Because it has 10 wires.
Because it does not mix the 10 audio streams.
Yes it can, actually, but never mind. I suggest you read the technical data sheets for these parts before making such statements.
Look, I love Paula's unique sound more than most and still compose for it. However, the stuff you come out with at times is just mindbogglingly messed up.
Because it does not support any kind of useable sampling rates.
In the audio industry, 11025, 22050, 24000, 32000, 44100, 48000 and so on are all standard frequencies. Any dedicated sampler (back in the days when you bought dedicated samplers) even used these frequencies. Entire professionally captured sample libraries were published in them.
However, since you raise the issue, as a musician and coder I must point out that neither does Paula. Paula doesn't even have the concept of a sampling rate; it has a
sample period which is the inverse function, a physical duration of time measured in clocks between loading the sample registers. Being a 1/x function and x measured as an integer number of clocks, the error in this approximation gets worse for higher playback rates. I've tested this. Given a synthetic looped sinewave of say 64 samples, you can hear the off-pitch playback, particularly for higher notes when comparing to the same notes rendered by software into a buffer.
Also, Paula doesn't mix the four audio outputs it has, either. At least not digitally. The four channels are configured L R R L at the analogue stage and the audio filter happens after that.
Paula would have been far more useful if each of the 4 channels could be panned and if the audio filter could have been toggled per channel. Better still if the filter had more than one cut off frequency that could be selected.
Because nobody writes 10 channel mods.
Because nobody writes mods using 100% 44.1Khz sample speed.
Irrelevant. Who cares? All the Amiga mod based music I've written in the last 15 years have been OctaMED Mix Mode because as a musician it's *vastly* more useful than the old 4 channel mode. 4 channel mode is handy if all you care about is CPU time, but if you want to make music for it's own sake, no musician is going to turn down more channels and more control over them.
Of all the sample speeds used in a random mod, this sound chip supports between 0% and 1% of the required sample rates.
So what? You get a set of industry standard replay rates versus an ad-hoc set of non-standard rates that are the result of dividing your PAL/NTSC clock by some fixed integer value.
It only outputs what the CPU has already mixed itself into 7.1 format or into 2 channel stereo format.
Either way the sound chip does almost nothing. Its just 10 dumb DACs sitting there waiting to be spoonfed data from the CPU.
Er, no. It supports all sorts of hardware DSP for channel expansion, equalisation and many other things besides. These are standard features on all modern sound chips, not just the DACs.
The driver has the responsibility of playing a potentially limitless number of input audio streams through whatever hardware configuration the chip has been programmed for, and in that it has to do whatever mixing is required. However, the chip does far more than basic DAC duty. The incoming streams can be at any of the standard audio rates. They are resampled to the configured output rate using nth-order digital filtering, equalized, companded, maybe even a bit of DSP effects depending on whatever goofy preset I select, and formatted for whatever output configuration is selected.
On all but the most unfeatured chipsets, the CPU just does upstream mixing of samples into input streams. Yes that does use CPU time, but it's absolutely insignificant on a modern CPU.
Its not a Paula or anything of that technology level.
No, it isn't.
