{"id":177,"date":"2013-11-09T07:00:47","date_gmt":"2013-11-08T23:00:47","guid":{"rendered":"http:\/\/blog.stlover.org\/?p=177"},"modified":"2013-11-12T22:38:37","modified_gmt":"2013-11-12T14:38:37","slug":"cuda%e5%ad%a6%e4%b9%a0%e4%b8%8e%e5%b0%9d%e8%af%95helloworld","status":"publish","type":"post","link":"http:\/\/blog.xuhao1.me\/?p=177","title":{"rendered":"CUDA\u5b66\u4e60\u4e0e\u5c1d\u8bd5:Hello,World"},"content":{"rendered":"

\u5b66\u4e60\u4e00\u4e2a\u4e1c\u897f\u603b\u8981\u8bb0\u4e0b\u6765\uff0c\u6700\u8fd1\u8054\u7cfb\u4e86\u5b9e\u9a8c\u5ba4\u51c6\u5907\u505a\u6709\u5173\u7b49\u4f53\u7684\u8ba1\u7b97\uff0c\u6d69\u6d69\u7684\u7269\u7406\u7cfb\u975e\u8d44\u6df1\u7801\u519c\u751f\u6daf\u4e5f\u5c06\u8fdb\u5165\u6b63\u8f68\u3002\u4e07\u4e8b\u5f00\u5934\u96be\uff0c\u5f53\u7136\u4ece\u6784\u67b6\u8d70\u8d77\uff0c\u5f00\u59cb\u5b66\u4e60CUDA\uff0c\u7b2c\u4e00\u6b65\u4f7f\u7528CUDA\u91cd\u6784\u4ee5\u524d\u505a\u8fc7\u7684\u5efa\u8bae\u79d1\u5b66\u8fd0\u7b97\u7b97\u4f8b\uff0c\u8bf8\u5982\u6709\u9650\u5dee\u5206\u89e3\u70ed\u6d41\u5566\uff0c\u5feb\u901f\u5085\u91cc\u53f6\u53d8\u6362\u5566\uff0c\u591a\u4f53\u8fd0\u52a8\u6a21\u62df\u5566\uff0c\u54c8\u5bc6\u987f\u539f\u7406\u6a21\u62df\u554a\uff0c\u4ee5\u53ca\u5b8c\u6210\u8ba1\u7b97\u7269\u7406\u4f5c\u4e1a\u3002\u7136\u540e\u5728CFD\u7684\u5b9e\u8df5\u4e2d\u5168\u90e8\u91c7\u7528CUDA\u6784\u67b6\u3002<\/p>\n

<\/p>\n

\u76ee\u5f55<\/a><\/p>\n

CUDA\u7684\u4e00\u4e9b\u626f\u6de1<\/h2>\n

\"Screen<\/a><\/p>\n

\u65b0\u5165\u624b\u7684\u7535\u8111\u5df2\u7ecf\u8fbe\u5230\u4e86\u777f\u98914 cores 3.5Ghz\u7684\u80fd\u529b\uff0c\u4f5c\u4e3a\u516b\u7ebf\u7a0b\u51fa\u73b0\u5df2\u7ecf\u76f8\u5f53\u7684\u5f3a\u5927\uff0c\u4ee5\u524d\u4e5f\u5c1d\u8bd5\u5199\u8fc7\u4e00\u4e9b\u591a\u7ebf\u7a0b\u7684\u4ee3\u7801\uff0c\u4ee3\u8868\u662f<\/p>\n

\u4f7f\u7528Java\u505a\u521a\u4f53\u529b\u5b66\u6a21\u62df\uff0c\u7136\u540e\u4f7f\u7528\u9057\u4f20\u7b97\u6cd5\u7528\u521a\u4f53\u529b\u5b66\u6a21\u62df\u7ed3\u679c\u4f5c\u4e3a\u63a7\u5236\u5668\u7684\u8bc4\u4ef7\u503c\u6765\u6f14\u5316PIDs\u7684\u63a7\u5236\u5668\uff0c\u8bf4\u767d\u4e86\u4e5f\u5c31\u662f\u662f\u4e00\u4e2a16\u7ebf\u7a0b\u7684\u5927\u73a9\u5177\u3002\u6ca1\u6709\u4ec0\u4e48\u5b9e\u9645\u6548\u679c\u3002<\/p><\/blockquote>\n

\u4e0d\u8fc7\u6bd5\u7adf\u662f\u5e76\u884c\u7684\u73a9\u610f\uff0c\u76f8\u6bd4\u4e0b\u5de6\u56fe\u6240\u793a\u7684\u7ebf\u7a0b\u6c60\u7adf\u7136\u8981\u7528\u6548\u7387\u4f4e\u4e0b\u7684\u4eba\u8111\u62a2\u5360\u5f0fc(brain)pu\u65f6\u95f4\u5206\u7247\u6765\u5b8c\u6210\uff0c\u800c\u4e14\u5728\u7ebf\u7a0b\u5207\u6362\u65f6\u5019\u6548\u7387\u6781\u5176\u4f4e\u4e0b\uff0ccpu\u5728\u7ebf\u7a0b\u5207\u6362\u65f6\u5019\u9762\u4e34\u7740\u6548\u7387\u6781\u4f4e\u7684\u7a98\u5883\uff0c\u800c\u4e14\u5e38\u5e38\u56e0\u4e3a\u8fc7\u70ed\uff0c\u8fc7\u51b7\uff0c\u80fd\u91cf\u4e0d\u8db3\uff0c\u80fd\u91cf\u8fc7\u5269\u800c\u7f62\u5de5\uff0c\u751a\u81f3\u4f1a\u56e0\u4e3a\u89c6\u89c9\u7cfb\u7edf\u53d7\u5230\u7279\u5b9a\u4eba\u7269\u7684\u523a\u6fc0\u800c\u5f00\u59cb\u5927\u91cf\u7684\u5b95\u673a\uff08\u53d1\u5446\uff09\u3002\u4e0d\u5f97\u4e0d\u559f\u53f9\u4eba\u751f\u4e4b\u8270\u96be\uff0c\u6069\uff0c\u7b2c\u4e00\u90e8\u5206\u56e0\u4e3a\u548c\u989c\u8272\u76f8\u5173\u7684\u7406\u7531\u6d89\u53ca\u5230\u4e2a\u4eba\u9690\u79c1\uff0c\u4f5c\u4e3a\u672c\u671f\u671f\u95f4\u6700\u5927\u7684\u652f\u7ebf\u4efb\u52a1\u4e4b\u4e00\u5c31\u5148\u9690\u53bb\u4e86\u3002<\/p>\n

