[Ffmpeg-devel] [PATCH] SSE counterpart of ff_imdct_calc_3dn2
Thu Aug 24 08:41:58 CEST 2006
On Thu, 24 Aug 2006, Zuxy Meng wrote:
> Better than hardcoded inline asms when:
> 1. gcc knows better about what the code does: for example gcc will
> insert instructions that modify the loop counter and/or calculate
> addresses into intrinsics and helps hide latency. You can only write
> the whole loop in asm (quite rare in practice) to achieve this.
Hence Michael suggested writing the whole loop in asm...
btw, for-loops and asm-loops are equally numerous in the current lavc asm.
More information about the ffmpeg-devel