I'm struggling to make a delay effect that doesn't totally drain my
resources. My curent UDO looks like this:

opcode StereoDelay, aa, aaiii
ainL, ainR, idelayTaps, idelayTime, idelayGain xin

idelayTimes[] init idelayTaps
idelayTapsLeft init idelayTaps
idelayTapNum init 0
while (idelayTapsLeft > 0) do
    idelayTimes[idelayTapNum] = (idelayTime / idelayTaps) * (idelayTapNum + 1)
    idelayTapsLeft -= 1
    idelayTapNum += 1
od

adelayL = 0
adelayR = 0
idelayTapsLeft = idelayTaps
idelayTapNum = 0
while (idelayTapsLeft > 0) do
    aL delay ainL, idelayTimes[idelayTapNum]
    aR delay ainR, idelayTimes[idelayTapNum]
    adelayL += aL
    adelayR += aR
    idelayTapsLeft -= 1
    idelayTapNum += 1
od

xout adelayL, adelayR
endop

for some reason, with only 3 or 4 delay taps, I can't seem to go
beyond 1.5 seconds without using near maximum processor resources (on
all 4 cores!) and I get underruns:

    WARNING: Buffer underrun in real-time audio output

I get underuns whether I'm going through jack or alsa, and I have
tried adjusting the buffer size. I also got exactly the same underruns
when using the multiptap opcode.

It's worth mentioning that I have some other processing going on, but
the cores are all peaking at about 20% when I remove the delay lines.

Is there a more CPU efficient way of setting up delays? Or is there
something wrong with my code that's causing unnecessary processing? I
feel like my pc should be able to handle this. I've used much worse
computers than this and had projects in a DAW with bundles of fx
running on multiple instruments at once

Pete

--
http://algorythmradio.com
https://soundcloud.com/algorythmradio

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Your opcode doesn't run for me? I get nothing but silence?

On 13 April 2016 at 12:43, Peter Burgess <pete.soundtechnician@gmail.com> wrote:

I'm struggling to make a delay effect that doesn't totally drain my
resources. My curent UDO looks like this:

opcode StereoDelay, aa, aaiii
ainL, ainR, idelayTaps, idelayTime, idelayGain xin

idelayTimes[] init idelayTaps
idelayTapsLeft init idelayTaps
idelayTapNum init 0
while (idelayTapsLeft > 0) do
    idelayTimes[idelayTapNum] = (idelayTime / idelayTaps) * (idelayTapNum + 1)
    idelayTapsLeft -= 1
    idelayTapNum += 1
od

adelayL = 0
adelayR = 0
idelayTapsLeft = idelayTaps
idelayTapNum = 0
while (idelayTapsLeft > 0) do
    aL delay ainL, idelayTimes[idelayTapNum]
    aR delay ainR, idelayTimes[idelayTapNum]
    adelayL += aL
    adelayR += aR
    idelayTapsLeft -= 1
    idelayTapNum += 1
od

xout adelayL, adelayR
endop

for some reason, with only 3 or 4 delay taps, I can't seem to go
beyond 1.5 seconds without using near maximum processor resources (on
all 4 cores!) and I get underruns:

    WARNING: Buffer underrun in real-time audio output

I get underuns whether I'm going through jack or alsa, and I have
tried adjusting the buffer size. I also got exactly the same underruns
when using the multiptap opcode.

It's worth mentioning that I have some other processing going on, but
the cores are all peaking at about 20% when I remove the delay lines.

Is there a more CPU efficient way of setting up delays? Or is there
something wrong with my code that's causing unnecessary processing? I
feel like my pc should be able to handle this. I've used much worse
computers than this and had projects in a DAW with bundles of fx
running on multiple instruments at once

Pete

--
http://algorythmradio.com
https://soundcloud.com/algorythmradio

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

You could use a single delayr / delayw pair and then create the taps within them using a recursive UDO. I have an instrument that does that to create up to 300 taps, which runs with no problem.

