[Cs-dev] Wrong lines in error message report

[Cs-dev] Wrong lines in error message report

Date	2009-04-18 20:45
From	Andres Cabrera
Subject	[Cs-dev] Wrong lines in error message report
	Hi all, Csound is currently reporting errors with a 1 line difference. Try this file: sr = 44100 ksmps = 128 nchnls = 2 0dbfs = 1 error instr 1 asig oscil 0.5, 440, 1 endin fgfds f 1 0 1024 10 1 f 10 0 8 -2 60 62 64 65 67 69 71 72 i 1 0 20 e Cheers, Andrés ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-18 21:44
From	Anthony Kozar
Subject	Re: [Cs-dev] Wrong lines in error message report
	I am getting the correct line numbers with your example -------------- orch compiler: 14 lines read error: no legal opcode, line 11: error instr 1 error: no legal opcode, line 17: fgfds 2 syntax errors in orchestra. compilation invalid -------------- Anthony Andres Cabrera wrote on 4/18/09 3:45 PM: > > > > > > sr = 44100 > ksmps = 128 > nchnls = 2 > 0dbfs = 1 > > error > > instr 1 > asig oscil 0.5, 440, 1 > endin > > fgfds > > > > f 1 0 1024 10 1 > f 10 0 8 -2 60 62 64 65 67 69 71 72 > i 1 0 20 > e > > ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-19 01:58
From	Andres Cabrera
Subject	Re: [Cs-dev] Wrong lines in error message report
	Odd... I get: 14 lines read error: no legal opcode, line 12: error instr 1 error: no legal opcode, line 18: fgfds On the command line and: error: no legal opcode, line 13: error instr 1 error: no legal opcode, line 19: fgfds 2 syntax errors in orchestra. compilation invalid in QuteCsound.... Any one get the wrong line number as well? I tried copy pasting from my email in case it was some newline weirdness, but it's not... Cheers, Andrés On Sat, Apr 18, 2009 at 3:44 PM, Anthony Kozar wrote: > I am getting the correct line numbers with your example > > -------------- > orch compiler: > 14 lines read > error: no legal opcode, line 11: > error > instr 1 > error: no legal opcode, line 17: > fgfds > 2 syntax errors in orchestra. compilation invalid > -------------- > > Anthony > > Andres Cabrera wrote on 4/18/09 3:45 PM: > >> >> >> >> >> >> sr = 44100 >> ksmps = 128 >> nchnls = 2 >> 0dbfs = 1 >> >> error >> >> instr 1 >> asig oscil 0.5, 440, 1 >> endin >> >> fgfds >> >> >> >> f 1 0 1024 10 1 >> f 10 0 8 -2 60 62 64 65 67 69 71 72 >> i 1 0 20 >> e >> >> > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > -- Andrés ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-19 18:34
From	Anthony Kozar
Subject	Re: [Cs-dev] Wrong lines in error message report
	I get the correct line numbers with QuteCsound too. I am using the latest CVS code for Csound. Anthony Andres Cabrera wrote on 4/18/09 8:58 PM: > Odd... I get: > > 14 lines read > error: no legal opcode, line 12: > error > instr 1 > error: no legal opcode, line 18: > fgfds > > On the command line and: > error: no legal opcode, line 13: > error > instr 1 > error: no legal opcode, line 19: > fgfds > 2 syntax errors in orchestra. compilation invalid > > in QuteCsound.... > Any one get the wrong line number as well? > I tried copy pasting from my email in case it was some newline > weirdness, but it's not... > > Cheers, > Andrés > > On Sat, Apr 18, 2009 at 3:44 PM, Anthony Kozar > wrote: >> I am getting the correct line numbers with your example >> >> -------------- >> orch compiler: >> 14 lines read >> error: no legal opcode, line 11: >> error >> instr 1 >> error: no legal opcode, line 17: >> fgfds >> 2 syntax errors in orchestra. compilation invalid >> -------------- ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-20 02:13
From	Andres Cabrera
Subject	Re: [Cs-dev] Wrong lines in error message report
	I just updated from cvs and rebuilt, but get the same problem.... I'm stumped... Any one else see this? Cheers, Andrés On Sun, Apr 19, 2009 at 12:34 PM, Anthony Kozar wrote: > I get the correct line numbers with QuteCsound too. I am using the latest > CVS code for Csound. > > Anthony > > Andres Cabrera wrote on 4/18/09 8:58 PM: > >> Odd... I get: >> >> 14 lines read >> error: no legal opcode, line 12: >> error >> instr 1 >> error: no legal opcode, line 18: >> fgfds >> >> On the command line and: >> error: no legal opcode, line 13: >> error >> instr 1 >> error: no legal opcode, line 19: >> fgfds >> 2 syntax errors in orchestra. compilation invalid >> >> in QuteCsound.... >> Any one get the wrong line number as well? >> I tried copy pasting from my email in case it was some newline >> weirdness, but it's not... >> >> Cheers, >> Andrés >> >> On Sat, Apr 18, 2009 at 3:44 PM, Anthony Kozar >> wrote: >>> I am getting the correct line numbers with your example >>> >>> -------------- >>> orch compiler: >>> 14 lines read >>> error: no legal opcode, line 11: >>> error >>> instr 1 >>> error: no legal opcode, line 17: >>> fgfds >>> 2 syntax errors in orchestra. compilation invalid >>> -------------- > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > -- Andrés ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-20 22:48
