QUOTE(SebastianG @ Aug 14 2005, 03:39 PM)
The amiga has 4 seperate "hardware channels" which do the resampling and mixing is done in hardware IIRC.
In standard 4 channel modules with 8bit samples there wasn't really such a thing as resampling or mixing.
Paula chip (now nearly 25 years old) which among other things was responsible for Amiga sound, as you said had 4 separate hardware channels (2 channels per speaker), you could feed a channel with 1 sample at a time, it wasn't possible to mix more with hardware.
As for resampling, the chip wasn't bound to a given frequency like it is today, you could play any sample at any sample-rate (different for every of the 4 channels) without any resampling nor interpolation. Funny thing is that the highest possible sample-rate was a function of horizontal frequency of video-mode, in PAL you could play samples at sample-rates up to ~28KHz (in NTSC a bit more), in a so called DblPAL/NTSC (modes with no interlacing in high video resolutions - for VGA & MultiSync monitors) it was ~56KHz, and there was a whole bunch more of video-modes. Funny machines was those Amigas.
To have more than 4 channels & 8bits you had to do software mixing & resampling. Since you had 2 channels per speaker you could do a trick with playing 2 slightly adjusted 8bit samples to gain ~14bit output. To have more than 4 channels you had to mix them into these 4 channels, but all the processing for these tricks had to be done by the CPU.
I remember making 8 channel modules with 14bit output on 68020/14MHz based Amiga. Later when I got 68040/50MHz I could throw in more channels and some DSP, and still there was a plenty of spare CPU time.
I know that playing MODs on PCs or portables is different (you always need to mix & resample), but it's still nothing in comparison to MP3 decoding.
BTW, when you play a standard 4-channel Amiga MOD, turn off any interpolation - then it sounds like it should. Interpolation should be used only with MODs with more than 4 channels and the ones coming from PC trackers.