Quote:
Originally Posted by Churst
P.S. (the more ranty, less relevant bit)
It seems to me that among the music fraternity there is just general hostility towards the apparently "scary" move to 64 bit. I read a piece in Sound on Sound, not all that long ago, that claimed that 64 bit computing has little or no benefit to audio production on computers. That shocked me somewhat, being as every single other computer based industry on the planet is embracing the phenomenal power enhancements that 64 bit computing will bring.
|
There is truth to what S.O.S says.
1) No current or planned audio codec records or plays at 64bit. We're lucky to get to 32. The more bits you have, the more of a problem denormal numbers become. a 64 bit floating calculation would hit your cpu like a small moon hitting the earth. The memory required is also quite staggering.
2) Bearing in mind the above, the dynamic range of of 32 bit system is in excess of what we can detect (even using fixed integers). Therefore, its pointless making extra hardwork for your computer...when we're not gonna hear the benefit of it. On paper, you'd expect double the performance at 64, but thats not strictly the case.
64 bits is useful for some mixing type applications. Specifically, applications that seek to mimic the response of analogue devices. the extra accuracy is useful in mapping and recreating smaller details that might go under the radar at lower bitdepths (the question is...if they are that small, are they worth picking up). at 64, you got a bit more room to mix mix in, but its minimal gain for a lot of work. If you're good at making and mixing music, you simply don't need to use 64bit mix engines.
Some apps (such as Izoptope, Reaper & Sonar) do run with 64bit mix engines in them. But the proof of the pudding is that none of these are holy grails for audio engineers. The most important thing is how the engineer works...32 bits is more then ample
Now, if you want to talk about 64bit addressing. That makes much more sense. Basically moving data in blocks of 64 instead of 32 will obviously move data twice as fast (per clock cycle, anyway)..providing your data bus supports 64 parallel bits, and your cpu can also do the same. This is why operating systems and computers are moving to 64 bits, but audio production isn't. Its useful for faster computers, but pointless for better audio.