I find some of this info quite misleading, to say that you have 1.7 million numbers (or parameters or whatever) per second is at best case very misleading.... a bit is not a number, nor has it been or will never be, a number can be 1, 4, 8, 16, 32, 64 or 128 bit, or it can even be in 10 bits, but it's never like this becausze it practlcally must be an even number of bytes or we're just wasting space in the digital processor or cpu or whatever it is that willl process or store the number. Then to suggest that processing a 32 bit number takes longer time than processing a 16 bit number is incorrect, this all depends on the hardware architecture....
Wadia is currently making digital processors that upsamples to 1.4MHz with 32 bit samples, I don't wanna suggest that they would improve performance by going to 24 bit, it all comes at a price tag though....
There's no way that I believe that a 24 bit DAC works better or quicker than a 32 bit, I just believe there will be a price penalty for the latter.
There's a lot of assumptions in this article, some of them may be correct but some of them certainly wrong if I may dare to say so, and it's probably very well written to sell his products.....
Edited by haraldo, 06 May 2012 - 03:16 AM.