Hi All,

I'm following up with what was on the user's list regarding realtime
performance and dropouts.  What I've noticed so far is that in some
cases, Csound6 has demonstrated better performance, and in others,
worse.  I have to use reproducible cases so far, one in realtime and
one in rendering to disk:

1. In my piece "TimeSphere"[1], using the CSD that is available in the
project.zip posted on the site[2], I had much worse rendering times to
disk than with Csound5.  I had not checked with the released CSD,
which I ended up rendering with ksmps=16.  With ksmps=1, I had seen
something like a 13 minute render with CS5 and a 24 minute render with
CS6.

2. In my tied notes example from my article in the Csound Journal[3],
running the CSD[4] exhibited dropouts with CS6 while it did not with
CS5. I believe I ran both with Core Audio, but would need to double
check that.  The dropouts occurred with the tied notes section, and so
it was easiest to run that with:

csound yi_tied_notes_example.csd -+skip_seconds=36

To note, I just tried to render in realtime and did not get breakups
this last time.  However, rendering to disk I get:


CS5
Elapsed time at end of performance: real: 2.346s, CPU: 2.116s

CS6
Elapsed time at end of performance: real: 13.516s, CPU: 7.042s


Thanks!
steven


[1] - http://kunstmusik.com/2013/07/27/timesphere/
[2] - http://www.kunstmusik.com/projects/timeSphere.zip
[3] - http://www.csounds.com/journal/2005fall/tiedNotes.html
[4] - http://www.csounds.com/journal/2005fall/examples/yi_tied_notes_example.csd

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent 
caught up. So what steps can you take to put your SQL databases under 
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=49501711&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I had a look at your tied note example. It appears that with ksmps=1 the difference in performance is significant,
but with ksmps=10 and ksmps=100 the difference is small or nonexistent. 

The ties are not to blame, setting iskip to 0 does not seem to change the performance.

So, in conclusion, yes, there is a penalty, but that is only with very small ksmps, 
which should be avoided anyway.

Victor

On 5 Aug 2013, at 22:39, Steven Yi wrote:

> Hi All,
> 
> I'm following up with what was on the user's list regarding realtime
> performance and dropouts.  What I've noticed so far is that in some
> cases, Csound6 has demonstrated better performance, and in others,
> worse.  I have to use reproducible cases so far, one in realtime and
> one in rendering to disk:
> 
> 1. In my piece "TimeSphere"[1], using the CSD that is available in the
> project.zip posted on the site[2], I had much worse rendering times to
> disk than with Csound5.  I had not checked with the released CSD,
> which I ended up rendering with ksmps=16.  With ksmps=1, I had seen
> something like a 13 minute render with CS5 and a 24 minute render with
> CS6.
> 
> 2. In my tied notes example from my article in the Csound Journal[3],
> running the CSD[4] exhibited dropouts with CS6 while it did not with
> CS5. I believe I ran both with Core Audio, but would need to double
> check that.  The dropouts occurred with the tied notes section, and so
> it was easiest to run that with:
> 
> csound yi_tied_notes_example.csd -+skip_seconds=36
> 
> To note, I just tried to render in realtime and did not get breakups
> this last time.  However, rendering to disk I get:
> 
> 
> CS5
> Elapsed time at end of performance: real: 2.346s, CPU: 2.116s
> 
> CS6
> Elapsed time at end of performance: real: 13.516s, CPU: 7.042s
> 
> 
> Thanks!
> steven
> 
> 
> [1] - http://kunstmusik.com/2013/07/27/timesphere/
> [2] - http://www.kunstmusik.com/projects/timeSphere.zip
> [3] - http://www.csounds.com/journal/2005fall/tiedNotes.html
> [4] - http://www.csounds.com/journal/2005fall/examples/yi_tied_notes_example.csd
> 
> ------------------------------------------------------------------------------
> Get your SQL database under version control now!
> Version control is standard for application code, but databases havent 
> caught up. So what steps can you take to put your SQL databases under 
> version control? Why should you start doing it? Read more to find out.
> http://pubads.g.doubleclick.net/gampad/clk?id=49501711&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

Dr Victor Lazzarini
Senior Lecturer
Dept. of Music
NUI Maynooth Ireland
tel.: +353 1 708 3545
Victor dot Lazzarini AT nuim dot ie




------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent 
caught up. So what steps can you take to put your SQL databases under 
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I don't think we can just say "don't use ksmps=1" and ignore this
problem.  I use ksmps=1 as part of my composing workflow when
rendering to disk to have higher quality renders.  The increase in
time lead me to not even use Csound 6 to finish my last piece. Also
using a-var's for everything does not work either as most filters in
Csound don't accept a-rate controls (cutoff, resonance, etc.).

If anything, I think we need to determine why it has become orders of
magnitude slower at lower ksmps.  I think there's a real issue here
and that we should try to improve the situation if possible.