\u626f\u4e86\u8fd9\u4e48\u591a\uff0c\u51e0\u767e\u4e2a\u7ebf\u7a0b\u8dd1\u4ee3\u7801\u786e\u5b9e\u662f\u4e00\u4ef6\u5f88\u723d\u7684\u4e8b\u60c5\uff0c\u662f\u60f3\u5982\u679c\u6211\u6709\u5206\u8eab\uff0c\u4e00\u4e2a\u53bb\u505a\u4f5c\u4e1a\u4e00\u4e2a\u53bb\u590d\u4e60\u8003\u8bd5\u4e00\u4e2a\u53bb\u5b66CFD\u4e00\u4e2a\u53bb\u5b66CUDA\uff0c\u7136\u540e\u672c\u4f53\u53bb\u6069\uff08zhui\uff09\u4f60\uff08mou\uff09\u61c2\uff08gu\uff09\u5f97\uff08niang\uff09\uff0c\u90a3\u4e48\u4eba\u751f\u5e94\u8be5\u5982\u4f55\u7684\u7f8e\u597d\u3002\u4e0d\u8fc7\u4e8b\u60c5\u603b\u5f97\u4e00\u4ef6\u4ef6\u6765\u6211\u4e5f\u6ca1\u6709\u5206\u8eab\u3002\u73b0\u5728\u5c31\u4eceCUDA\u7684hello,world\u5b66\u8d77\uff0c\u8bb2\u8d77\u3002<\/p>\n

\u5b66\u4e60\u8d44\u6e90\uff1a<\/p>\n

http:\/\/docs.nvidia.com\/cuda\/cuda-c-programming-guide\/index.html<\/p>\n

\u6211\u8fd9\u8fb9\u5e76\u4e0d\u8981\u505a\u4e00\u4e2a\u7ffb\u8bd1\u5de5\u4f5c\uff0c\u800c\u662f\u5728\u5b66\u4e60\u7684\u8fc7\u7a0b\u4e2d\u8bb2\u51fa\u6765\uff0c\u8bb2\u51fa\u6765\u81ea\u5df1\u66f4\u61c2\uff0c\u4e5f\u80fd\u5206\u4eab\u77e5\u8bc6\u3002<\/p>\n

Hello,World\u4ee5\u53ca\u7ebf\u7a0b\u6a21\u578b<\/h2>\n

\u5148\u770b\u5b98\u65b9\u7ed9\u51fa\u7684\u4e00\u4e2a\u6837\u4f8b\uff1a<\/p>\n

    \n
  1. \/\/\u00a0Kernel\u00a0definition<\/span><\/li>\n
  2. __global__\u00a0<\/span>void<\/span>\u00a0<\/span>VecAdd<\/span>(<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
  3. {<\/span><\/li>\n
  4. \u00a0\u00a0\u00a0\u00a0<\/span>int<\/span>\u00a0i\u00a0<\/span>=<\/span>\u00a0threadIdx<\/span>.<\/span>x<\/span>;<\/span><\/li>\n
  5. \u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>+<\/span>\u00a0B<\/span>[<\/span>i<\/span>];<\/span><\/li>\n
  6. }<\/span><\/li>\n
  7. \u00a0<\/span><\/li>\n
  8. int<\/span>\u00a0main<\/span>()<\/span><\/li>\n
  9. {<\/span><\/li>\n
  10. \u00a0\u00a0\u00a0\u00a0<\/span>…<\/span><\/li>\n
  11. \u00a0\u00a0\u00a0\u00a0<\/span>\/\/\u00a0Kernel\u00a0invocation\u00a0with\u00a0N\u00a0threads<\/span><\/li>\n
  12. \u00a0\u00a0\u00a0\u00a0<\/span>VecAdd<\/span><<<<\/span>1<\/span>,<\/span>\u00a0N<\/span>>>>(<\/span>A<\/span>,<\/span>\u00a0B<\/span>,<\/span>\u00a0C<\/span>);<\/span><\/li>\n
  13. \u00a0\u00a0\u00a0\u00a0<\/span>…<\/span><\/li>\n
  14. }<\/span><\/li>\n<\/ol>\n

    \u5341\u5206\u7b80\u5355\uff0c\u8fd9\u662f\u4e00\u4e2a\u8ba1\u7b97\u957f\u77e2\u91cf\u76f8\u52a0\u7684\u4ee3\u7801\u3002\u5341\u5206\u6709\u6548\u7684\u9732\u51fa\u4e86\u6a21\u578b\u7684\u771f\u9762\u76ee\uff1a<\/p>\n

    \u6839\u636e\u76ee\u524d\u7684\u89c2\u5bdfCUDA\u57fa\u4e8e\u4e00\u79cd\u975e\u5e38\u63a5\u8fd1\u79d1\u5b66\u8fd0\u7b97\u7684\u6a21\u578b\uff1a\u5757-\u7ebf\u7a0b\u6a21\u578b\uff0c<\/p>\n

    \u6bcf\u4e2a\u5757\u6709\u82e5\u5e72\u7ebf\u7a0b\uff081024\uff09\uff0c\u53ef\u4ee5\u6709\u5f88\u591a\u5f88\u591a\u5757\uff0c\u90fd\u53ef\u4ee5\u4f7f\u7528\u4e00\u4e2a\u4e09\u7ef4\u5411\u91cf\u6765\u5b9a\u4f4d\u3002<\/p>\n

