Page 2 of 2

Re: Float Anomaly

PostPosted: Sat Feb 04, 2023 6:33 pm
by Tepeix
No problem, and thanks for the advice !!)

I was thinking that you probably know the precision scaling think, but it's surprising when we come to this number and precision become less than a int.
Not a common situation or number !)

Memory thing, trying to get 10 second even if the user is at 192000 sample rate.
And with asm we then need to have (2^21 * 16)-1.
(this is for using a counter that increment at 16 then with andps with this number the magic happen and the counter reset to 0.) (only way is then to declare as an int, or make some calculation based on good float/int translation..)

I will recall the advice and try to not mix those type !
That's sometime a bad obsession that we need to avoid when getting experience at flowstone !
Even if i know it in some way when i was using synthmaker, when i come back to flowstone after a long pause, i could not resist to play with tick and blue stream !) !) lol ! And Spock telled me that i go to the wrong/dark path !)
Héhé the fantasm of mixing different types in flowstone could be very dangerous !) !)

Will go back maybe soon with another topic and (maybe dangerous) experiment/hack about tick and delay/dethread...

Re: Float Anomaly

PostPosted: Sun Feb 05, 2023 10:12 am
by Spogg
I may be wasting your time with this, but I get the idea that you might find the Double Precision prims useful to deal with the big numbers for higher resolution. This higher resolution is needed to play longer wavs without getting pitch errors and artefacts at longer times.
I’ve attached a wav player (not by me) with a 20 second clip loaded and if you look inside you’ll see the Precise Counter module.

Apologies if you’re already aware of this!

Re: Float Anomaly

PostPosted: Sun Feb 05, 2023 4:39 pm
by Tepeix
Thanks !
I was beginning to think about double, but i never used it, so it's cool to get an example !)