On Tue, Aug 6, 2013 at 12:56 AM, Victor Lazzarini
 wrote:
> I had a look at your tied note example. It appears that with ksmps=1 the difference in performance is significant,
> but with ksmps=10 and ksmps=100 the difference is small or nonexistent.
>
> The ties are not to blame, setting iskip to 0 does not seem to change the performance.
>
> So, in conclusion, yes, there is a penalty, but that is only with very small ksmps,
> which should be avoided anyway.
>
> Victor
>
> On 5 Aug 2013, at 22:39, Steven Yi wrote:
>
>> Hi All,
>>
>> I'm following up with what was on the user's list regarding realtime
>> performance and dropouts.  What I've noticed so far is that in some
>> cases, Csound6 has demonstrated better performance, and in others,
>> worse.  I have to use reproducible cases so far, one in realtime and
>> one in rendering to disk:
>>
>> 1. In my piece "TimeSphere"[1], using the CSD that is available in the
>> project.zip posted on the site[2], I had much worse rendering times to
>> disk than with Csound5.  I had not checked with the released CSD,
>> which I ended up rendering with ksmps=16.  With ksmps=1, I had seen
>> something like a 13 minute render with CS5 and a 24 minute render with
>> CS6.
>>
>> 2. In my tied notes example from my article in the Csound Journal[3],
>> running the CSD[4] exhibited dropouts with CS6 while it did not with
>> CS5. I believe I ran both with Core Audio, but would need to double
>> check that.  The dropouts occurred with the tied notes section, and so
>> it was easiest to run that with:
>>
>> csound yi_tied_notes_example.csd -+skip_seconds=36
>>
>> To note, I just tried to render in realtime and did not get breakups
>> this last time.  However, rendering to disk I get:
>>
>>
>> CS5
>> Elapsed time at end of performance: real: 2.346s, CPU: 2.116s
>>
>> CS6
>> Elapsed time at end of performance: real: 13.516s, CPU: 7.042s
>>
>>
>> Thanks!
>> steven
>>
>>
>> [1] - http://kunstmusik.com/2013/07/27/timesphere/
>> [2] - http://www.kunstmusik.com/projects/timeSphere.zip
>> [3] - http://www.csounds.com/journal/2005fall/tiedNotes.html
>> [4] - http://www.csounds.com/journal/2005fall/examples/yi_tied_notes_example.csd
>>
>> ------------------------------------------------------------------------------
>> Get your SQL database under version control now!
>> Version control is standard for application code, but databases havent
>> caught up. So what steps can you take to put your SQL databases under
>> version control? Why should you start doing it? Read more to find out.
>> http://pubads.g.doubleclick.net/gampad/clk?id=49501711&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>
> Dr Victor Lazzarini
> Senior Lecturer
> Dept. of Music
> NUI Maynooth Ireland
> tel.: +353 1 708 3545
> Victor dot Lazzarini AT nuim dot ie
>
>
>
>
> ------------------------------------------------------------------------------
> Get your SQL database under version control now!
> Version control is standard for application code, but databases havent
> caught up. So what steps can you take to put your SQL databases under
> version control? Why should you start doing it? Read more to find out.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent 
caught up. So what steps can you take to put your SQL databases under 
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

While I am not saying to let it go, I think this is might not be easily fixed. I remember John ffitch
counting operations and noting that there were more in Csound 6. 

Victor
On 6 Aug 2013, at 09:44, Steven Yi wrote:

> I don't think we can just say "don't use ksmps=1" and ignore this
> problem.  I use ksmps=1 as part of my composing workflow when
> rendering to disk to have higher quality renders.  The increase in
> time lead me to not even use Csound 6 to finish my last piece. Also
> using a-var's for everything does not work either as most filters in
> Csound don't accept a-rate controls (cutoff, resonance, etc.).
> 
> If anything, I think we need to determine why it has become orders of
> magnitude slower at lower ksmps.  I think there's a real issue here
> and that we should try to improve the situation if possible.
> 
> On Tue, Aug 6, 2013 at 12:56 AM, Victor Lazzarini
>  wrote:
>> I had a look at your tied note example. It appears that with ksmps=1 the difference in performance is significant,
>> but with ksmps=10 and ksmps=100 the difference is small or nonexistent.
>> 
>> The ties are not to blame, setting iskip to 0 does not seem to change the performance.
>> 
>> So, in conclusion, yes, there is a penalty, but that is only with very small ksmps,
>> which should be avoided anyway.
>> 
>> Victor
>> 
>> On 5 Aug 2013, at 22:39, Steven Yi wrote:
>> 
>>> Hi All,
>>> 
>>> I'm following up with what was on the user's list regarding realtime
>>> performance and dropouts.  What I've noticed so far is that in some
>>> cases, Csound6 has demonstrated better performance, and in others,
>>> worse.  I have to use reproducible cases so far, one in realtime and
>>> one in rendering to disk:
>>> 
>>> 1. In my piece "TimeSphere"[1], using the CSD that is available in the
>>> project.zip posted on the site[2], I had much worse rendering times to
>>> disk than with Csound5.  I had not checked with the released CSD,
>>> which I ended up rendering with ksmps=16.  With ksmps=1, I had seen
>>> something like a 13 minute render with CS5 and a 24 minute render with
>>> CS6.
>>> 
>>> 2. In my tied notes example from my article in the Csound Journal[3],
>>> running the CSD[4] exhibited dropouts with CS6 while it did not with
>>> CS5. I believe I ran both with Core Audio, but would need to double
>>> check that.  The dropouts occurred with the tied notes section, and so
>>> it was easiest to run that with:
>>> 
>>> csound yi_tied_notes_example.csd -+skip_seconds=36
>>> 
>>> To note, I just tried to render in realtime and did not get breakups
>>> this last time.  However, rendering to disk I get:
>>> 
>>> 
>>> CS5
>>> Elapsed time at end of performance: real: 2.346s, CPU: 2.116s
>>> 
>>> CS6
>>> Elapsed time at end of performance: real: 13.516s, CPU: 7.042s
>>> 
>>> 
>>> Thanks!
>>> steven
>>> 
>>> 
>>> [1] - http://kunstmusik.com/2013/07/27/timesphere/
>>> [2] - http://www.kunstmusik.com/projects/timeSphere.zip
>>> [3] - http://www.csounds.com/journal/2005fall/tiedNotes.html
>>> [4] - http://www.csounds.com/journal/2005fall/examples/yi_tied_notes_example.csd
>>> 
>>> ------------------------------------------------------------------------------
>>> Get your SQL database under version control now!
>>> Version control is standard for application code, but databases havent
>>> caught up. So what steps can you take to put your SQL databases under
>>> version control? Why should you start doing it? Read more to find out.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=49501711&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>> 
>> Dr Victor Lazzarini
>> Senior Lecturer
>> Dept. of Music
>> NUI Maynooth Ireland
>> tel.: +353 1 708 3545
>> Victor dot Lazzarini AT nuim dot ie
>> 
>> 
>> 
>> 
>> ------------------------------------------------------------------------------
>> Get your SQL database under version control now!
>> Version control is standard for application code, but databases havent
>> caught up. So what steps can you take to put your SQL databases under
>> version control? Why should you start doing it? Read more to find out.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
> 
> ------------------------------------------------------------------------------
> Get your SQL database under version control now!
> Version control is standard for application code, but databases havent 
> caught up. So what steps can you take to put your SQL databases under 
> version control? Why should you start doing it? Read more to find out.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

Dr Victor Lazzarini
Senior Lecturer
Dept. of Music
NUI Maynooth Ireland
tel.: +353 1 708 3545
Victor dot Lazzarini AT nuim dot ie




------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent 
caught up. So what steps can you take to put your SQL databases under 
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

> While I am not saying to let it go, I think this is might not be easily
> fixed. I remember John ffitch
> counting operations and noting that there were more in Csound 6.
>
> Victor

I could do a comparison between cs5 and cs6 if you were to provide a
simple example where there was a significant difference.  Instruction
counting is not too hard and we cab see if there is a group of slower
operations or ig it is spread.

I suspect that the --sample-accurate code could be a drag in ksmps=1
There are ways to modify the way it is done that might change performance

==John




------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent 
caught up. So what steps can you take to put your SQL databases under 
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Hi John and Victor,

Thanks for replies.  John, the CSD I had linked to at the beginning of
this thread is what I used recently for testing:

http://www.csounds.com/journal/2005fall/examples/yi_tied_notes_example.csd

And I ran it with:

csound yi_tied_notes_example.csd -+skip_seconds=36

to jump ahead to the tied notes section that would give audio breakups
in realtime.  The difference in performance was noticeable to disk as
well, and may be easier to use for profiling. I had used XCode's
instrument a week ago to look at this problem but did not come to a
definitive conclusion on what could be the difference.  I think a lot
of things are still on the table, including: sample-accurate code,
initialization code changes, missing compile-time optimizations in
CS6, possibly any changes for UDO init and calling. There may be other
things that could be contributing too, and I have no idea if it's a
lot of small things or one big thing that could be the difference.

Thanks!
steven



On Tue, Aug 6, 2013 at 12:00 PM,   wrote:
>> While I am not saying to let it go, I think this is might not be easily
>> fixed. I remember John ffitch
>> counting operations and noting that there were more in Csound 6.
>>
>> Victor
>
> I could do a comparison between cs5 and cs6 if you were to provide a
> simple example where there was a significant difference.  Instruction
> counting is not too hard and we cab see if there is a group of slower
> operations or ig it is spread.
>
> I suspect that the --sample-accurate code could be a drag in ksmps=1
> There are ways to modify the way it is done that might change performance
>
> ==John
>
>
>
>
> ------------------------------------------------------------------------------
> Get your SQL database under version control now!
> Version control is standard for application code, but databases havent
> caught up. So what steps can you take to put your SQL databases under
> version control? Why should you start doing it? Read more to find out.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

------------------------------------------------------------------------------
Get your SQL database under version control now!
Version control is standard for application code, but databases havent 
caught up. So what steps can you take to put your SQL databases under 
version control? Why should you start doing it? Read more to find out.
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net