> Date: Wed, 13 Apr 2016 15:58:27 +0100
> From: pete.soundtechnician@GMAIL.COM
> Subject: Re: [Csnd] Delay Effects
> To: CSOUND@LISTSERV.HEANET.IE

>
> arrays of delay tap times and gains, so I can have any number of delay
> taps and I calculate them all on the init pass, and just access them
> during the performance loop.
>
> I'm not displeased with the amount of delay opcodes, but I can't work
> out why my setup is causing so much CPU drag. I've tried a number of
> different setups now, the current one must be completely wrong cos
> it's not producing any sound, haha!
>
> On Wed, Apr 13, 2016 at 3:50 PM, jpff <jpff@codemist.co.uk> wrote:
> > Why are you using arrays here? Are the opcodes
> > delay
> > delay1
> > delayk
> > delayr
> > delayw
> > deltap
> > deltap3
> > deltapi
> > deltapn
> > deltapx
> > deltapxw
> > vdelay
> > vdelay3
> > vdelayx
> > vdelayxs
> > vdelayxq
> > vdelayxw
> > vdelayxwq
> > vdelayxws
> > multitap
> >
> > insufficient?
> >
> > Csound mailing list
> > Csound@listserv.heanet.ie
> > https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> > Send bugs reports to
> > https://github.com/csound/csound/issues
> > Discussions of bugs and features can be posted here
>
>
>
> --
> http://algorythmradio.com
> https://soundcloud.com/algorythmradio
>
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
> https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list Csound@listserv.heanet.ie https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to https://github.com/csound/csound/issues Discussions of bugs and features can be posted here

Hi Steven,

Is there a full list of "stateful opcode" somewhere in the manuals?

or is there a thorough discussion about do's and don'ts in Csound loops?

It will help newbies tremendously, especially who are used to loops in other programming languages like myself.

I had similar problems with the OP when implementing filter banks. 

The general advice I got from someone on the list was that 

it's in general a bad idea to use while loop in Csound. It may not be what he literally said, 

but I think the spirit was that. I'm sure whoever said that while loops are taboos. 

But it sends the message: extreme precautions are needed.

It would be great if there is a recipe for batch calling opcodes, like filter banks.

So far the way I came up with was to generate that sort of code from another language.

Thanks,

Beinan

On Wed, Apr 13, 2016 at 12:28 PM, Steven Yi <stevenyi@gmail.com> wrote:

The original code looks to have a problem of using a stateful opcode
within a loop. I wrote a little bit about this problem a long while
ago in:

http://csoundjournal.com/2006summer/controlFlow_part2.html

(Section "Technical Information" just before the end)

If you're going to use arrays to store parameters for multiple delays,
you may want to use recursion and a UDO to do so.

On Wed, Apr 13, 2016 at 11:35 AM, Rory Walsh <rorywalsh@ear.ie> wrote:
> I have to make a few edits for it to work, mostly changed i variables to k.
> Anyhow, I am not getting any crazy spike in CPU. On my mac mini is runs less
> than 1%.
>
> On 13 April 2016 at 16:29, Rory Walsh <rorywalsh@ear.ie> wrote:
>>
>> Your opcode doesn't run for me? I get nothing but silence?
>>
>> On 13 April 2016 at 12:43, Peter Burgess <pete.soundtechnician@gmail.com>
>> wrote:
>>>
>>> I'm struggling to make a delay effect that doesn't totally drain my
>>> resources. My curent UDO looks like this:
>>>
>>> opcode StereoDelay, aa, aaiii
>>> ainL, ainR, idelayTaps, idelayTime, idelayGain xin
>>>
>>> idelayTimes[] init idelayTaps
>>> idelayTapsLeft init idelayTaps
>>> idelayTapNum init 0
>>> while (idelayTapsLeft > 0) do
>>>     idelayTimes[idelayTapNum] = (idelayTime / idelayTaps) * (idelayTapNum
>>> + 1)
>>>     idelayTapsLeft -= 1
>>>     idelayTapNum += 1
>>> od
>>>
>>> adelayL = 0
>>> adelayR = 0
>>> idelayTapsLeft = idelayTaps
>>> idelayTapNum = 0
>>> while (idelayTapsLeft > 0) do
>>>     aL delay ainL, idelayTimes[idelayTapNum]
>>>     aR delay ainR, idelayTimes[idelayTapNum]
>>>     adelayL += aL
>>>     adelayR += aR
>>>     idelayTapsLeft -= 1
>>>     idelayTapNum += 1
>>> od
>>>
>>> xout adelayL, adelayR
>>> endop
>>>
>>> for some reason, with only 3 or 4 delay taps, I can't seem to go
>>> beyond 1.5 seconds without using near maximum processor resources (on
>>> all 4 cores!) and I get underruns:
>>>
>>>     WARNING: Buffer underrun in real-time audio output
>>>
>>> I get underuns whether I'm going through jack or alsa, and I have
>>> tried adjusting the buffer size. I also got exactly the same underruns
>>> when using the multiptap opcode.
>>>
>>> It's worth mentioning that I have some other processing going on, but
>>> the cores are all peaking at about 20% when I remove the delay lines.
>>>
>>> Is there a more CPU efficient way of setting up delays? Or is there
>>> something wrong with my code that's causing unnecessary processing? I
>>> feel like my pc should be able to handle this. I've used much worse
>>> computers than this and had projects in a DAW with bundles of fx
>>> running on multiple instruments at once
>>>
>>> Pete
>>>
>>> --
>>> http://algorythmradio.com
>>> https://soundcloud.com/algorythmradio
>>>
>>> Csound mailing list
>>> Csound@listserv.heanet.ie
>>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>>> Send bugs reports to
>>>         https://github.com/csound/csound/issues
>>> Discussions of bugs and features can be posted here
>>
>>
>
> Csound mailing list Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to
> https://github.com/csound/csound/issues Discussions of bugs and features can
> be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Hi Beinan,

