Hi, On Tue, Jun 28, 2011 at 3:03 PM, Luca Barbato <[email protected]> wrote: > On 6/28/11 3:27 PM, Ronald S. Bultje wrote: >> On Tue, Jun 28, 2011 at 12:36 PM, Diego Biurrun<[email protected]> wrote: >>> >>> + * Really, the only reason this type of scaling (fast bilinear) >>> exists >>> + * is because it allowed for certain types of MMX/MMX2 >>> optimizations. >>> + * In most practical cases, there is no reason to use this type of >>> + * scaling. >> >> I think we should remove this part, it's not very positive and doesn't >> add anything... > > What about something along the lines of: > > "Fast bilinear scaling works best on MMX, more advanced SIMD allows > optimization of better scalers"
I don't think that's true. Plus, I hate to reference specific optimizations / implementation details in api docs. Ronald _______________________________________________ libav-devel mailing list [email protected] https://lists.libav.org/mailman/listinfo/libav-devel
