Kyma Forum
  Kyma Support
  Spectrum Gallery Aggregate Sounds

Post New Topic  Post A Reply
profile | register | preferences | faq | search

next newest topic | next oldest topic
Author Topic:   Spectrum Gallery Aggregate Sounds
Denis Goekdag
Member
posted 03 October 2009 05:25         Edit/Delete Message   Reply w/Quote
I'm currently working with a stereo piano spectrum at 96kHz on my 8 DSP Capy. Both Aggregate Synthesis type sounds in the Spectrum Gallery run out of realtime. What's odd about that is that they seem to grab just DSP 1, overloading it, while not even touching the other 7. As the Sounds are stereo (means two synthesis branches), this seems odd....at the very least I'd expect to see each synthesis branch grab its own DS. I've reduced the number of tracks to 64 and tried changing the bank size to a small number to force the grains etc to be rendered on different DSPs but that changes nothing. Any ideas?

IP: Logged

Denis Goekdag
Member
posted 03 October 2009 05:30         Edit/Delete Message   Reply w/Quote
Well, it seems that the aggregate sounds just aren't made for such a "lowly" box as mine....setting tracks to 4 and bank size to default cured the issue. I do think that the gallery might not want to have number of tracks at the value it has now (156).... unless of course the wicked plan of world domination is to frustrate me into buying a rack full of Pacaranas ;-)

IP: Logged

tuscland
Member
posted 03 October 2009 07:48         Edit/Delete Message   Reply w/Quote
Hello Denis,

Maybe you should try to feed your spectral analysis with two different copies of the sample memory. I think using a single source of memory can be a concurrency contention for shared DSP access.

Best,
Cam

IP: Logged

SSC
Administrator
posted 03 October 2009 09:50         Edit/Delete Message   Reply w/Quote
OK, I will look at making the Gallery examples depend on the processor you are using. In the meantime, you might consider analyzing a mono-mix to produce the CloudBank(?) And reducing the number of grains per track.

The CloudBank is pretty juicy (consider that it has a GrainCloud on each track). And a stereo CloudBank is double (possibly a little more than double due to the extra processor-crossing overhead)

On the Capybara-320 you can get about 28 simultaneous grains per processor, so you could use a rule of thumb calculation:

Tracks * GrainsPerTrack / 28 ~= processors

IP: Logged

All times are CT (US)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply

Contact Us | Symbolic Sound Home

This forum is provided solely for the support and edification of the customers of Symbolic Sound Corporation.


Ultimate Bulletin Board 5.45c