There is no list of stateful opcodes that I know of.  In general, I
use loops when doing very low-level code that primarily deals with
math operations, such as when implementing a filter or envelope
generator.  Anything higher-level -- such as pre-made filters and
oscillators -- are likely to have state. As mentioned in the posted
article, if you want to do bank-like operations, you should currently
use recursion and UDOs in Csound.

If you are working in other programming languages and used loops as
well as developed a processing network of objects similarly to how
Csound operates, you would run in the same problem.  For example, if
you had classes called Filter and AudioSignal, imagine this:

AudioSignal output;
AudioSignal inputs[16];
Filter f = new Filter();
for (int i =0; i < 16; i++) {
  output += f.process(inputs[i]);
}

You would have a problem as most filters have state. The single filter
instance will depend upon the state of the last run, but then it is
being used with multiple signals, not the previous filtering of the
same signal. If instead you had:

AudioSignal output;
AudioSignal inputs[16];
Filter f[16];
for (int i =0; i < 16; i++) {
  output += f[i].process(inputs[i]);
}

You would be okay.  I think with the Csound language, people think
when using a loop they are doing the latter example, but it's really
more like the first example.

When using a recursive UDO, you are getting a new filter opcode
instance per step of the recursion.  That is is the important part as
to why that works when loops do not.

Loops shouldn't be considered taboo in any way, but they do require
caution.  As mentioned before, I use loops to implement filters and
envelope generators in UDO code. Usually in these cases, one is
calculating on a sample-by-sample basis, and one has to have state
information from the previous sample calculation to generate the next
sample.  These things are often not efficient, easy, nor possible to
calculate using only vector operations (i.e., opcodes that process a
block of samples at a time). In these cases, I think loops are very
appropriate.