From	victor
Subject	[Cs-dev] lpc on doubles is buggy
	Hi everyone, I was just alerted by a user that lpc did not seem to work on doubles. In his case, it appeared to be just the fact that the lpc file he used had been created by a floats csound. However, on testing this, I am getting a whistling sound instead of the correct resynthesis on csound doubles version. If anyone wants to try, you can use the CSD below with any 7-second analysis file (or any other duration, if you change p3) sr=44100 kr=4410 ksmps=10 nchnls=1 instr 1 ktim line 0, p3, p3 krmsr,krmso,kerr,kcps lpread ktim,"flute.lpc" asig buzz 100, 200, 30, 1 afil lpreson asig out afil endin f1 0 16384 10 1 i1 0 7 I'll see if I can find where the problem is (could be in the analysis or in the synthesis, who knows?), but the clue is that the problem is not on floats csound. So it might be a case of chasing a place where MYFLT is assumed to be single precision. If anyone has some suggestions, let me know. Victor ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-20 23:27
From	joachim heintz
Subject	Re: [Cs-dev] Wrong lines in error message report
	I'm getting the same as Anthony: 14 lines read error: no legal opcode, line 11: error instr 1 error: no legal opcode, line 17: fgfds 2 syntax errors in orchestra. compilation invalid in Qutecsound and in the Terminal. joachim Am 20.04.2009 um 03:13 schrieb Andres Cabrera: > I just updated from cvs and rebuilt, but get the same problem.... I'm > stumped... Any one else see this? > > > Cheers, > Andrés > > On Sun, Apr 19, 2009 at 12:34 PM, Anthony Kozar > wrote: >> I get the correct line numbers with QuteCsound too. I am using the >> latest >> CVS code for Csound. >> >> Anthony >> >> Andres Cabrera wrote on 4/18/09 8:58 PM: >> >>> Odd... I get: >>> >>> 14 lines read >>> error: no legal opcode, line 12: >>> error >>> instr 1 >>> error: no legal opcode, line 18: >>> fgfds >>> >>> On the command line and: >>> error: no legal opcode, line 13: >>> error >>> instr 1 >>> error: no legal opcode, line 19: >>> fgfds >>> 2 syntax errors in orchestra. compilation invalid >>> >>> in QuteCsound.... >>> Any one get the wrong line number as well? >>> I tried copy pasting from my email in case it was some newline >>> weirdness, but it's not... >>> >>> Cheers, >>> Andrés >>> >>> On Sat, Apr 18, 2009 at 3:44 PM, Anthony Kozar >>> wrote: >>>> I am getting the correct line numbers with your example >>>> >>>> -------------- >>>> orch compiler: >>>> 14 lines read >>>> error: no legal opcode, line 11: >>>> error >>>> instr 1 >>>> error: no legal opcode, line 17: >>>> fgfds >>>> 2 syntax errors in orchestra. compilation invalid >>>> -------------- >> >> >> ------------------------------------------------------------------------------ >> Stay on top of everything new and different, both inside and >> around Java (TM) technology - register by April 22, and save >> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >> 300 plus technical and hands-on sessions. Register today. >> Use priority code J9JMT32. http://p.sf.net/sfu/p >> _______________________________________________ >> Csound-devel mailing list >> Csound-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/csound-devel >> > > > > -- > > > Andrés > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 03:54
From	Anthony Kozar
Subject	Re: [Cs-dev] lpc on doubles is buggy
	Did you test this on the latest CVS? I fixed exactly such a problem with lpread on doubles back on Feb. 28 (or at least I thought I fixed it!). Anthony victor wrote on 4/20/09 5:48 PM: > I was just alerted by a user that lpc did not seem to work on > doubles. In his case, it appeared to be just the fact that the > lpc file he used had been created by a floats csound. > > However, on testing this, I am getting a whistling sound instead > of the correct resynthesis on csound doubles version. ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 08:31
