[Cs-dev] linseg bug?
| Date | 2011-07-17 10:58 |
| From | aum |
| Subject | [Cs-dev] linseg bug? |
"linseg 1 p3 2" gives constant value 1.000 for long P3. Is it my fault or the bug? Here is csd file: |
| Date | 2011-07-17 12:57 |
| From | jpff@cs.bath.ac.uk |
| Subject | Re: [Cs-dev] linseg bug? |
Cannot reproduce. Or in other words works for me. What platform? I notice you are using 5.19 which i a a littl eold, but libnseg has not changed for ages (I think) ==John ff > "linseg 1 p3 2" gives constant value 1.000 for long P3. Is it my fault or > the > bug? > Here is csd file: > > |
| Date | 2011-07-17 13:52 |
| From | peiman khosravi |
| Subject | Re: [Cs-dev] linseg bug? |
| Attachments | None None |
| It may be an old printks issue? On 17 July 2011 12:57, <jpff@cs.bath.ac.uk> wrote: Cannot reproduce. Or in other words works for me. |
| Date | 2011-07-17 21:24 |
| From | aum |
| Subject | Re: [Cs-dev] linseg bug? |
It is not the printks bug. linseg output is sometimes wrong. Constant or wrong range (e.g. linseg 4, p3, 14 goes over 15), depending on p3, kr, ksmps. My system is Fedora 12 with Planet CCRMA and csound 5.10 beta. Upgrade will be probably the solution. Thanks. aum -- View this message in context: http://csound.1045644.n5.nabble.com/linseg-bug-tp4596059p4597238.html Sent from the Csound - Dev mailing list archive at Nabble.com. ------------------------------------------------------------------------------ AppSumo Presents a FREE Video for the SourceForge Community by Eric Ries, the creator of the Lean Startup Methodology on "Lean Startup Secrets Revealed." This video shows you how to validate your ideas, optimize your ideas and identify your business strategy. http://p.sf.net/sfu/appsumosfdev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2011-07-17 23:51 |
| From | Victor Lazzarini |
| Subject | Re: [Cs-dev] linseg bug? |
I think this might not be a bug, but a limitation of 32bit floats. The increment is too small for a 32bit float. If you run the same csd with a doubles version of csound, it will work. That is why John says it works for him. Maybe we should consider making all envelope increments double precision by default? On 17 Jul 2011, at 21:24, aum wrote: > It is not the printks bug. linseg output is sometimes wrong. > Constant or > wrong range (e.g. linseg 4, p3, 14 goes over 15), depending on p3, kr, > ksmps. My system is Fedora 12 with Planet CCRMA and csound 5.10 beta. > Upgrade will be probably the solution. Thanks. aum > > -- > View this message in context: http://csound.1045644.n5.nabble.com/linseg-bug-tp4596059p4597238.html > Sent from the Csound - Dev mailing list archive at Nabble.com. > > ------------------------------------------------------------------------------ > AppSumo Presents a FREE Video for the SourceForge Community by Eric > Ries, the creator of the Lean Startup Methodology on "Lean Startup > Secrets Revealed." This video shows you how to validate your ideas, > optimize your ideas and identify your business strategy. > http://p.sf.net/sfu/appsumosfdev2dev > _______________________________________________ > 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 ------------------------------------------------------------------------------ AppSumo Presents a FREE Video for the SourceForge Community by Eric Ries, the creator of the Lean Startup Methodology on "Lean Startup Secrets Revealed." This video shows you how to validate your ideas, optimize your ideas and identify your business strategy. http://p.sf.net/sfu/appsumosfdev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2011-07-18 11:17 |
| From | aum |
| Subject | Re: [Cs-dev] linseg bug? |
| Attachments | None None |
| I think you are right. 1/(4410*5000) is about 2^-28, near the limit of
32bit float precision. I do not know what the default precision should be. I do not know the effect of 64bit arithmetic in 32bit csound on the speed of calculations. I think for most users with relatively short sounds 32bit envelopes are ok. For long sounds 64bit version of csound is the option (this is my case, too). Maybe it would help others to mention it somewhere in the documentation, maybe with some practical tips. Thanks aum On 07/18/2011 12:51 AM, Victor Lazzarini [via Csound] wrote: > I think this might not be a bug, but a limitation of 32bit floats. The > increment is too small for a 32bit float. If you run the > same csd with a doubles version of csound, it will work. That is why > John says it works for him. > > Maybe we should consider making all envelope increments double > precision by default? View this message in context: Re: linseg bug? Sent from the Csound - Dev mailing list archive at Nabble.com. |