Hope that helps!
steven

On Thu, Apr 14, 2016 at 11:42 AM, Beinan Li <li.beinan@gmail.com> wrote:
> Hi Steven,
>
> Is there a full list of "stateful opcode" somewhere in the manuals?
> or is there a thorough discussion about do's and don'ts in Csound loops?
> It will help newbies tremendously, especially who are used to loops in other
> programming languages like myself.
>
> I had similar problems with the OP when implementing filter banks.
> The general advice I got from someone on the list was that
> it's in general a bad idea to use while loop in Csound. It may not be what
> he literally said,
> but I think the spirit was that. I'm sure whoever said that while loops are
> taboos.
> But it sends the message: extreme precautions are needed.
>
> It would be great if there is a recipe for batch calling opcodes, like
> filter banks.
> So far the way I came up with was to generate that sort of code from another
> language.
>
> Thanks,
> Beinan
>
>
> On Wed, Apr 13, 2016 at 12:28 PM, Steven Yi <stevenyi@gmail.com> wrote:
>>
>> The original code looks to have a problem of using a stateful opcode
>> within a loop. I wrote a little bit about this problem a long while
>> ago in:
>>
>> http://csoundjournal.com/2006summer/controlFlow_part2.html
>>
>> (Section "Technical Information" just before the end)
>>
>> If you're going to use arrays to store parameters for multiple delays,
>> you may want to use recursion and a UDO to do so.
>>
>> On Wed, Apr 13, 2016 at 11:35 AM, Rory Walsh <rorywalsh@ear.ie> wrote:
>> > I have to make a few edits for it to work, mostly changed i variables to
>> > k.
>> > Anyhow, I am not getting any crazy spike in CPU. On my mac mini is runs
>> > less
>> > than 1%.
>> >
>> > On 13 April 2016 at 16:29, Rory Walsh <rorywalsh@ear.ie> wrote:
>> >>
>> >> Your opcode doesn't run for me? I get nothing but silence?
>> >>
>> >> On 13 April 2016 at 12:43, Peter Burgess
>> >> <pete.soundtechnician@gmail.com>
>> >> wrote:
>> >>>
>> >>> I'm struggling to make a delay effect that doesn't totally drain my
>> >>> resources. My curent UDO looks like this:
>> >>>
>> >>> opcode StereoDelay, aa, aaiii
>> >>> ainL, ainR, idelayTaps, idelayTime, idelayGain xin
>> >>>
>> >>> idelayTimes[] init idelayTaps
>> >>> idelayTapsLeft init idelayTaps
>> >>> idelayTapNum init 0
>> >>> while (idelayTapsLeft > 0) do
>> >>>     idelayTimes[idelayTapNum] = (idelayTime / idelayTaps) *
>> >>> (idelayTapNum
>> >>> + 1)
>> >>>     idelayTapsLeft -= 1
>> >>>     idelayTapNum += 1
>> >>> od
>> >>>
>> >>> adelayL = 0
>> >>> adelayR = 0
>> >>> idelayTapsLeft = idelayTaps
>> >>> idelayTapNum = 0
>> >>> while (idelayTapsLeft > 0) do
>> >>>     aL delay ainL, idelayTimes[idelayTapNum]
>> >>>     aR delay ainR, idelayTimes[idelayTapNum]
>> >>>     adelayL += aL
>> >>>     adelayR += aR
>> >>>     idelayTapsLeft -= 1
>> >>>     idelayTapNum += 1
>> >>> od
>> >>>
>> >>> xout adelayL, adelayR
>> >>> endop
>> >>>
>> >>> for some reason, with only 3 or 4 delay taps, I can't seem to go
>> >>> beyond 1.5 seconds without using near maximum processor resources (on
>> >>> all 4 cores!) and I get underruns:
>> >>>
>> >>>     WARNING: Buffer underrun in real-time audio output
>> >>>
>> >>> I get underuns whether I'm going through jack or alsa, and I have
>> >>> tried adjusting the buffer size. I also got exactly the same underruns
>> >>> when using the multiptap opcode.
>> >>>
>> >>> It's worth mentioning that I have some other processing going on, but
>> >>> the cores are all peaking at about 20% when I remove the delay lines.
>> >>>
>> >>> Is there a more CPU efficient way of setting up delays? Or is there
>> >>> something wrong with my code that's causing unnecessary processing? I
>> >>> feel like my pc should be able to handle this. I've used much worse
>> >>> computers than this and had projects in a DAW with bundles of fx
>> >>> running on multiple instruments at once
>> >>>
>> >>> Pete
>> >>>
>> >>> --
>> >>> http://algorythmradio.com
>> >>> https://soundcloud.com/algorythmradio
>> >>>
>> >>> Csound mailing list
>> >>> Csound@listserv.heanet.ie
>> >>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>> >>> Send bugs reports to
>> >>>         https://github.com/csound/csound/issues
>> >>> Discussions of bugs and features can be posted here
>> >>
>> >>
>> >
>> > Csound mailing list Csound@listserv.heanet.ie
>> > https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to
>> > https://github.com/csound/csound/issues Discussions of bugs and features
>> > can
>> > be posted here
>>
>> Csound mailing list
>> Csound@listserv.heanet.ie
>> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
>> Send bugs reports to
>>         https://github.com/csound/csound/issues
>> Discussions of bugs and features can be posted here
>
>
> Csound mailing list Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to
> https://github.com/csound/csound/issues Discussions of bugs and features can
> be posted here