From	victor
Subject	Re: [Cs-dev] lpc on doubles is buggy
	You're quite right! I tested this at home and only realised now how long it's been since I'd updated (I thought it had been not too long ago.). Thanks. I'm glad I asked. Victor ----- Original Message ----- From: "Anthony Kozar" To: "New Csound Developer list" Sent: Tuesday, April 21, 2009 3:54 AM Subject: Re: [Cs-dev] lpc on doubles is buggy > Did you test this on the latest CVS? I fixed exactly such a problem with > lpread on doubles back on Feb. 28 (or at least I thought I fixed it!). > > Anthony > > victor wrote on 4/20/09 5:48 PM: > >> I was just alerted by a user that lpc did not seem to work on >> doubles. In his case, it appeared to be just the fact that the >> lpc file he used had been created by a floats csound. >> >> However, on testing this, I am getting a whistling sound instead >> of the correct resynthesis on csound doubles version. > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 10:52
From	victor
Subject	[Cs-dev] --num-threads on Windows
Attachments	choir.csd None None
	Hi everyone, just thought I'd check the status of --num-threads, since talking to Steven last week, I have realised I had not looked into it lately. I am getting a crash if I set my --num-threads > 1, with the attached CSD. It uses a UDO, if I remove it, it works; but I am also getting a distorted result (say if I substitute the UDOs for simple oscillators). With a different CSD, if I have this code: if isrc == 1 goto noise asig diskin "flutec3.wav", 1, 0, 1 asig = asig/2000 goto next noise: asig rand 1 next: csound complains 'diskin not initialised' and hangs. If I make the if ...goto into if... igoto, it complains and crashes. If I remove the if ... goto, it works. I am not sure what the code does, so I can't really play with it; perhaps someone else could comment on this. Regards Victor

Date	2009-04-21 11:03
From	victor
Subject	Re: [Cs-dev] --num-threads on Windows
	Further to this, I realised the distortion has been caused by the global reverb, so I expect this is a problem with global vars. I think it is probably a well-known issue, too. In summary: the detected issues were: 1. UDOs 2. Global variables 3. control of flow (if... etc) Otherwise on a modest multicore (2 core) like the one I have here at home, it runs faster. Some stats: The choir.csd with oscils instead of UDOs 10.103 secs on 1 thread 7.253 secs on 2 threads 7.301 secs on 4 threads 8.385 secs on 6 threads It looks good. Victor ----- Original Message ----- From: "victor" To: "Developer discussions" Sent: Tuesday, April 21, 2009 10:52 AM Subject: [Cs-dev] --num-threads on Windows > Hi everyone, > > just thought I'd check the status of --num-threads, since talking > to Steven last week, I have realised I had not looked into it > lately. > > I am getting a crash if I set my --num-threads > 1, with the attached > CSD. It uses a UDO, if I remove it, it works; but I am also getting > a distorted result (say if I substitute the UDOs for simple oscillators). > > With a different CSD, if I have this code: > > if isrc == 1 goto noise > asig diskin "flutec3.wav", 1, 0, 1 > asig = asig/2000 > goto next > noise: asig rand 1 > next: > > csound complains 'diskin not initialised' and hangs. > If I make the if ...goto into if... igoto, it complains and crashes. > > If I remove the if ... goto, it works. > > I am not sure what the code does, so I can't really play with it; perhaps > someone else could comment on this. > > Regards > > Victor > > > > > -------------------------------------------------------------------------------- > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p -------------------------------------------------------------------------------- > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 11:26
From	victor
Subject	Re: [Cs-dev] --num-threads on Windows
Attachments	tests.csd None None
