I have installed the recent 5.15 on windows xp laptop, from the binary
distribution (thanks Michael). The processor is an i5 M520 2.4 Ghz.
The windows task manager show 4 threads.

I have tested several csd files generated by AVSynthesis with the
following option "--new-parser -j4 -d -m0 -g -odac3", and several
ksmps settings (64,256). The audio output is through the internal
laptop audio device, and the tests are done using the csound API.

The sound output is randomly noisy. By comparison, the sound is clean
with the following option "--new-parser -d -m0 -g -odac3".  I monitor
the cpu usage using the windows task manager, and don't see any high
usage for any of the 4 threads, either with -j4 or without.

I have attached the csd file if anybody wants to test if the sound is
clean with multicore option on, in other environments.

Best
jp

Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Thanks Alex for these tests.
Did you try ksmps=64 without -j4 ? Any click in this case ?

On Wed, Jan 4, 2012 at 1:47 PM, Alex AB  wrote:
> Hello,
> I just installed 5.15 on a win 7 laptop (i5 processor).
> I ran your .csd in winXound using the laptop's soundcard.
> I got a few clicks at  sr = 44100 ksmps = 64 but none at ksmps = 128.
> It also worked at sr = 48000 ksmps 128.
> However, I had an error message concerning tclcsound.dll (report from
> winXsound attached). I check the csound directory and tclcsound.dll is
> present.
> winXound also couldn't find csound5gui.exe. I searched the csound directory
> and couldn't find it.
>
> I then tried running the .csd using an Emu 0204USB soundcard. It worked well
> at sr = 44100, but I got a blue screen when I requested a higher sampling
> rate (on the previous versions, I never encountered that issue and could
> request sampling rates up to sr = 192000).
>
> My wife is out, so I had to leave the noise room to care for the kids, I'll
> try again tonight with the Emu card, setting the sampling rate manually
> before launching winXound and see if it works that way or not.
>
> CsoundQT couldn't run the .csd because of an issue in instr 2 (report also
> attached, also, same tclcsound error).
>
> Like I said, I shall be doing further tests tonight when I can be in the
> noise room, I'd be glad to try out whatever actions you think might be
> useful.
>
> Hope that helps.
>
> Alex
>
>
> On 4 January 2012 11:17, jean-pierre lemoine  wrote:
>>
>> I have installed the recent 5.15 on windows xp laptop, from the binary
>> distribution (thanks Michael). The processor is an i5 M520 2.4 Ghz.
>> The windows task manager show 4 threads.
>>
>> I have tested several csd files generated by AVSynthesis with the
>> following option "--new-parser -j4 -d -m0 -g -odac3", and several
>> ksmps settings (64,256). The audio output is through the internal
>> laptop audio device, and the tests are done using the csound API.
>>
>> The sound output is randomly noisy. By comparison, the sound is clean
>> with the following option "--new-parser -d -m0 -g -odac3".  I monitor
>> the cpu usage using the windows task manager, and don't see any high
>> usage for any of the 4 threads, either with -j4 or without.
>>
>> I have attached the csd file if anybody wants to test if the sound is
>> clean with multicore option on, in other environments.
>>
>> Best
>> jp
>>
>> Send bugs reports to the Sourceforge bug tracker
>>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>
>


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Due to the nature of multicore, expect small ksmps to be less
efficient than a single thread. Multicore will save significant time
for some Csound pieces, may take longer than single threading for
other pieces.

The reason is that starting a thread and managing switches between
threads, accesses to shared data, etc. involve overhead as compared to
a single thread. If this overhead is small compared to the ability to
run long stretches of code at the same time, multicore will run
faster. Otherwise, slower.