    \u800c\u5728\u672c\u4f8b\u4e2d\uff0c\u957f\u77e2\u91cf\u4f7f\u7528threadIdx.x\u6765\u5b9a\u4f4d\uff0c\u8fd9\u4e5f\u5c31\u662fcuda\u6a21\u578b\u7684\u6f02\u4eae\u4e4b\u5904\uff0c\u53ea\u9700\u8981\u4e00\u4e2a\u4e09\u7ef4\u5411\u91cf\u5c31\u53ef\u4ee5\u786e\u5b9a\u6bcf\u4e2a\u6a21\u5757\u7684\u4f4d\u7f6e\uff0c\u8fd9\u4e2a\u7406\u5ff5\u867d\u7136\u548c\u666e\u901a\u610f\u4e49\u4e0a\u7684\u7f16\u7a0b\u76f8\u53bb\u751a\u8fdc\uff0c\u4f46\u662f\u5bf9\u4e8e\u79d1\u5b66\u8fd0\u7b97\uff08\u5927\u91cf\u7684\u77e9\u9635\u8fd0\u7b97\uff09\u662f\u975e\u5e38\u6f02\u4eae\u7684\uff0c\u800c\u4e14\u5728\u5927\u591a\u6570\u65f6\u5019\u77e9\u9635\u6b62\u4e8e\u4e8c\u7ef4\u4e09\u7ef4\u5373\u53ef\uff0c\u56e0\u4e3a\u7528cuda\u505a\u6a21\u62df\u7684\u5316\u7a7a\u95f4\u4e5f\u4e0d\u8fc7\u4e09\u7ef4\u3002<\/p>\n

    \u81f3\u4e8e\u5177\u4f53\u7684\u7ebf\u7a0b\u8fd0\u884c\u987a\u5e8f\uff0c\u90a3\u662fNVIDIA\u5e94\u8be5\u8003\u8651\u7684\u4e8b\u60c5\uff0c\u4e0d\u8fc7\u6839\u636eguide\u4e2d\u7684\u4ecb\u7ecd\uff0c\u5176\u5b9e\u548cc\u7684\u591a\u7ef4\u6570\u7ec4\u50a8\u5b58\u6ca1\u4ec0\u4e48\u533a\u522b\u3002<\/p>\n

    \u6069\uff0c\u73b0\u5728\u6211\u4eec\u6709\u4e86\u4e00\u4e2a\u6a21\u578b\u548c\u4e00\u4e2a\u5199\u597d\u7684\u4ee3\u7801\uff0c\u5982\u4f55\u7f16\u8bd1\u5b83\u5e76\u4e14\u8fd0\u884c\u5462\uff0c\u6211\u4eec\u4ee5\u5982\u4e0b\u73af\u5883\u4e3a\u4f8b<\/p>\n

    Macbook pro retina<\/p>\n

    System<\/strong>\u00a0: OS X 10.9<\/p>\n

    CPU<\/strong>:\u00a0i7-4850HQ<\/p>\n

    Memory<\/b>\u00a0 16 GB 1600 MHz DDR3<\/p>\n

    Graphics<\/b>\u00a0 NVIDIA GeForce GT 750M 2048 MB<\/p>\n

    \u7f16\u8bd1\u73af\u5883\uff1a<\/p>\n

    gcc (MacPorts gcc49 4.9-20131027_0) 4.9.0 20131027 (experimental),GNU Make 3.81<\/p>\n

    \u5f00\u53d1\u5de5\u5177\uff1avim<\/p><\/blockquote>\n

    \u5148\u81ea\u5df1\u53bb\u88c5toolkit\u3002\u53c2\u7167<\/p>\n

    http:\/\/docs.nvidia.com\/cuda\/cuda-getting-started-guide-for-mac-os-x\/index.html<\/p>\n

    \u914d\u7f6e~\/.bash_profile\uff0c\u52a0\u5165<\/p>\n

      \n
    1. export\u00a0PATH<\/span>=\/<\/span>Developer<\/span>\/<\/span>NVIDIA<\/span>\/<\/span>CUDA<\/span>–<\/span>5.5<\/span>\/<\/span>bin<\/span>:<\/span>$PATH<\/span><\/li>\n
    2. export\u00a0DYLD_LIBRARY_PATH<\/span>=\/<\/span>Developer<\/span>\/<\/span>NVIDIA<\/span>\/<\/span>CUDA<\/span>–<\/span>5.5<\/span>\/<\/span>lib<\/span>:<\/span>$DYLD_LIBRARY_PATH<\/span><\/li>\n<\/ol>\n

      \u91cd\u65b0\u6253\u5f00\u7ec8\u7aef\u6216\u8005\u91cd\u542f\u7535\u8111\uff0c\u7136\u540e<\/p>\n

        \n
      1. ~<\/span>$\u00a0nvcc\u00a0<\/span>—<\/span>version<\/span><\/li>\n
      2. nvcc<\/span>:<\/span>\u00a0NVIDIA\u00a0<\/span>(<\/span>R<\/span>)<\/span>\u00a0<\/span>Cuda<\/span>\u00a0compiler\u00a0driver<\/span><\/li>\n
      3. Copyright<\/span>\u00a0<\/span>(<\/span>c<\/span>)<\/span>\u00a0<\/span>2005<\/span>–<\/span>2013<\/span>\u00a0NVIDIA\u00a0<\/span>Corporation<\/span><\/li>\n
      4. Built<\/span>\u00a0on\u00a0<\/span>Thu_Sep__5_10<\/span>:<\/span>17<\/span>:<\/span>14<\/span>_PDT_2013<\/span><\/li>\n<\/ol>\n

        Cuda<\/span>\u00a0compilation\u00a0tools<\/span>,<\/span>\u00a0release\u00a0<\/span>5.5<\/span>,<\/span>\u00a0V5<\/span>.<\/span>5.0<\/span><\/p>\n

        \u6069\uff0c\u5b89\u88c5\u5b8c\u6210\u4e86\uff0c\u4e8e\u662f\u6211\u4eec\u6765hello\u4e2aworld<\/p>\n

        \u628a\u4ee3\u7801\u5199\u5165(vim \u6709cuda\u7684\u9ad8\u4eae\uff0c\u8d5e\u4e00\u4e2a)<\/p>\n