	I was a little rushed in my assessment. There are further problems, the attached CSD demonstrates: with --num-threads > 1 it produces lots of clicks/discontinuities in the output waveform. It seems to be a problem in the spout mixing. Victor ----- Original Message ----- From: "victor" To: "Developer discussions" Sent: Tuesday, April 21, 2009 11:03 AM Subject: Re: [Cs-dev] --num-threads on Windows > Further to this, I realised the distortion has been caused by the > global reverb, so I expect this is a problem with global vars. I think > it is probably a well-known issue, too. > > In summary: the detected issues were: > > 1. UDOs > 2. Global variables > 3. control of flow (if... etc) > > Otherwise on a modest multicore (2 core) like the one I have here > at home, it runs faster. Some stats: > > The choir.csd with oscils instead of UDOs > 10.103 secs on 1 thread > 7.253 secs on 2 threads > 7.301 secs on 4 threads > 8.385 secs on 6 threads > > It looks good. > > Victor > > ----- Original Message ----- > From: "victor" > To: "Developer discussions" > Sent: Tuesday, April 21, 2009 10:52 AM > Subject: [Cs-dev] --num-threads on Windows > > >> Hi everyone, >> >> just thought I'd check the status of --num-threads, since talking >> to Steven last week, I have realised I had not looked into it >> lately. >> >> I am getting a crash if I set my --num-threads > 1, with the attached >> CSD. It uses a UDO, if I remove it, it works; but I am also getting >> a distorted result (say if I substitute the UDOs for simple oscillators). >> >> With a different CSD, if I have this code: >> >> if isrc == 1 goto noise >> asig diskin "flutec3.wav", 1, 0, 1 >> asig = asig/2000 >> goto next >> noise: asig rand 1 >> next: >> >> csound complains 'diskin not initialised' and hangs. >> If I make the if ...goto into if... igoto, it complains and crashes. >> >> If I remove the if ... goto, it works. >> >> I am not sure what the code does, so I can't really play with it; perhaps >> someone else could comment on this. >> >> Regards >> >> Victor >> >> >> >> >> > > > -------------------------------------------------------------------------------- > > >> ------------------------------------------------------------------------------ >> Stay on top of everything new and different, both inside and >> around Java (TM) technology - register by April 22, and save >> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >> 300 plus technical and hands-on sessions. Register today. >> Use priority code J9JMT32. http://p.sf.net/sfu/p > > > -------------------------------------------------------------------------------- > > >> _______________________________________________ >> Csound-devel mailing list >> Csound-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/csound-devel >> > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel

Date	2009-04-21 17:48
From	Anthony Kozar
Subject	Re: [Cs-dev] lpc on doubles is buggy
	Well, I'm glad that it really is working now :) BTW, hetro/adsyn seem to have a similar problem with doubles that I have not looked into yet, if anyone else wants to take a look. http://www.nabble.com/hetro64-to22308770.html#a22308770 Anthony victor wrote on 4/21/09 3:31 AM: > You're quite right! I tested this at home and only realised now how long > it's been since I'd updated (I thought it had been not too long ago.). > Thanks. I'm glad I asked. > > Victor > > ----- Original Message ----- > From: "Anthony Kozar" > To: "New Csound Developer list" > Sent: Tuesday, April 21, 2009 3:54 AM > Subject: Re: [Cs-dev] lpc on doubles is buggy > > >> Did you test this on the latest CVS? I fixed exactly such a problem with >> lpread on doubles back on Feb. 28 (or at least I thought I fixed it!). >> >> Anthony >> >> victor wrote on 4/20/09 5:48 PM: >> >>> I was just alerted by a user that lpc did not seem to work on >>> doubles. In his case, it appeared to be just the fact that the >>> lpc file he used had been created by a floats csound. >>> >>> However, on testing this, I am getting a whistling sound instead >>> of the correct resynthesis on csound doubles version. ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 19:05
From	joachim heintz
Subject	Re: [Cs-dev] --num-threads on Windows