Regards,
Mike

On Thu, Jan 5, 2012 at 6:21 AM, Alex AB  wrote:
> I'm sorry, I wrote kr = ... when I should have written ksmps = ... in my
> previous message.
>
>
>
> On 5 January 2012 12:17, Alex AB  wrote:
>>
>> Hello,
>> I tried without the -j4 flag and it worked well down to kr = 64 at sr =
>> 44100.
>>
>> I had a few clicks at kr = 32, but less than kr = 64 with the -j4 flag.
>>
>> At sr = 48000 I needed kr = 128 for it to work well.
>>
>> All this using the laptop's built-in sound device.
>>
>> I shall be creating an other thread for higher frame rates since I can't
>> test them with the laptop's soundcard.
>>
>> Hope that helps.
>>
>> Alex
>>
>>
>> On 4 January 2012 16:01, jean-pierre lemoine  wrote:
>>>
>>> Thanks Alex for these tests.
>>> Did you try ksmps=64 without -j4 ? Any click in this case ?
>>>
>>> On Wed, Jan 4, 2012 at 1:47 PM, Alex AB 
>>> wrote:
>>> > Hello,
>>> > I just installed 5.15 on a win 7 laptop (i5 processor).
>>> > I ran your .csd in winXound using the laptop's soundcard.
>>> > I got a few clicks at  sr = 44100 ksmps = 64 but none at ksmps = 128.
>>> > It also worked at sr = 48000 ksmps 128.
>>> > However, I had an error message concerning tclcsound.dll (report from
>>> > winXsound attached). I check the csound directory and tclcsound.dll is
>>> > present.
>>> > winXound also couldn't find csound5gui.exe. I searched the csound
>>> > directory
>>> > and couldn't find it.
>>> >
>>> > I then tried running the .csd using an Emu 0204USB soundcard. It worked
>>> > well
>>> > at sr = 44100, but I got a blue screen when I requested a higher
>>> > sampling
>>> > rate (on the previous versions, I never encountered that issue and
>>> > could
>>> > request sampling rates up to sr = 192000).
>>> >
>>> > My wife is out, so I had to leave the noise room to care for the kids,
>>> > I'll
>>> > try again tonight with the Emu card, setting the sampling rate manually
>>> > before launching winXound and see if it works that way or not.
>>> >
>>> > CsoundQT couldn't run the .csd because of an issue in instr 2 (report
>>> > also
>>> > attached, also, same tclcsound error).
>>> >
>>> > Like I said, I shall be doing further tests tonight when I can be in
>>> > the
>>> > noise room, I'd be glad to try out whatever actions you think might be
>>> > useful.
>>> >
>>> > Hope that helps.
>>> >
>>> > Alex
>>> >
>>> >
>>> > On 4 January 2012 11:17, jean-pierre lemoine 
>>> > wrote:
>>> >>
>>> >> I have installed the recent 5.15 on windows xp laptop, from the binary
>>> >> distribution (thanks Michael). The processor is an i5 M520 2.4 Ghz.
>>> >> The windows task manager show 4 threads.
>>> >>
>>> >> I have tested several csd files generated by AVSynthesis with the
>>> >> following option "--new-parser -j4 -d -m0 -g -odac3", and several
>>> >> ksmps settings (64,256). The audio output is through the internal
>>> >> laptop audio device, and the tests are done using the csound API.
>>> >>
>>> >> The sound output is randomly noisy. By comparison, the sound is clean
>>> >> with the following option "--new-parser -d -m0 -g -odac3".  I monitor
>>> >> the cpu usage using the windows task manager, and don't see any high
>>> >> usage for any of the 4 threads, either with -j4 or without.
>>> >>
>>> >> I have attached the csd file if anybody wants to test if the sound is
>>> >> clean with multicore option on, in other environments.
>>> >>
>>> >> Best
>>> >> jp
>>> >>
>>> >> Send bugs reports to the Sourceforge bug tracker
>>> >>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>> >> Discussions of bugs and features can be posted here
>>> >> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>> >> "unsubscribe
>>> >> csound"
>>> >>
>>> >
>>>
>>>
>>> Send bugs reports to the Sourceforge bug tracker
>>>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>> Discussions of bugs and features can be posted here
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>> csound"
>>>
>>
>



-- 
Michael Gogins
Irreducible Productions
http://www.michael-gogins.com
Michael dot Gogins at gmail dot com


