73b0\u8fd9\u4e2a\u6587\u6863\u7cfb\u7edf\u5f88\u76f4\u89c2\u548c\u7b80\u5355\u3002\u7136\u800c\uff0c\u6211\u8ba4\u4e3a\u4f5c\u8005\u5728\u4e00\u7bc7\u540e\u7eed\u7684\u535a\u5ba2\u4e2d\u63d0\u4f9b\u4e86\u66f4\u591a\u7684\u4fe1\u606f\uff0c\u6211\u8ba4\u4e3a\u4f5c\u8005\u63d0\u51fa\u4e86\u4e00\u4e2a\u4e0d\u9519\u7684\u89c2\u70b9\u3002\u770b\u6765\u610f\u89c1\u662f\u53d1\u751f\u4e86\u5206\u6b67\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \u5728Julia\u7684\u6848\u4f8b\u4e2d\uff0c\u8fd9\u4e2a\u9009\u62e9\u662f\u5728\u65e9\u671f\u4f7f\u7528reST\u7684\u51e0\u4e2a\u75db\u70b9\u4e0b\u51b3\u5b9a\u7684\u3002\u5728\u505a\u4e86\u5404\u79cd\u5404\u6837\u7684\u7814\u7a76\u4e4b\u540e\u3002\u5728\u6211\u770b\u6765\uff0cReStructuredText\u66f4\u52a0\u4e25\u683c\u548c\u5b9e\u7528\uff0c\u800cMarkdown\u66f4\u52a0\u5bb9\u6613\u5b66\u4e60\u548c\u5b9e\u7528\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \u6700\u540e\uff0c\u8fd9\u91cc\u5176\u5b9e\u5e76\u6ca1\u6709\u5bf9\u4e0e\u9519\u3002\u6211\u4eec\u77e5\u9053\u5f00\u53d1\u4eba\u5458\u4ece\u6765\u6ca1\u6709\u5728\u4e00\u4ef6\u4e8b\u4e0a\u8fbe\u6210\u8fc7\u4e00\u81f4\u3002\u5e78\u8fd0\u7684\u662f\uff0c\u5f00\u53d1\u4eba\u5458\u603b\u6570\u53ef\u4ee5\u9009\u62e9\u81ea\u5df1\u505a\u51b3\u5b9a\u3002\u6211\u4eec\u5b8c\u5168\u5c0a\u91cd\u90a3\u4e9b\u9009\u62e9ReStructuredText\u800c\u4e0d\u662fMarkdown\u7684\u9879\u76ee\uff0c\u53cd\u4e4b\u4ea6\u7136\u3002<\/span><\/p>\n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \u5b89\u5168<\/span><\/strong><\/span><\/strong><\/section>\n<\/section>\n
    \u00a0<\/strong><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \u4f5c\u8005\u8ba4\u4e3a\uff0c\u4f7f\u7528ccall\u65f6Julia\u5f88\u5bb9\u6613\u4ea7\u751f\u5185\u5b58\u533a\u6bb5\u9519\u8bef\u3002\u6211\u53d1\u73b0\u5b83\u4e0e\u73b0\u72b6\u7684Julia\u7248\u672c\u4e0d\u7b26\u3002\u7531\u4e8e\u9519\u8bef\u7684\u5e93\u540d\u79f0\u548c\u6570\u636e\u7c7b\u578b\uff0c\u6211\u65e0\u6cd5\u751f\u6210\u4e00\u4e2a\u5185\u5b58\u533a\u6bb5\u9519\u8bef\u3002\u7136\u800c\uff0c\u6211\u53ef\u4ee5\u6536\u5230\u4e00\u4e2a\u5f88\u597d\u7684\u9519\u8bef\u4fe1\u606f\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    julia> val = ccall((:getenv, \"libc.so.6\"), Ptr{UInt8}, (Ptr{UInt8},), var)<\/span><\/pre>\n
     ERROR: error compiling anonymous: could not load library \"libc.so.6\"<\/span><\/pre>\n
     dlopen(libc.so.6.dylib, 1): image not found\n \n<\/span><\/pre>\n
    julia> val = ccall((:getenv, \"libc.dylib\"), Ptr{UInt8}, (Ptr{UInt8},), 123)<\/span><\/pre>\n
     ERROR: MethodError: no method matching unsafe_convert(::Type{Ptr{UInt8}}, ::Int64)<\/span><\/pre>\n
     Closest candidates are:<\/span><\/pre>\n
     unsafe_convert(::Type{Ptr{UInt8}}, ::Symbol) at pointer.jl:35<\/span><\/pre>\n
     unsafe_convert(::Type{Ptr{UInt8}}, ::String) at pointer.jl:37<\/span><\/pre>\n
     unsafe_convert(::Type{T<:Ptr}, ::T<:Ptr) where T<:Ptr at essentials.jl:152 ...<\/span><\/pre>\n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \u00a0Printf\/Sprintf<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \u5bf9\u4e8e\u751f\u6210\u5927\u91cf\u5b8f\u7684@sprintf\u548c@printf\uff0c\u4f5c\u8005\u597d\u50cf\u5bf9\u6b64\u6709\u70b9\u4e0d\u6ee1\u3002
    \n<\/span><\/p>\n