          \n
        1. \/\/test.cu<\/span><\/li>\n
        2. #include<\/span>“stdio.h”<\/span><\/li>\n
        3. #include<\/span><cuda_runtime.h><\/span><\/li>\n
        4. #define<\/span>\u00a0N\u00a0<\/span>100<\/span><\/li>\n
        5. \/\/\u00a0Kernel\u00a0definition<\/span><\/li>\n
        6. __global__\u00a0<\/span>void<\/span>\u00a0<\/span>VecAdd<\/span>(<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
        7. {<\/span><\/li>\n
        8. \u00a0\u00a0\u00a0\u00a0<\/span>int<\/span>\u00a0i\u00a0<\/span>=<\/span>\u00a0threadIdx<\/span>.<\/span>x<\/span>;<\/span><\/li>\n
        9. \u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>+<\/span>\u00a0B<\/span>[<\/span>i<\/span>];<\/span><\/li>\n
        10. }<\/span><\/li>\n
        11. \u00a0<\/span><\/li>\n
        12. int<\/span>\u00a0main<\/span>()<\/span><\/li>\n
        13. {<\/span><\/li>\n
        14. \u00a0\u00a0\u00a0\u00a0<\/span>\/\/\u00a0Kernel\u00a0invocation\u00a0with\u00a0N\u00a0threads<\/span><\/li>\n
        15. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“Hello,World\\n”<\/span>);<\/span><\/li>\n
        16. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>\u00a0<\/span>*<\/span>A<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>100<\/span>],*<\/span>B<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>100<\/span>],*<\/span>C<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>100<\/span>];<\/span><\/li>\n
        17. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
        18. \u00a0\u00a0\u00a0\u00a0<\/span>{<\/span><\/li>\n
        19. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0A<\/span>[<\/span>i<\/span>]=<\/span>i<\/span>;<\/span><\/li>\n
        20. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0B<\/span>[<\/span>i<\/span>]=<\/span>i<\/span>*<\/span>i<\/span>;<\/span><\/li>\n
        21. \u00a0\u00a0\u00a0\u00a0<\/span>}<\/span><\/li>\n
        22. \u00a0\u00a0\u00a0\u00a0<\/span>VecAdd<\/span><<<<\/span>1<\/span>,<\/span>\u00a0N<\/span>>>>(<\/span>A<\/span>,<\/span>\u00a0B<\/span>,<\/span>\u00a0C<\/span>);<\/span><\/li>\n
        23. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
        24. \u00a0\u00a0\u00a0\u00a0<\/span>{<\/span><\/li>\n
        25. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“%f\u00a0“<\/span>,<\/span>C<\/span>[<\/span>i<\/span>]);<\/span><\/li>\n
        26. \u00a0\u00a0\u00a0\u00a0<\/span>}<\/span><\/li>\n
        27. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“\\n”<\/span>);<\/span><\/li>\n
        28. }<\/span><\/li>\n<\/ol>\n

          \u4ee5\u53camakefile<\/p>\n

            \n
          1. test<\/span>:<\/span>test<\/span>.<\/span>o<\/span><\/li>\n
          2. \u00a0\u00a0\u00a0\u00a0nvcc\u00a0<\/span>–<\/span>o\u00a0test\u00a0test<\/span>.<\/span>o<\/span><\/li>\n
          3. test<\/span>.<\/span>o<\/span>:<\/span>test<\/span>.<\/span>cu<\/span><\/li>\n
          4. \u00a0\u00a0\u00a0\u00a0nvcc\u00a0<\/span>–<\/span>I\u00a0<\/span>\/<\/span>Developer<\/span>\/<\/span>NVIDIA<\/span>\/<\/span>CUDA<\/span>–<\/span>5.5<\/span>\/<\/span>include\u00a0<\/span>–<\/span>c\u00a0test<\/span>.<\/span>cu<\/span><\/li>\n
          5. clean<\/span>:<\/span><\/li>\n
          6. \u00a0\u00a0\u00a0\u00a0rm\u00a0<\/span>*.<\/span>o\u00a0test<\/span><\/li>\n<\/ol>\n

            make\uff0c\u6f47\u6d12\u901a\u8fc7<\/p>\n

            \u8fd0\u884c\uff0c<\/p>\n

              \n
            1. CUDA$\u00a0<\/span>.\/<\/span>test<\/span><\/li>\n
            2. Hello<\/span>,<\/span>World<\/span><\/li>\n
            3. 0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>0.000000<\/span>\u00a0<\/span>…………<\/span><\/li>\n<\/ol>\n

              \u54a6\u3002\u3002\u5b8c\u5168\u6ca1\u6709\u8fd0\u884c\u3002\u3002\u3002\u4f3c\u4e4e\u662f\u5185\u5b58\u7ba1\u7406\u7684\u95ee\u9898<\/p>\n

              \u6ce8\u610f\u8fd9\u6837\u4e00\u4e2a\u95ee\u9898\uff1a\u521a\u624d\u7684\u7a0b\u5e8f\u5b8c\u5168\u662f\u628a\u7535\u8111\u91cc\u9762\u7684\u4e3bram\u6487\u7ed9gpu\u8dd1\uff0c\u8fd9\u4e0d\u662f\u626f\u6de1\u561b\uff01<\/p>\n

              \u5176\u5b9ecuda\u7684gpu\u5185\u5b58\u5206\u4e3a\u51e0\u79cd\uff0c\u5168\u5c40\u5185\u5b58\uff0c\u5171\u4eab\u5185\u5b58\uff0c\u7ebf\u7a0b\u72ec\u5360\u5185\u5b58\uff0c\u4f46\u662f\u65e0\u8bba\u5982\u4f55\u3002\u3002\u3002host\u8bbe\u5907\u7684\u5185\u5b58\u5f53\u7136\u662f\u4e0d\u80fd\u8dd1\u7684,<\/p>\n

              GPU\u7ebf\u6027\u5b58\u50a8\u5668\u4f7f\u7528\u00a0cudaMalloc()\u5206\u914d,\u901a\u8fc7\u00a0cudaFree()\u91ca\u653e,\u4f7f\u7528\u00a0cudaMemcpy()\u5728\u8bbe\u5907\u548c\u4e3b\u673a\u95f4\u4f20\u8f93<\/p>\n