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Yes, I understand, but do you have more information on how to
characterize a csound piece regrading multicore behaviours?
Also, is a ksmps of 256 small? I found the performance very bad, but
maybe this is because of the csound piece (many globals, use of zak)

Best
jp


On Thu, Jan 5, 2012 at 2:29 PM, Michael Gogins  wrote:
> Due to the nature of multicore, expect small ksmps to be less
> efficient than a single thread. Multicore will save significant time
> for some Csound pieces, may take longer than single threading for
> other pieces.
>
> The reason is that starting a thread and managing switches between
> threads, accesses to shared data, etc. involve overhead as compared to
> a single thread. If this overhead is small compared to the ability to
> run long stretches of code at the same time, multicore will run
> faster. Otherwise, slower.
>
> Regards,
> Mike
>
> On Thu, Jan 5, 2012 at 6:21 AM, Alex AB  wrote:
>> I'm sorry, I wrote kr = ... when I should have written ksmps = ... in my
>> previous message.
>>
>>
>>
>> On 5 January 2012 12:17, Alex AB  wrote:
>>>
>>> Hello,
>>> I tried without the -j4 flag and it worked well down to kr = 64 at sr =
>>> 44100.
>>>
>>> I had a few clicks at kr = 32, but less than kr = 64 with the -j4 flag.
>>>
>>> At sr = 48000 I needed kr = 128 for it to work well.
>>>
>>> All this using the laptop's built-in sound device.
>>>
>>> I shall be creating an other thread for higher frame rates since I can't
>>> test them with the laptop's soundcard.
>>>
>>> Hope that helps.
>>>
>>> Alex
>>>
>>>
>>> On 4 January 2012 16:01, jean-pierre lemoine  wrote:
>>>>
>>>> Thanks Alex for these tests.
>>>> Did you try ksmps=64 without -j4 ? Any click in this case ?
>>>>
>>>> On Wed, Jan 4, 2012 at 1:47 PM, Alex AB 
>>>> wrote:
>>>> > Hello,
>>>> > I just installed 5.15 on a win 7 laptop (i5 processor).
>>>> > I ran your .csd in winXound using the laptop's soundcard.
>>>> > I got a few clicks at  sr = 44100 ksmps = 64 but none at ksmps = 128.
>>>> > It also worked at sr = 48000 ksmps 128.
>>>> > However, I had an error message concerning tclcsound.dll (report from
>>>> > winXsound attached). I check the csound directory and tclcsound.dll is
>>>> > present.
>>>> > winXound also couldn't find csound5gui.exe. I searched the csound
>>>> > directory
>>>> > and couldn't find it.
>>>> >
>>>> > I then tried running the .csd using an Emu 0204USB soundcard. It worked
>>>> > well
>>>> > at sr = 44100, but I got a blue screen when I requested a higher
>>>> > sampling
>>>> > rate (on the previous versions, I never encountered that issue and
>>>> > could
>>>> > request sampling rates up to sr = 192000).
>>>> >
>>>> > My wife is out, so I had to leave the noise room to care for the kids,
>>>> > I'll
>>>> > try again tonight with the Emu card, setting the sampling rate manually
>>>> > before launching winXound and see if it works that way or not.
>>>> >
>>>> > CsoundQT couldn't run the .csd because of an issue in instr 2 (report
>>>> > also
>>>> > attached, also, same tclcsound error).
>>>> >
>>>> > Like I said, I shall be doing further tests tonight when I can be in
>>>> > the
>>>> > noise room, I'd be glad to try out whatever actions you think might be
>>>> > useful.
>>>> >
>>>> > Hope that helps.
>>>> >
>>>> > Alex
>>>> >
>>>> >
>>>> > On 4 January 2012 11:17, jean-pierre lemoine 
>>>> > wrote:
>>>> >>
>>>> >> I have installed the recent 5.15 on windows xp laptop, from the binary
>>>> >> distribution (thanks Michael). The processor is an i5 M520 2.4 Ghz.
>>>> >> The windows task manager show 4 threads.
>>>> >>
>>>> >> I have tested several csd files generated by AVSynthesis with the
>>>> >> following option "--new-parser -j4 -d -m0 -g -odac3", and several
>>>> >> ksmps settings (64,256). The audio output is through the internal
>>>> >> laptop audio device, and the tests are done using the csound API.
>>>> >>
>>>> >> The sound output is randomly noisy. By comparison, the sound is clean
>>>> >> with the following option "--new-parser -d -m0 -g -odac3".  I monitor
>>>> >> the cpu usage using the windows task manager, and don't see any high
>>>> >> usage for any of the 4 threads, either with -j4 or without.
>>>> >>
>>>> >> I have attached the csd file if anybody wants to test if the sound is
>>>> >> clean with multicore option on, in other environments.
>>>> >>
>>>> >> Best
>>>> >> jp
>>>> >>
>>>> >> Send bugs reports to the Sourceforge bug tracker
>>>> >>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>> >> Discussions of bugs and features can be posted here
>>>> >> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>> >> "unsubscribe
>>>> >> csound"
>>>> >>
>>>> >
>>>>
>>>>
>>>> Send bugs reports to the Sourceforge bug tracker
>>>>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>> Discussions of bugs and features can be posted here
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>>> csound"
>>>>
>>>
>>
>
>
>
> --
> Michael Gogins
> Irreducible Productions
> http://www.michael-gogins.com
> Michael dot Gogins at gmail dot com
>
>
> Send bugs reports to the Sourceforge bug tracker
>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
>


