posted 11 January 2002 03:34            

I just had the most amazing thing happen here on my Kyma system...

I had defined a triple of 3 delay lines fed by bandpass filters with a MemoryWriter tapping off the output of the delay and feeding back to a Sample player reading from the memory and feeding back into a mixer with the source sound. The memory buffers were each 4096 samples long. All three channels were then fed to a stereo mixer with different pannings and amplitudes.

When I played this sound on my Capy I had to resort to forced processor assignments and even then it took up nearly all of my numerous DSP's.

But when I created a class from a prototype of one of the 3 filter/delay strips and then used this class 3 times in a reformulated equivalent of the first sound, it used a mere 2.5 DSP's. And I think the sound has a higher quality as well = much brighter and crisper sounding.

Is it generally true that you gain runtime efficiency by grouping sound blocks into a new Class?

I also notice that the Class editor in 5.2 still doubles up many ?args, sometimes with a capitalized and a lowercase version of the same name, and other times with just plain duplicate names. But I found that as long as I make duplicate names act the same way, and occupy the same location in the sound editor panel, then everything seems to work just fine. By default the Class editor wants to place these duplicates in their own subpanels on the sound editor screen.

Cheers,