              –\u6458\u81eaGUIDE book\uff0c\u98ce\u8fb0\u7ffb\u8bd1<\/p><\/blockquote>\n

              \u6240\u4ee5\uff0c\u65e0\u8bba\u5982\u4f55\uff0c\u5148\u628a\u6570\u636e\u8dd1\u5230GPU\u91cc\u9762\u5427\uff01<\/p>\n

              \u4e8e\u662f\u4fee\u6539\u4ee3\u7801\u5982\u4e0b<\/p>\n

                \n
              1. __global__\u00a0<\/span>void<\/span>\u00a0<\/span>VecAdd<\/span>(<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
              2. {<\/span><\/li>\n
              3. \u00a0\u00a0\u00a0\u00a0<\/span>int<\/span>\u00a0i\u00a0<\/span>=<\/span>\u00a0threadIdx<\/span>.<\/span>x<\/span>;<\/span><\/li>\n
              4. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0j<\/span>=<\/span>0<\/span>;<\/span>j<\/span><<\/span>1000<\/span>;<\/span>j<\/span>++)<\/span><\/li>\n
              5. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0<\/span>(<\/span>A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>*<\/span>\u00a0B<\/span>[<\/span>i<\/span>]);<\/span><\/li>\n
              6. }<\/span><\/li>\n
              7. \u00a0<\/span><\/li>\n
              8. void<\/span>\u00a0cpu_VecAdd<\/span>(<\/span>int<\/span>\u00a0i<\/span>,<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
              9. {<\/span><\/li>\n
              10. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0j<\/span>=<\/span>0<\/span>;<\/span>j<\/span><<\/span>1000<\/span>;<\/span>j<\/span>++)<\/span><\/li>\n
              11. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0<\/span>(<\/span>A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>*<\/span>\u00a0B<\/span>[<\/span>i<\/span>]);<\/span><\/li>\n
              12. }<\/span><\/li>\n
              13. int<\/span>\u00a0main<\/span>()<\/span><\/li>\n
              14. {<\/span><\/li>\n
              15. \u00a0\u00a0\u00a0\u00a0<\/span>\/\/\u00a0Kernel\u00a0invocation\u00a0with\u00a0N\u00a0threads<\/span><\/li>\n
              16. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“Hello,World\\n”<\/span>);<\/span><\/li>\n
              17. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>\u00a0<\/span>*<\/span>A<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>],*<\/span>B<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>],*<\/span>C<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>];<\/span><\/li>\n
              18. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
              19. \u00a0\u00a0\u00a0\u00a0<\/span>{<\/span><\/li>\n
              20. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0A<\/span>[<\/span>i<\/span>]=<\/span>i<\/span>;<\/span><\/li>\n
              21. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0B<\/span>[<\/span>i<\/span>]=<\/span>2<\/span>*<\/span>i<\/span>;<\/span><\/li>\n
              22. \u00a0\u00a0\u00a0\u00a0<\/span>}<\/span><\/li>\n
              23. \u00a0<\/span><\/li>\n
              24. \u00a0\u00a0\u00a0\u00a0<\/span>size_t<\/span>\u00a0size\u00a0<\/span>=<\/span>\u00a0N\u00a0<\/span>*<\/span>\u00a0<\/span>sizeof<\/span>(<\/span>float<\/span>);<\/span><\/li>\n
              25. \u00a0<\/span><\/li>\n
              26. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>*<\/span>\u00a0d_A<\/span>;<\/span><\/li>\n
              27. \u00a0\u00a0\u00a0\u00a0cudaMalloc<\/span>(><\/span>d_A<\/span>,<\/span>\u00a0size<\/span>);<\/span><\/li>\n
              28. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>*<\/span>\u00a0d_B<\/span>;<\/span><\/li>\n
              29. \u00a0\u00a0\u00a0\u00a0cudaMalloc<\/span>(><\/span>d_B<\/span>,<\/span>\u00a0size<\/span>);<\/span><\/li>\n
              30. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>*<\/span>\u00a0d_C<\/span>;<\/span><\/li>\n
              31. \u00a0\u00a0\u00a0\u00a0cudaMalloc<\/span>(><\/span>d_C<\/span>,<\/span>\u00a0size<\/span>);<\/span><\/li>\n
              32. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>\u00a0<\/span>*<\/span>e<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>];<\/span><\/li>\n
              33. \u00a0\u00a0\u00a0\u00a0<\/span>long<\/span>\u00a0st<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
              34. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>d_A<\/span>,<\/span>\u00a0A<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyHostToDevice<\/span>);<\/span><\/li>\n
              35. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>d_B<\/span>,<\/span>\u00a0B<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyHostToDevice<\/span>);<\/span><\/li>\n
              36. \u00a0\u00a0\u00a0\u00a0<\/span>VecAdd<\/span><<<<\/span>1<\/span>,<\/span>\u00a0N<\/span>>>>(<\/span>d_A<\/span>,<\/span>\u00a0d_B<\/span>,<\/span>\u00a0d_C<\/span>);<\/span><\/li>\n
              37. \u00a0\u00a0\u00a0\u00a0<\/span>long<\/span>\u00a0ed<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
              38. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“gpu\u00a0running\u00a0time:%ld\\n”<\/span>,<\/span>ed<\/span>–<\/span>st<\/span>);<\/span><\/li>\n
              39. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>C<\/span>,<\/span>\u00a0d_C<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyDeviceToHost<\/span>);<\/span><\/li>\n
              40. \u00a0\u00a0\u00a0\u00a0st<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
              41. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
              42. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0cpu_VecAdd<\/span>(<\/span>i<\/span>,<\/span>A<\/span>,<\/span>B<\/span>,<\/span>e<\/span>);<\/span><\/li>\n
              43. \u00a0\u00a0\u00a0\u00a0ed<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
              44. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“cpu\u00a0running\u00a0time:%ld\\n”<\/span>,<\/span>ed<\/span>–<\/span>st<\/span>);<\/span><\/li>\n
              45. \u00a0\u00a0\u00a0\u00a0cudaFree<\/span>(<\/span>d_A<\/span>);<\/span><\/li>\n
              46. \u00a0\u00a0\u00a0\u00a0cudaFree<\/span>(<\/span>d_B<\/span>);<\/span><\/li>\n
              47. \u00a0\u00a0\u00a0\u00a0cudaFree<\/span>(<\/span>d_C<\/span>);<\/span><\/li>\n
              48. \u00a0<\/span><\/li>\n
              49. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
              50. \u00a0\u00a0\u00a0\u00a0<\/span>{<\/span><\/li>\n
              51. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/span>\/\/printf(“%f\u00a0“,C[i]);<\/span><\/li>\n
              52. \u00a0\u00a0\u00a0\u00a0<\/span>}<\/span><\/li>\n
              53. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“\\n”<\/span>);<\/span><\/li>\n
              54. }<\/span><\/li>\n<\/ol>\n

