truncate or not truncate?

aliasant · 23 Sept 2007

Good question.

I have a similar one. But I would also like to know more of what this Double precision actually does.

I have many guesses but nowere have I found any info on what it actually does.

This could come with a couple of quick examples on when and were to use this and what about cpu consumtion?

Pleas. Some more info would be great.

oddson · 23 Sept 2007

say i have a float component containing the value ..0.03790693583186333....

You don't, you have a float component with the value 0.0379069 as that's the best you can do with a single precision float. So it's already truncated.

I know what you saying though... but yes, it would have to be truncated because all that extra precision is only stored as text in the float display and isn't really there in the float output.

...when and were to use this

It seems to have been implemented solely to fix the index problem with long sample playback.

Normally an index would be stored as an integer but SM uses SSE which uses floats... so to keep track of the index value in a sample playback you need to use a float but a single precision doesn't retain integer accuracy to a large enough value to be usable (the pitch shift problem noted elsewhere). AFAIK this is the only reason of double-floats in SM and the number of components is too limited to do much else with them.

exonerate · 23 Sept 2007

aliasant wrote: Good question.

I have a similar one. But I would also like to know more of what this Double precision actually does.

I have many guesses but nowere have I found any info on what it actually does.

This could come with a couple of quick examples on when and were to use this and what about cpu consumtion?

Pleas. Some more info would be great.

The double components where added to solve the issue with the pitch changing in the wave player module on long samples.

Basically you could use them where ever you would use the normal stream mulitpies, addition ect...

Although i wouldn't go just using them anywhere, only in places where they are benifical, like for instance they can be used (and are used in the updated wave player) to increase the accuracy of interpolation.

they are very basic components, but can be used in areas where the extra accuracy is benifical.

cheers 😉

Exo

exonerate · 23 Sept 2007

the reason i asked the question in the first place was because i'm planning on doing interpolation completley with doubles (which is why i got malc to include a double multiply 😉 thanks malc!).

And this is why i asked the original question because i can still do the interpolation using doubles but i have static double floats to multiply the streams with. it seems then that i'll be losing a bit of the accuracy in these areas.

But still though the double components just set up like in the wave player really increase the quality of the interpolation quite a lot because the fractional part is now a lot more accurate.

cheers 😉

aliasant · 24 Sept 2007

Basically you could use them where ever you would use the normal stream mulitpies, addition ect...

Although i wouldn't go just using them anywhere, only in places where they are benifical, like for instance they can be used (and are used in the updated wave player) to increase the accuracy of interpolation.

OK. I was thinking about 2 places were I might get a better result using them.

First place is in a synth Im working on. I use OSC1 to modulate OSC2's amplitude. AM So higher precision might give a better sound? Also in a ringmodulator or any other place were one polyphonic signal directly modifies another.

Or?

Second place is in my Inerthia compressor. hmm. Or maybe not. This one uses mono signals. One monosignal multiplies the dry monosignal to create the compression but I guess I cant use these poly goodies for that. Maybe there is an other way to get higher precision while multiplying monosignals?

exonerate · 24 Sept 2007

OK. I was thinking about 2 places were I might get a better result using them.

First place is in a synth Im working on. I use OSC1 to modulate OSC2's amplitude. AM So higher precision might give a better sound? Also in a ringmodulator or any other place were one polyphonic signal directly modifies another.

Yes you could use them for AM or ring mod, and yes i'd only use them where one stream is affecting another.

Second place is in my Inerthia compressor. hmm. Or maybe not. This one uses mono signals. One monosignal multiplies the dry monosignal to create the compression but I guess I cant use these poly goodies for that. Maybe there is an other way to get higher precision while multiplying monosignals?

Well your in luck because they work with the mono stream as well as mono pack 4, 'stream' in SM means poly or mono like for instance in a code component when you type 'streamin' (for your inputs) you can connect either poly or mono or even mono pack 4.

cheers 😉

Exo

aliasant · 24 Sept 2007

Well your in luck because they work with the mono stream as well as mono pack 4, 'stream' in SM means poly or mono like for instance in a code component when you type 'streamin' (for your inputs) you can connect either poly or mono or even mono pack 4.

'

Yeeeehaaaaaa! =)

exonerate · 24 Sept 2007

Hey i've just posted an interpolation algorithm using the double components here.....HQ 6 point Interpolation using doubles

Check it out if you want to see the benifits of double precision 😉

cheers 😉

Exo

aliasant · 24 Sept 2007

Checking very soon. I just tried to update my bandpass filter and my Parabolic and Hyperbolic saturators.

The filter sounds amazing but my ears are kind of whacky. I have to do more comparing.

Man this filter sounds good.

Gonna check your osm now.

Update: HolySmoke""!!!!!#!€

wow.... stunned