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

In general, I have found that so far, few pieces benefit from
multicore. It depends more on the compute intensive nature of the
instruments, how many instances of the same instrument are running at
the same time, etc., than on ksmps but ksmps also is important. In
this context 256 is medium.

I expect multicore will become increasingly important as (a) we finish
the "costing" component of the Csound implementation, which should
speed things up some more, (b) more computers have multiple cores, and
(c) composers become more ambitious with the computer power.

Hope this helps,
Mike

On Thu, Jan 5, 2012 at 8:42 AM, jean-pierre lemoine
 wrote:
> Yes, I understand, but do you have more information on how to
> characterize a csound piece regrading multicore behaviours?
> Also, is a ksmps of 256 small? I found the performance very bad, but
> maybe this is because of the csound piece (many globals, use of zak)
>
> Best
> jp
>
>
> On Thu, Jan 5, 2012 at 2:29 PM, Michael Gogins  wrote:
>> Due to the nature of multicore, expect small ksmps to be less
>> efficient than a single thread. Multicore will save significant time
>> for some Csound pieces, may take longer than single threading for
>> other pieces.
>>
>> The reason is that starting a thread and managing switches between
>> threads, accesses to shared data, etc. involve overhead as compared to
>> a single thread. If this overhead is small compared to the ability to
>> run long stretches of code at the same time, multicore will run
>> faster. Otherwise, slower.
>>
>> Regards,
>> Mike
>>
>> On Thu, Jan 5, 2012 at 6:21 AM, Alex AB  wrote:
>>> I'm sorry, I wrote kr = ... when I should have written ksmps = ... in my
>>> previous message.
>>>
>>>
>>>
>>> On 5 January 2012 12:17, Alex AB  wrote:
>>>>
>>>> Hello,
>>>> I tried without the -j4 flag and it worked well down to kr = 64 at sr =
>>>> 44100.
>>>>
>>>> I had a few clicks at kr = 32, but less than kr = 64 with the -j4 flag.
>>>>
>>>> At sr = 48000 I needed kr = 128 for it to work well.
>>>>
>>>> All this using the laptop's built-in sound device.
>>>>
>>>> I shall be creating an other thread for higher frame rates since I can't
>>>> test them with the laptop's soundcard.
>>>>
>>>> Hope that helps.
>>>>
>>>> Alex
>>>>
>>>>
>>>> On 4 January 2012 16:01, jean-pierre lemoine  wrote:
>>>>>
>>>>> Thanks Alex for these tests.
>>>>> Did you try ksmps=64 without -j4 ? Any click in this case ?
>>>>>
>>>>> On Wed, Jan 4, 2012 at 1:47 PM, Alex AB 
>>>>> wrote:
>>>>> > Hello,
>>>>> > I just installed 5.15 on a win 7 laptop (i5 processor).
>>>>> > I ran your .csd in winXound using the laptop's soundcard.
>>>>> > I got a few clicks at  sr = 44100 ksmps = 64 but none at ksmps = 128.
>>>>> > It also worked at sr = 48000 ksmps 128.
>>>>> > However, I had an error message concerning tclcsound.dll (report from
>>>>> > winXsound attached). I check the csound directory and tclcsound.dll is
>>>>> > present.
>>>>> > winXound also couldn't find csound5gui.exe. I searched the csound
>>>>> > directory
>>>>> > and couldn't find it.
>>>>> >