                \u4e8e\u662f<\/p>\n

                  \n
                1. make<\/span><\/li>\n
                2. .\/<\/span>test<\/span><\/li>\n
                3. gpu\u00a0running\u00a0time<\/span>:<\/span>102<\/span><\/li>\n
                4. cpu\u00a0running\u00a0time<\/span>:<\/span>2263<\/span><\/li>\n
                5. 0.000000<\/span>\u00a0<\/span>2.000000<\/span>\u00a0<\/span>8.000000<\/span><\/li>\n<\/ol>\n

                  \u54c7\uff01\uff01GPU\u597d\u5feb\u554a<\/p>\n

                  \u4e8e\u662f\u628a\u7b2c\u5341\u884c\u7684j\u8303\u56f4\u589e\u52a0\u5341\u500d<\/p>\n

                    \n
                  1. CUDA$\u00a0<\/span>.\/<\/span>test<\/span><\/li>\n
                  2. Hello<\/span>,<\/span>World<\/span><\/li>\n
                  3. gpu\u00a0running\u00a0time<\/span>:<\/span>59<\/span><\/li>\n
                  4. cpu\u00a0running\u00a0time<\/span>:<\/span>2121<\/span><\/li>\n
                  5. 0.000000<\/span>\u00a0<\/span>2.000000<\/span>\u00a0<\/span>8.000000<\/span>\u00a0<\/span>.<\/span><\/li>\n<\/ol>\n

                    .<\/p>\n

                    \"u=3674647973,796439839&fm=21&gp=0\"<\/a><\/p>\n

                    \u662f\u4e0d\u662f\u6beb\u4e0d\u79d1\u5b66\uff1f<\/p>\n

                    \u5176\u5b9e\u662f\u56e0\u4e3a\uff0c\u5bf9\u4e8e\u5185\u6838\u542f\u52a8\u8fd9\u4ef6\u4e8b\u60c5\uff0c\u5b9e\u9645\u4e0a\u662f\u5f02\u6b65\u7684\uff0c\u5373\u6211\u4eec\u542f\u52a8\u5185\u6838\u4ee5\u540e\u63a7\u5236\u6743\u5c31\u4ea4\u7ed9\u4e86host\uff0c\u6240\u4ee5\uff0c\u6211\u4eec\u4f3c\u4e4e\u9700\u8981\u4e00\u4e2a\u51fd\u6570\u6765\u7b49\u5f85\u4e0b\u5185\u6838\u7684\u6267\u884c\u3002<\/p>\n

                    \u5176\u5b9e\u5012\u4e0d\u9700\u8981\u7279\u522b\u4fee\u6539\uff0c\u56e0\u4e3acudaMemcpy()\u5c31\u662f\u4e00\u4e2a\u540c\u6b65\u51fd\u6570<\/p>\n

                    \u4fee\u6539\u8fd0\u884c\u4ee3\u7801\u5982\u4e0b\uff1a<\/p>\n

                      \n
                    1. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>d_A<\/span>,<\/span>\u00a0A<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyHostToDevice<\/span>);<\/span><\/li>\n
                    2. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>d_B<\/span>,<\/span>\u00a0B<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyHostToDevice<\/span>);<\/span><\/li>\n
                    3. \u00a0\u00a0\u00a0\u00a0<\/span>VecAdd<\/span><<<<\/span>1<\/span>,<\/span>\u00a0N<\/span>>>>(<\/span>d_A<\/span>,<\/span>\u00a0d_B<\/span>,<\/span>\u00a0d_C<\/span>);<\/span><\/li>\n
                    4. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>C<\/span>,<\/span>\u00a0d_C<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyDeviceToHost<\/span>);<\/span><\/li>\n
                    5. \u00a0\u00a0\u00a0\u00a0<\/span>long<\/span>\u00a0ed<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
                    6. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“gpu\u00a0running\u00a0time:%ld\\n”<\/span>,<\/span>ed<\/span>–<\/span>st<\/span>);<\/span><\/li>\n<\/ol>\n

                      \u6267\u884c\uff0c\u5c06\u90a3\u4e2a\u6beb\u65e0\u610f\u4e49\u7684j\u5faa\u73af\u8bbe\u5230\u540c\u7b49\u5927\u5c0f<\/p>\n

                        \n
                      1. Hello<\/span>,<\/span>World<\/span><\/li>\n
                      2. gpu\u00a0running\u00a0time<\/span>:<\/span>515<\/span><\/li>\n
                      3. cpu\u00a0running\u00a0time<\/span>:<\/span>2674<\/span><\/li>\n<\/ol>\n

                        \u53c8\u6216\u8005\u63d0\u9ad8j\u5faa\u73af\u6b21\u6570\u5206\u522b\u63d0\u9ad8\u5341\u500d<\/p>\n

                          \n
                        1. CUDA$\u00a0<\/span>.\/<\/span>test<\/span><\/li>\n
                        2. Hello<\/span>,<\/span>World<\/span><\/li>\n
                        3. gpu\u00a0running\u00a0time<\/span>:<\/span>4064<\/span><\/li>\n
                        4. cpu\u00a0running\u00a0time<\/span>:<\/span>24683<\/span><\/li>\n<\/ol>\n