	Did you try it with diskin2 instead of diskin? I think, diskin isn't working correctly at the moment. Am 21.04.2009 um 12:26 schrieb victor: > I was a little rushed in my assessment. There are further problems, > the > attached CSD demonstrates: with --num-threads > 1 it produces > lots of clicks/discontinuities in the output waveform. It seems to be > a problem in the spout mixing. > > Victor > > ----- Original Message ----- From: "victor" > To: "Developer discussions" > Sent: Tuesday, April 21, 2009 11:03 AM > Subject: Re: [Cs-dev] --num-threads on Windows > > >> Further to this, I realised the distortion has been caused by the >> global reverb, so I expect this is a problem with global vars. I >> think >> it is probably a well-known issue, too. >> >> In summary: the detected issues were: >> >> 1. UDOs >> 2. Global variables >> 3. control of flow (if... etc) >> >> Otherwise on a modest multicore (2 core) like the one I have here >> at home, it runs faster. Some stats: >> >> The choir.csd with oscils instead of UDOs >> 10.103 secs on 1 thread >> 7.253 secs on 2 threads >> 7.301 secs on 4 threads >> 8.385 secs on 6 threads >> >> It looks good. >> >> Victor >> >> ----- Original Message ----- From: "victor" >> >> To: "Developer discussions" >> Sent: Tuesday, April 21, 2009 10:52 AM >> Subject: [Cs-dev] --num-threads on Windows >> >> >>> Hi everyone, >>> >>> just thought I'd check the status of --num-threads, since talking >>> to Steven last week, I have realised I had not looked into it >>> lately. >>> >>> I am getting a crash if I set my --num-threads > 1, with the >>> attached >>> CSD. It uses a UDO, if I remove it, it works; but I am also getting >>> a distorted result (say if I substitute the UDOs for simple >>> oscillators). >>> >>> With a different CSD, if I have this code: >>> >>> if isrc == 1 goto noise >>> asig diskin "flutec3.wav", 1, 0, 1 >>> asig = asig/2000 >>> goto next >>> noise: asig rand 1 >>> next: >>> >>> csound complains 'diskin not initialised' and hangs. >>> If I make the if ...goto into if... igoto, it complains and crashes. >>> >>> If I remove the if ... goto, it works. >>> >>> I am not sure what the code does, so I can't really play with it; >>> perhaps >>> someone else could comment on this. >>> >>> Regards >>> >>> Victor >>> >>> >>> >>> >>> >> >> >> -------------------------------------------------------------------------------- >> >> >>> ------------------------------------------------------------------------------ >>> Stay on top of everything new and different, both inside and >>> around Java (TM) technology - register by April 22, and save >>> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >>> 300 plus technical and hands-on sessions. Register today. >>> Use priority code J9JMT32. http://p.sf.net/sfu/p >> >> >> -------------------------------------------------------------------------------- >> >> >>> _______________________________________________ >>> Csound-devel mailing list >>> Csound-devel@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/csound-devel >>> >> >> >> ------------------------------------------------------------------------------ >> Stay on top of everything new and different, both inside and >> around Java (TM) technology - register by April 22, and save >> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >> 300 plus technical and hands-on sessions. Register today. >> Use priority code J9JMT32. http://p.sf.net/sfu/p >> _______________________________________________ >> Csound-devel mailing list >> Csound-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/csound-devel > < > tests > .csd > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p_______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 19:48
From	Steven Yi
Subject	Re: [Cs-dev] lpc on doubles is buggy
	Hi All, I know LPC did not work with doubles for a very long time. So are they now fully functioning? If so, that'd be great! :) steven On Tue, Apr 21, 2009 at 9:48 AM, Anthony Kozar wrote: > Well, I'm glad that it really is working now :) > > BTW, hetro/adsyn seem to have a similar problem with doubles that I have not > looked into yet, if anyone else wants to take a look. > > http://www.nabble.com/hetro64-to22308770.html#a22308770 > > Anthony > > victor wrote on 4/21/09 3:31 AM: > >> You're quite right! I tested this at home and only realised now how long >> it's been since I'd updated (I thought it had been not too long ago.). >> Thanks. I'm glad I asked. >> >> Victor >> >> ----- Original Message ----- >> From: "Anthony Kozar" >> To: "New Csound Developer list" >> Sent: Tuesday, April 21, 2009 3:54 AM >> Subject: Re: [Cs-dev] lpc on doubles is buggy >> >> >>> Did you test this on the latest CVS? I fixed exactly such a problem with >>> lpread on doubles back on Feb. 28 (or at least I thought I fixed it!). >>> >>> Anthony >>> >>> victor wrote on 4/20/09 5:48 PM: >>> >>>> I was just alerted by a user that lpc did not seem to work on >>>> doubles. In his case, it appeared to be just the fact that the >>>> lpc file he used had been created by a floats csound. >>>> >>>> However, on testing this, I am getting a whistling sound instead >>>> of the correct resynthesis on csound doubles version. > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 20:07
