可以参考 http://stackoverflow.com/questions/5729964/why-arm-neon-not-faster-than-plain-c http://stackoverflow.com/questions/6034966/neon-asm-code-running-much-slower-than-c-code On Fri, May 23, 2014 at 2:32 PM, xunxun <xunxun1982@xxxxxxxxx> wrote: > 优化的效果很难说,这个大多是和你的平台或者硬件架构有关,比如导致的cache miss之类的 > > > On Sun, May 4, 2014 at 5:50 PM, Shuduo Sang <sangshuduo@xxxxxxxxx> wrote: > >> >> http://www.learnopengles.com/a-performance-comparison-between-java-and-c-on-the-nexus-5/?utm_source=rss&utm_medium=rss&utm_campaign=a-performance-comparison-between-java-and-c-on-the-nexus-5 >> >> 为什么打开(LOCAL_ARM_NEON := true) -ffast-math -O3编译选项的gcc >> 4.8比不打开的时候还差一点(1.02 vs 1.00) ? >> >> Thanks, >> Shuduo >> >> > > > -- > Best Regards, > xunxun > -- Best Regards, xunxun