iOSã®BLAS(Accelerate.framework)ã¨Eigenã®è¡Œåˆ—ã®æŽ›ã‘ç®—ã®å®Ÿè¡Œé€Ÿåº¦ãƒ†ã‚¹ãƒˆ - 開発ã®ãƒ’ホçµè«–：BLASã®æ–¹ãŒé€Ÿã„。 ã“ã®å‰ã“ã‚“ãªè¨˜äº‹ã‚’書ãã¾ã—ãŸã€‚ iOSã®vDSP(Accelerate.framework)ã¨Eigenã®è¡Œåˆ—ã®æŽ›ã‘ç®—ã®å®Ÿè¡Œé€Ÿåº¦ãƒ†ã‚¹ãƒˆ - 開発ã®ãƒ’ホ ã§ã€ã“ã‚“ãªã‚³ãƒ¡ãƒ³ãƒˆã‚’è²°ã„ã¾ã—ãŸã€‚ Steve Apologies in advance for not writing in Japanese, but I assume you’ll get the gist: vDSP_mmul is a legacy routine that has never been aggressively tuned for any platform. If you want to perform a matrix multiply using Accelerate (or actually any linear algebra operations), use the BLAS
生活や実務ã«å½¹ç«‹ã¤é«˜ç²¾åº¦è¨ˆç®—サイト
http://www.flashandmath.com/
{{#tags}}- {{label}}
{{/tags}}