>>>>> > I then tried running the .csd using an Emu 0204USB soundcard. It worked
>>>>> > well
>>>>> > at sr = 44100, but I got a blue screen when I requested a higher
>>>>> > sampling
>>>>> > rate (on the previous versions, I never encountered that issue and
>>>>> > could
>>>>> > request sampling rates up to sr = 192000).
>>>>> >
>>>>> > My wife is out, so I had to leave the noise room to care for the kids,
>>>>> > I'll
>>>>> > try again tonight with the Emu card, setting the sampling rate manually
>>>>> > before launching winXound and see if it works that way or not.
>>>>> >
>>>>> > CsoundQT couldn't run the .csd because of an issue in instr 2 (report
>>>>> > also
>>>>> > attached, also, same tclcsound error).
>>>>> >
>>>>> > Like I said, I shall be doing further tests tonight when I can be in
>>>>> > the
>>>>> > noise room, I'd be glad to try out whatever actions you think might be
>>>>> > useful.
>>>>> >
>>>>> > Hope that helps.
>>>>> >
>>>>> > Alex
>>>>> >
>>>>> >
>>>>> > On 4 January 2012 11:17, jean-pierre lemoine 
>>>>> > wrote:
>>>>> >>
>>>>> >> I have installed the recent 5.15 on windows xp laptop, from the binary
>>>>> >> distribution (thanks Michael). The processor is an i5 M520 2.4 Ghz.
>>>>> >> The windows task manager show 4 threads.
>>>>> >>
>>>>> >> I have tested several csd files generated by AVSynthesis with the
>>>>> >> following option "--new-parser -j4 -d -m0 -g -odac3", and several
>>>>> >> ksmps settings (64,256). The audio output is through the internal
>>>>> >> laptop audio device, and the tests are done using the csound API.
>>>>> >>
>>>>> >> The sound output is randomly noisy. By comparison, the sound is clean
>>>>> >> with the following option "--new-parser -d -m0 -g -odac3".  I monitor
>>>>> >> the cpu usage using the windows task manager, and don't see any high
>>>>> >> usage for any of the 4 threads, either with -j4 or without.
>>>>> >>
>>>>> >> I have attached the csd file if anybody wants to test if the sound is
>>>>> >> clean with multicore option on, in other environments.
>>>>> >>
>>>>> >> Best
>>>>> >> jp
>>>>> >>
>>>>> >> Send bugs reports to the Sourceforge bug tracker
>>>>> >>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>>> >> Discussions of bugs and features can be posted here
>>>>> >> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>>>>> >> "unsubscribe
>>>>> >> csound"
>>>>> >>
>>>>> >
>>>>>
>>>>>
>>>>> Send bugs reports to the Sourceforge bug tracker
>>>>>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>>> Discussions of bugs and features can be posted here
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>>>> csound"
>>>>>
>>>>
>>>
>>
>>
>>
>> --
>> Michael Gogins
>> Irreducible Productions
>> http://www.michael-gogins.com
>> Michael dot Gogins at gmail dot com
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
>>
>
>
> Send bugs reports to the Sourceforge bug tracker
>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
>



-- 
Michael Gogins
Irreducible Productions
http://www.michael-gogins.com
Michael dot Gogins at gmail dot com


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"