    \n
    \u90a3\u65f6\u5019\u53ef\u80fd\u662f\u771f\u7684\uff0c\u4f46\u662f\u73b0\u5728\u6709\u5f88\u591a\u89e3\u51b3\u65b9\u6cd5\u3002\u8fd9\u662f\u4e00\u4e2a\u6b63\u5728\u7ecf\u5386\u5feb\u901f\u521b\u65b0\u7684\u9886\u57df\u3002\u4e00\u4e9bJulia\u7684\u5305\u53ef\u4ee5\u89e3\u51b3\u8fd9\u4e2a\u95ee\u9898\u3002\u6bd4\u5982\uff1aFormatting.jl\u548cStringLiterals.jl\u3002<\/span><\/p>\n
    \n
    \u4e0b\u9762\u8fd9\u884c\u4ee3\u7801\u53ea\u751f\u621063\u884c\u672c\u5730\u4ee3\u7801\uff0c\u76f8\u6bd4\u4e4b\u4e0b\uff0c\u4f5c\u8005\u5728\u8fc7\u53bb\u8bf4\u6709500+\u884c\u3002<\/span><\/p>\n
    julia> @code_native Formatting.printfmt(“{1:>4s} {2:.2f}”, “abc”, 12)<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \u5982\u679c\u6211\u7f16\u8bd1\u4e0b\u9762\u7684C\u4ee3\u7801\uff0c\u6211\u53ea\u4f1a\u5f97\u523021\u884c\u6c47\u7f16\u6307\u4ee4\u3002\u60f3\u8981\u6253\u8d25C\u8bed\u8a00\u8fd8\u662f\u6709\u70b9\u56f0\u96be\u7684\u3002<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    void f(char *buffer, const char *a, double b) {<\/span><\/pre>\n
      sprintf(buffer, \"this is a %s %g\", a, b);<\/span><\/pre>\n
    }<\/span><\/pre>\n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \u00a0\u5355\u5143\u6d4b\u8bd5<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \u4f5c\u8005\u8ba4\u4e3a\uff0cJulia\u7684\u5355\u5143\u6d4b\u8bd5\u5e93\u662f\u53d7\u9650\u7684\uff1a
    \n<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \n
    \u5355\u5143\u6d4b\u8bd5\u5e93\u662f\u975e\u5e38\u57fa\u672c\u7684\uff0c\u81f3\u5c11\u5bf9\u4e8eC++\u548cJava\u800c\u8a00\u662f\u8fd9\u6837\u7684\u3002\u53ef\u4ee5\u8bf4\uff0cFactCheck\u662f\u6700\u53d7\u6b22\u8fce\u7684\u9009\u62e9\uff0c\u4f46\u662f\u9664\u4e86\u5947\u602a\u7684API\u4e4b\u5916\uff0c\u5b83\u662f\u76f8\u5f53\u53d7\u9650\u7684\uff0c\u800c\u4e14\u4e5f\u57fa\u672c\u4e0d\u518d\u5f00\u53d1\u3002<\/span><\/p>\n<\/blockquote>\n
    \u00a0<\/span><\/p>\n
    \u5f53\u6211\u5728\u5f00\u53d1SASLib.jl\u7684\u65f6\u5019\uff0c\u6211\u53ea\u4f7f\u7528Base.Test\u3002\u6211\u89c9\u5f97\u5b83\u5bf9\u7528\u6237\u662f\u975e\u5e38\u53cb\u597d\u7684\uff0c\u5e76\u4e14\u5b83\u505a\u4e86\u6211\u6b63\u9700\u8981\u7684\u4e8b\u60c5\u3002\u4f5c\u8005\u63d0\u5230\u4e86FactCheck\u5305\uff0c\u8fd9\u4f3c\u4e4e\u662f\u4e00\u4e2a\u5408\u7406\u7684\u9644\u52a0\u6761\u4ef6\u3002\u6211\u4ee5\u524d\u6ca1\u7528\u8fc7FactCheck\uff0c\u6240\u4ee5\u6211\u5728\u8fd9\u91cc\u5c31\u4e0d\u53d1\u8868\u989d\u5916\u7684\u89c1\u89e3\u4e86\u3002<\/span><\/p>\n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \n
    \u00a0\u53d1\u5c55<\/span><\/p>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n<\/section>\n
    \u6709\u8da3\u7684\u662f\uff0c\u4f5c\u8005\u8bf4\u4ed6\u8003\u8651\u8fc7\u4e3aJulia\u9879\u76ee\u505a\u8d21\u732e\uff0c\u4f46\u662f\u4ed6\u4e0d\u559c\u6b22Julia\u7684\u8fd0\u884c\u65f6\u662f\u7531\u51e0\u79cd\u7f16\u7a0b\u8bed\u8a00C\/C++\uff0cLisp\uff0c\u548cJulia\u6784\u5efa\u7684\u3002
    \n<\/span><\/p>\n
    \u00a0<\/span><\/p>\n
    \u6211\u4e0d\u660e\u767d\u8fd9\u6709\u4ec0\u4e48\u5927\u4e0d\u4e86\u7684\u3002\u5982\u679c\u6211\u7edf\u8ba1\u4e00\u4e0b\u5bf9\u4e8eJulia 0.6\u5404\u4e2a\u8bed\u8a00\u5206\u522b\u6709\u591a\u5c11\u884c\u4ee3\u7801\uff1a<\/span><\/p>\n