                          \u8fd9\u5c31\u6bd4\u8f83\u79d1\u5b66\u5566~<\/p>\n

                          \u7136\u800c\u9700\u8981\u6ce8\u610f\u7684\u662f\u5185\u6838\u542f\u52a8\u548c\u6570\u636e\u62f7\u8d1d\u5e76\u975e\u4e0d\u9700\u8981\u65f6\u95f4\uff0c\u6bd4\u5982\u5982\u679c\u4fee\u6539add\u4e3a\u5982\u4e0b\u4ee3\u7801<\/p>\n

                            \n
                          1. __global__\u00a0<\/span>void<\/span>\u00a0<\/span>VecAdd<\/span>(<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
                          2. {<\/span><\/li>\n
                          3. \u00a0\u00a0\u00a0\u00a0<\/span>int<\/span>\u00a0i\u00a0<\/span>=<\/span>\u00a0threadIdx<\/span>.<\/span>x<\/span>;<\/span><\/li>\n
                          4. \u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0<\/span>(<\/span>A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>*<\/span>\u00a0B<\/span>[<\/span>i<\/span>]);<\/span><\/li>\n
                          5. }<\/span><\/li>\n
                          6. void<\/span>\u00a0cpu_VecAdd<\/span>(<\/span>int<\/span>\u00a0i<\/span>,<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
                          7. {<\/span><\/li>\n
                          8. \u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0<\/span>(<\/span>A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>*<\/span>\u00a0B<\/span>[<\/span>i<\/span>]);<\/span><\/li>\n
                          9. }<\/span><\/li>\n<\/ol>\n

                            \u53cd\u5012\u662fGPU\u6162\u4e86<\/p>\n

                            \u4e8e\u662f\uff0c\u6211\u4eec\u5df2\u7ecf\u5bf9CUDA\u6709\u4e86\u521d\u6b65\u7684\u5c1d\u8bd5<\/p>\n