| Date | 2011-07-18 11:20 |
| From | Victor Lazzarini |
| Subject | Re: [Cs-dev] linseg bug? |
| Attachments | None None |
| In my opinion all increments should be double, as in oscillators. I have changed line/expon but all other envelope opcodes should be changed too. Victor On 18 Jul 2011, at 11:17, aum wrote: I think you are right. 1/(4410*5000) is about 2^-28, near the limit of Dr Victor Lazzarini Senior Lecturer Dept. of Music NUI Maynooth Ireland tel.: +353 1 708 3545 Victor dot Lazzarini AT nuim dot ie |
| Date | 2011-07-18 12:05 |
| From | jpff@cs.bath.ac.uk |
| Subject | Re: [Cs-dev] linseg bug? |
In my opinion we should be using doubles everywhere! ==John ff > In my opinion all increments should be double, as in oscillators. I > have changed line/expon but all other envelope opcodes should be > changed too. > > Victor > On 18 Jul 2011, at 11:17, aum wrote: > >> I think you are right. 1/(4410*5000) is about 2^-28, near the limit of >> 32bit float precision. >> >> I do not know what the default precision should be. I do not know the >> effect of 64bit arithmetic in 32bit csound on the speed of >> calculations. >> I think for most users with relatively short sounds 32bit envelopes >> are >> ok. For long sounds 64bit version of csound is the option (this is my >> case, too). Maybe it would help others to mention it somewhere in the >> documentation, maybe with some practical tips. >> Thanks >> aum >> >> On 07/18/2011 12:51 AM, Victor Lazzarini [via Csound] wrote: >> > I think this might not be a bug, but a limitation of 32bit floats. >> The >> > increment is too small for a 32bit float. If you run the >> > same csd with a doubles version of csound, it will work. That is why >> > John says it works for him. >> > >> > Maybe we should consider making all envelope increments double >> > precision by default? >> >> >> View this message in context: Re: linseg bug? >> Sent from the Csound - Dev mailing list archive at Nabble.com. >> ------------------------------------------------------------------------------ >> AppSumo Presents a FREE Video for the SourceForge Community by Eric >> Ries, the creator of the Lean Startup Methodology on "Lean Startup >> Secrets Revealed." This video shows you how to validate your ideas, >> optimize your ideas and identify your business strategy. >> http://p.sf.net/sfu/appsumosfdev2dev_______________________________________________ >> 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 > > > > ------------------------------------------------------------------------------ > AppSumo Presents a FREE Video for the SourceForge Community by Eric > Ries, the creator of the Lean Startup Methodology on "Lean Startup > Secrets Revealed." This video shows you how to validate your ideas, > optimize your ideas and identify your business strategy. > http://p.sf.net/sfu/appsumosfdev2dev_______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > ------------------------------------------------------------------------------ AppSumo Presents a FREE Video for the SourceForge Community by Eric Ries, the creator of the Lean Startup Methodology on "Lean Startup Secrets Revealed." This video shows you how to validate your ideas, optimize your ideas and identify your business strategy. http://p.sf.net/sfu/appsumosfdev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2011-07-18 12:08 |
| From | Victor Lazzarini |
| Subject | Re: [Cs-dev] linseg bug? |
yes, but we have not binned the floats version yet, so I think we need to change the increments. Victor On 18 Jul 2011, at 12:05, jpff@cs.bath.ac.uk wrote: > In my opinion we should be using doubles everywhere! > ==John ff > >> In my opinion all increments should be double, as in oscillators. I >> have changed line/expon but all other envelope opcodes should be >> changed too. >> >> Victor >> On 18 Jul 2011, at 11:17, aum wrote: >> >>> I think you are right. 