Csound mailing list
Csound@listserv.heanet.ie
https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
Send bugs reports to
        https://github.com/csound/csound/issues
Discussions of bugs and features can be posted here

Hi Richard,

It's this one:

http://www.iainmccurdy.org/CsoundRealtimeExamples/Cabbage/Effects/GateReverb.csd

If you activate the filter you might not get to 300. 300 x moogvcf!

bye,

Iain

---

Date: Thu, 14 Apr 2016 11:48:03 -0400
From: rboulanger@BERKLEE.EDU
Subject: Re: [Csnd] Delay Effects
To: CSOUND@LISTSERV.HEANET.IE

Iain,

Which of your instruments is the 300 tap recursive delay?

Would love to check it out and show it off in class.

-dB

On Wed, Apr 13, 2016 at 4:58 PM, Iain McCurdy <i_mccurdy@hotmail.com> wrote:

You could use a single delayr / delayw pair and then create the taps within them using a recursive UDO. I have an instrument that does that to create up to 300 taps, which runs with no problem.

> Date: Wed, 13 Apr 2016 15:58:27 +0100
> From: pete.soundtechnician@GMAIL.COM
> Subject: Re: [Csnd] Delay Effects
> To: CSOUND@LISTSERV.HEANET.IE

>
> arrays of delay tap times and gains, so I can have any number of delay
> taps and I calculate them all on the init pass, and just access them
> during the performance loop.
>
> I'm not displeased with the amount of delay opcodes, but I can't work
> out why my setup is causing so much CPU drag. I've tried a number of
> different setups now, the current one must be completely wrong cos
> it's not producing any sound, haha!
>
> On Wed, Apr 13, 2016 at 3:50 PM, jpff <jpff@codemist.co.uk> wrote:
> > Why are you using arrays here? Are the opcodes
> > delay
> > delay1
> > delayk
> > delayr
> > delayw
> > deltap
> > deltap3
> > deltapi
> > deltapn
> > deltapx
> > deltapxw
> > vdelay
> > vdelay3
> > vdelayx
> > vdelayxs
> > vdelayxq
> > vdelayxw
> > vdelayxwq
> > vdelayxws
> > multitap
> >
> > insufficient?
> >
> > Csound mailing list
> > Csound@listserv.heanet.ie
> > https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> > Send bugs reports to
> > https://github.com/csound/csound/issues
> > Discussions of bugs and features can be posted here
>
>
>
> --
> http://algorythmradio.com
> https://soundcloud.com/algorythmradio
>
> Csound mailing list
> Csound@listserv.heanet.ie
> https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND
> Send bugs reports to
> https://github.com/csound/csound/issues
> Discussions of bugs and features can be posted here

Csound mailing list Csound@listserv.heanet.ie https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to https://github.com/csound/csound/issues Discussions of bugs and features can be posted here

--

_____________________________________________
Dr. Richard Boulanger
Professor of Electronic Production and Design
Professional Writing and Music Technology Division
Berklee College of Music

______________________________________________
President of Boulanger Labs - http://boulangerlabs.com
Author & Editor of The Csound Book - http://mitpress.mit.edu/books/csound-book
Author & Editor of The Audio Programming Book - http://mitpress.mit.edu/books/audio-programming-book
______________________________________________
about: http://www.boulangerlabs.com/about/richardboulanger/
about: http://www.csounds.com/community/developers/dr-richard-boulanger/
music: http://www.csounds.com/community/developers/dr-richard-boulanger/dr-richard-boulanger-music/

______________________________________________
email: rboulanger@berklee.edu
facebook: https://www.facebook.com/richard.boulanger.58

Csound mailing list Csound@listserv.heanet.ie https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to https://github.com/csound/csound/issues Discussions of bugs and features can be posted here

Csound mailing list Csound@listserv.heanet.ie https://listserv.heanet.ie/cgi-bin/wa?A0=CSOUND Send bugs reports to https://github.com/csound/csound/issues Discussions of bugs and features can be posted here