                            \u8fd9\u4e2aHelloworld\u5b8c\u6574\u4ee3\u7801\u5982\u4e0b<\/p>\n

                              \n
                            1. #include<\/span>“stdio.h”<\/span><\/li>\n
                            2. #include<\/span><cuda_runtime.h><\/span><\/li>\n
                            3. #include<\/span>\u00a0<\/span><sys\/time.h><\/span><\/li>\n
                            4. \u00a0<\/span><\/li>\n
                            5. #define<\/span>\u00a0N\u00a0<\/span>1000<\/span><\/li>\n
                            6. \/\/\u00a0Kernel\u00a0definition<\/span><\/li>\n
                            7. __global__\u00a0<\/span>void<\/span>\u00a0<\/span>VecAdd<\/span>(<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
                            8. {<\/span><\/li>\n
                            9. \u00a0\u00a0\u00a0\u00a0<\/span>int<\/span>\u00a0i\u00a0<\/span>=<\/span>\u00a0threadIdx<\/span>.<\/span>x<\/span>;<\/span><\/li>\n
                            10. \u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0<\/span>(<\/span>A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>*<\/span>\u00a0B<\/span>[<\/span>i<\/span>]);<\/span><\/li>\n
                            11. }<\/span><\/li>\n
                            12. \u00a0<\/span><\/li>\n
                            13. long<\/span>\u00a0getCurrentTime<\/span>()<\/span><\/li>\n
                            14. {<\/span><\/li>\n
                            15. \u00a0\u00a0\u00a0<\/span>struct<\/span>\u00a0timeval\u00a0tv<\/span>;<\/span><\/li>\n
                            16. \u00a0\u00a0\u00a0gettimeofday<\/span>(><\/span>tv<\/span>,<\/span>NULL<\/span>);<\/span><\/li>\n
                            17. \u00a0\u00a0\u00a0<\/span>return<\/span>\u00a0tv<\/span>.<\/span>tv_sec\u00a0<\/span>*<\/span>\u00a0<\/span>1000000<\/span>\u00a0<\/span>+<\/span>\u00a0tv<\/span>.<\/span>tv_usec<\/span>;<\/span><\/li>\n
                            18. }<\/span><\/li>\n
                            19. void<\/span>\u00a0cpu_VecAdd<\/span>(<\/span>int<\/span>\u00a0i<\/span>,<\/span>float<\/span>*<\/span>\u00a0A<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0B<\/span>,<\/span>\u00a0<\/span>float<\/span>*<\/span>\u00a0C<\/span>)<\/span><\/li>\n
                            20. {<\/span><\/li>\n
                            21. \u00a0\u00a0\u00a0\u00a0C<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>=<\/span>\u00a0<\/span>(<\/span>A<\/span>[<\/span>i<\/span>]<\/span>\u00a0<\/span>*<\/span>\u00a0B<\/span>[<\/span>i<\/span>]);<\/span><\/li>\n
                            22. }<\/span><\/li>\n
                            23. int<\/span>\u00a0main<\/span>()<\/span><\/li>\n
                            24. {<\/span><\/li>\n
                            25. \u00a0\u00a0\u00a0\u00a0<\/span>\/\/\u00a0Kernel\u00a0invocation\u00a0with\u00a0N\u00a0threads<\/span><\/li>\n
                            26. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“Hello,World\\n”<\/span>);<\/span><\/li>\n
                            27. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>\u00a0<\/span>*<\/span>A<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>],*<\/span>B<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>],*<\/span>C<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>];<\/span><\/li>\n
                            28. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
                            29. \u00a0\u00a0\u00a0\u00a0<\/span>{<\/span><\/li>\n
                            30. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0A<\/span>[<\/span>i<\/span>]=<\/span>i<\/span>;<\/span><\/li>\n
                            31. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0B<\/span>[<\/span>i<\/span>]=<\/span>2<\/span>*<\/span>i<\/span>;<\/span><\/li>\n
                            32. \u00a0\u00a0\u00a0\u00a0<\/span>}<\/span><\/li>\n
                            33. \u00a0<\/span><\/li>\n
                            34. \u00a0\u00a0\u00a0\u00a0<\/span>size_t<\/span>\u00a0size\u00a0<\/span>=<\/span>\u00a0N\u00a0<\/span>*<\/span>\u00a0<\/span>sizeof<\/span>(<\/span>float<\/span>);<\/span><\/li>\n
                            35. \u00a0<\/span><\/li>\n
                            36. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>*<\/span>\u00a0d_A<\/span>;<\/span><\/li>\n
                            37. \u00a0\u00a0\u00a0\u00a0cudaMalloc<\/span>(><\/span>d_A<\/span>,<\/span>\u00a0size<\/span>);<\/span><\/li>\n
                            38. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>*<\/span>\u00a0d_B<\/span>;<\/span><\/li>\n
                            39. \u00a0\u00a0\u00a0\u00a0cudaMalloc<\/span>(><\/span>d_B<\/span>,<\/span>\u00a0size<\/span>);<\/span><\/li>\n
                            40. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>*<\/span>\u00a0d_C<\/span>;<\/span><\/li>\n
                            41. \u00a0\u00a0\u00a0\u00a0cudaMalloc<\/span>(><\/span>d_C<\/span>,<\/span>\u00a0size<\/span>);<\/span><\/li>\n
                            42. \u00a0\u00a0\u00a0\u00a0<\/span>float<\/span>\u00a0<\/span>*<\/span>e<\/span>=<\/span>new<\/span>\u00a0<\/span>float<\/span>[<\/span>N<\/span>];<\/span><\/li>\n
                            43. \u00a0\u00a0\u00a0\u00a0<\/span>long<\/span>\u00a0st<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
                            44. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>d_A<\/span>,<\/span>\u00a0A<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyHostToDevice<\/span>);<\/span><\/li>\n
                            45. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>d_B<\/span>,<\/span>\u00a0B<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyHostToDevice<\/span>);<\/span><\/li>\n
                            46. \u00a0\u00a0\u00a0\u00a0<\/span>VecAdd<\/span><<<<\/span>1<\/span>,<\/span>\u00a0N<\/span>>>>(<\/span>d_A<\/span>,<\/span>\u00a0d_B<\/span>,<\/span>\u00a0d_C<\/span>);<\/span><\/li>\n
                            47. \u00a0\u00a0\u00a0\u00a0cudaMemcpy<\/span>(<\/span>C<\/span>,<\/span>\u00a0d_C<\/span>,<\/span>\u00a0size<\/span>,<\/span>\u00a0cudaMemcpyDeviceToHost<\/span>);<\/span><\/li>\n
                            48. \u00a0\u00a0\u00a0\u00a0<\/span>long<\/span>\u00a0ed<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
                            49. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“gpu\u00a0running\u00a0time:%ld\\n”<\/span>,<\/span>ed<\/span>–<\/span>st<\/span>);<\/span><\/li>\n
                            50. \u00a0\u00a0\u00a0\u00a0st<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
                            51. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
                            52. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0cpu_VecAdd<\/span>(<\/span>i<\/span>,<\/span>A<\/span>,<\/span>B<\/span>,<\/span>e<\/span>);<\/span><\/li>\n
                            53. \u00a0\u00a0\u00a0\u00a0ed<\/span>=<\/span>getCurrentTime<\/span>();<\/span><\/li>\n
                            54. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“cpu\u00a0running\u00a0time:%ld\\n”<\/span>,<\/span>ed<\/span>–<\/span>st<\/span>);<\/span><\/li>\n
                            55. \u00a0\u00a0\u00a0\u00a0cudaFree<\/span>(<\/span>d_A<\/span>);<\/span><\/li>\n
                            56. \u00a0\u00a0\u00a0\u00a0cudaFree<\/span>(<\/span>d_B<\/span>);<\/span><\/li>\n
                            57. \u00a0\u00a0\u00a0\u00a0cudaFree<\/span>(<\/span>d_C<\/span>);<\/span><\/li>\n
                            58. \u00a0<\/span><\/li>\n
                            59. \u00a0\u00a0\u00a0\u00a0<\/span>for<\/span>(<\/span>int<\/span>\u00a0i<\/span>=<\/span>0<\/span>;<\/span>i<\/span><<\/span>N<\/span>;<\/span>i<\/span>++)<\/span><\/li>\n
                            60. \u00a0\u00a0\u00a0\u00a0<\/span>{<\/span><\/li>\n
                            61. \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/span>\/\/printf(“%f\u00a0“,C[i]);<\/span><\/li>\n
                            62. \u00a0\u00a0\u00a0\u00a0<\/span>}<\/span><\/li>\n
                            63. \u00a0\u00a0\u00a0\u00a0printf<\/span>(<\/span>“\\n”<\/span>);<\/span><\/li>\n
                            64. }<\/span><\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"

                              \u5b66\u4e60\u4e00\u4e2a\u4e1c\u897f\u603b\u8981\u8bb0\u4e0b\u6765\uff0c\u6700\u8fd1\u8054\u7cfb\u4e86\u5b9e\u9a8c\u5ba4\u51c6\u5907\u505a\u6709\u5173\u7b49\u4f53\u7684\u8ba1\u7b97\uff0c\u6d69\u6d69\u7684\u7269\u7406\u7cfb\u975e\u8d44\u6df1\u7801\u519c\u751f\u6daf\u4e5f\u5c06\u8fdb\u5165\u6b63\u8f68\u3002\u4e07\u4e8b\u5f00\u5934 … <\/p>\n

                              \u7ee7\u7eed\u9605\u8bfb\u201cCUDA\u5b66\u4e60\u4e0e\u5c1d\u8bd5:Hello,World\u201d<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":[],"categories":[7],"tags":[],"_links":{"self":[{"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=\/wp\/v2\/posts\/177"}],"collection":[{"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=177"}],"version-history":[{"count":6,"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=\/wp\/v2\/posts\/177\/revisions"}],"predecessor-version":[{"id":187,"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=\/wp\/v2\/posts\/177\/revisions\/187"}],"wp:attachment":[{"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=177"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=177"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/blog.xuhao1.me\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=177"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}