1/(4410*5000) is about 2^-28, near the >>> limit of >>> 32bit float precision. >>> >>> I do not know what the default precision should be. I do not know >>> the >>> effect of 64bit arithmetic in 32bit csound on the speed of >>> calculations. >>> I think for most users with relatively short sounds 32bit envelopes >>> are >>> ok. For long sounds 64bit version of csound is the option (this is >>> my >>> case, too). Maybe it would help others to mention it somewhere in >>> the >>> documentation, maybe with some practical tips. >>> Thanks >>> aum >>> >>> On 07/18/2011 12:51 AM, Victor Lazzarini [via Csound] wrote: >>>> I think this might not be a bug, but a limitation of 32bit floats. >>> The >>>> increment is too small for a 32bit float. If you run the >>>> same csd with a doubles version of csound, it will work. That is >>>> why >>>> John says it works for him. >>>> >>>> Maybe we should consider making all envelope increments double >>>> precision by default? >>> >>> >>> View this message in context: Re: linseg bug? >>> Sent from the Csound - Dev mailing list archive at Nabble.com. >>> ------------------------------------------------------------------------------ >>> AppSumo Presents a FREE Video for the SourceForge Community by Eric >>> Ries, the creator of the Lean Startup Methodology on "Lean Startup >>> Secrets Revealed." This video shows you how to validate your ideas, >>> optimize your ideas and identify your business strategy. >>> http://p.sf.net/sfu/appsumosfdev2dev_______________________________________________ >>> 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 >> >> >> >> ------------------------------------------------------------------------------ >> AppSumo Presents a FREE Video for the SourceForge Community by Eric >> Ries, the creator of the Lean Startup Methodology on "Lean Startup >> Secrets Revealed." This video shows you how to validate your ideas, >> optimize your ideas and identify your business strategy. >> http://p.sf.net/sfu/appsumosfdev2dev_______________________________________________ >> Csound-devel mailing list >> Csound-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/csound-devel >> > > > > ------------------------------------------------------------------------------ > AppSumo Presents a FREE Video for the SourceForge Community by Eric > Ries, the creator of the Lean Startup Methodology on "Lean Startup > Secrets Revealed." This video shows you how to validate your ideas, > optimize your ideas and identify your business strategy. > http://p.sf.net/sfu/appsumosfdev2dev > _______________________________________________ > 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 ------------------------------------------------------------------------------ AppSumo Presents a FREE Video for the SourceForge Community by Eric Ries, the creator of the Lean Startup Methodology on "Lean Startup Secrets Revealed." This video shows you how to validate your ideas, optimize your ideas and identify your business strategy. http://p.sf.net/sfu/appsumosfdev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2011-07-18 12:24 |
| From | jpff@cs.bath.ac.uk |
| Subject | Re: [Cs-dev] linseg bug? |
> yes, but we have not binned the floats version yet, so I think we need > to change the increments. > > Victor Do you have a list of which need change? I would rather not either of us repeat the work of the other. ------------------------------------------------------------------------------ AppSumo Presents a FREE Video for the SourceForge Community by Eric Ries, the creator of the Lean Startup Methodology on "Lean Startup Secrets Revealed." This video shows you how to validate your ideas, optimize your ideas and identify your business strategy. http://p.sf.net/sfu/appsumosfdev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |
| Date | 2011-07-18 12:37 |
| From | Victor Lazzarini |
| Subject | Re: [Cs-dev] linseg bug? |
I guess all of the ones in ugens1 to start with. I've done line and expon just to check it's worth it (in GIT now). Victor On 18 Jul 2011, at 12:24, jpff@cs.bath.ac.uk wrote: >> yes, but we have not binned the floats version yet, so I think we >> need >> to change the increments. >> >> Victor > > > Do you have a list of which need change? I would rather not either > of us > repeat the work of the other. > > > > ------------------------------------------------------------------------------ > AppSumo Presents a FREE Video for the SourceForge Community by Eric > Ries, the creator of the Lean Startup Methodology on "Lean Startup > Secrets Revealed." This video shows you how to validate your ideas, > optimize your ideas and identify your business strategy. > http://p.sf.net/sfu/appsumosfdev2dev > _______________________________________________ > 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 ------------------------------------------------------------------------------ AppSumo Presents a FREE Video for the SourceForge Community by Eric Ries, the creator of the Lean Startup Methodology on "Lean Startup Secrets Revealed." This video shows you how to validate your ideas, optimize your ideas and identify your business strategy. http://p.sf.net/sfu/appsumosfdev2dev _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net |