CUDAï¼Compute Unified Device Architectureï¼ã¯ã¼ãï¼ã¨ã¯ãNVIDIAãéçºã»æä¾ãã¦ãããGPUåãã®æ±ç¨ä¸¦åã³ã³ãã¥ã¼ãã£ã³ã°ãã©ãããã©ã¼ã ï¼ä¸¦åã³ã³ãã¥ã¼ãã£ã³ã°ã¢ã¼ããã¯ãã£ï¼ããã³ããã°ã©ãã³ã°ã¢ãã«ã§ãã[4][5][6]ãå°ç¨ã®C/C++ã³ã³ãã¤ã© (nvcc) ãã©ã¤ãã©ãª (API) ãªã©ãæä¾ããã¦ããããªãNVIDIA製GPUã«ããã¦ã¯ãOpenCL/DirectComputeãªã©ã®é¡ä¼¼APIã³ã¼ã«ã¯ããã¹ã¦å ±éã®GPGPUãã©ãããã©ã¼ã ã§ããCUDAãçµç±ãããã¨ã«ãªã[7]ã CUDAã®å¦çã®æµã 1. ã¡ã¤ã³ã¡ã¢ãªï¼ãã¹ãã¡ã¢ãªï¼ãããã¼ã¿ãGPUç¨ã¡ã¢ãªï¼ããã¤ã¹ã¡ã¢ãªï¼ã«ã³ãã¼ããã 2. CPUãGPUã«å¯¾ãã¦å¦çãæ示ããã 3. GPUãå¿ è¦ãªãã¼ã¿ãåãè¾¼ã¿åã³ã¢ã§ä¸¦åå®è¡ããã 4. çµæãGPUç¨ã¡ã¢ãªã
{{#tags}}- {{label}}
{{/tags}}