HI all, I'm looking at the Channel Conversion Header; in particular the "Target Channel (M) x Src Channel (N) Gain array". Why are these numbers some sort of 15.16 fixed point value and not a 32 bit float? Is that because we want to make it easy to implement this on CPUs without floating point units? Erik -- +-----------------------------------------------------------+ Erik de Castro Lopo +-----------------------------------------------------------+ "Linux is produced to be used, whereas the others are produced to be sold" -- Bobby D. Bryant
> I'm looking at the Channel Conversion Header; in particular the "Target > Channel (M) x Src Channel (N) Gain array". Why are these numbers some > sort of 15.16 fixed point value and not a 32 bit float? > > Is that because we want to make it easy to implement this on CPUs without > floating point units?That's right. I also kind of dislikes having float values in headers. Also, I don't think it's very, but I think there are still *some* archs that actually have non-IEEE floats. Just curious, what are the reasons that make you want a float instead? Jean-Marc
Jean-Marc Valin wrote:> That's right. I also kind of dislikes having float values in headers. > Also, I don't think it's very, but I think there are still *some* archs > that actually have non-IEEE floats. Just curious, what are the reasons > that make you want a float instead?I'm cool with the fixed point. I just wanted to make sure it was there for a good reason. The reason you give is fine. Erik -- +-----------------------------------------------------------+ Erik de Castro Lopo +-----------------------------------------------------------+ Failure is not an option. It comes bundled with your Microsoft product.