From	victor
Subject	Re: [Cs-dev] lpc on doubles is buggy
	yes, it seems to work to the same level of floats LPC. The main thing is that files are version (and platform) dependen. I think this has confused some users. We might need to put a note to that effect in the manual. Victor ----- Original Message ----- From: "Steven Yi" To: "Developer discussions" Sent: Tuesday, April 21, 2009 7:48 PM Subject: Re: [Cs-dev] lpc on doubles is buggy Hi All, I know LPC did not work with doubles for a very long time. So are they now fully functioning? If so, that'd be great! :) steven On Tue, Apr 21, 2009 at 9:48 AM, Anthony Kozar wrote: > Well, I'm glad that it really is working now :) > > BTW, hetro/adsyn seem to have a similar problem with doubles that I have > not > looked into yet, if anyone else wants to take a look. > > http://www.nabble.com/hetro64-to22308770.html#a22308770 > > Anthony > > victor wrote on 4/21/09 3:31 AM: > >> You're quite right! I tested this at home and only realised now how long >> it's been since I'd updated (I thought it had been not too long ago.). >> Thanks. I'm glad I asked. >> >> Victor >> >> ----- Original Message ----- >> From: "Anthony Kozar" >> To: "New Csound Developer list" >> Sent: Tuesday, April 21, 2009 3:54 AM >> Subject: Re: [Cs-dev] lpc on doubles is buggy >> >> >>> Did you test this on the latest CVS? I fixed exactly such a problem with >>> lpread on doubles back on Feb. 28 (or at least I thought I fixed it!). >>> >>> Anthony >>> >>> victor wrote on 4/20/09 5:48 PM: >>> >>>> I was just alerted by a user that lpc did not seem to work on >>>> doubles. In his case, it appeared to be just the fact that the >>>> lpc file he used had been created by a floats csound. >>>> >>>> However, on testing this, I am getting a whistling sound instead >>>> of the correct resynthesis on csound doubles version. > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 20:07
From	victor
Subject	Re: [Cs-dev] --num-threads on Windows
	diskin2 is the same. ----- Original Message ----- From: "joachim heintz" To: "Developer discussions" Sent: Tuesday, April 21, 2009 7:05 PM Subject: Re: [Cs-dev] --num-threads on Windows > Did you try it with diskin2 instead of diskin? I think, diskin isn't > working correctly at the moment. > > > > Am 21.04.2009 um 12:26 schrieb victor: > >> I was a little rushed in my assessment. There are further problems, >> the >> attached CSD demonstrates: with --num-threads > 1 it produces >> lots of clicks/discontinuities in the output waveform. It seems to be >> a problem in the spout mixing. >> >> Victor >> >> ----- Original Message ----- From: "victor" >> To: "Developer discussions" >> Sent: Tuesday, April 21, 2009 11:03 AM >> Subject: Re: [Cs-dev] --num-threads on Windows >> >> >>> Further to this, I realised the distortion has been caused by the >>> global reverb, so I expect this is a problem with global vars. I >>> think >>> it is probably a well-known issue, too. >>> >>> In summary: the detected issues were: >>> >>> 1. UDOs >>> 2. Global variables >>> 3. control of flow (if... etc) >>> >>> Otherwise on a modest multicore (2 core) like the one I have here >>> at home, it runs faster. Some stats: >>> >>> The choir.csd with oscils instead of UDOs >>> 10.103 secs on 1 thread >>> 7.253 secs on 2 threads >>> 7.301 secs on 4 threads >>> 8.385 secs on 6 threads >>> >>> It looks good. >>> >>> Victor >>> >>> ----- Original Message ----- From: "victor" >>> >>> To: "Developer discussions" >>> Sent: Tuesday, April 21, 2009 10:52 AM >>> Subject: [Cs-dev] --num-threads on Windows >>> >>> >>>> Hi everyone, >>>> >>>> just thought I'd check the status of --num-threads, since talking >>>> to Steven last week, I have realised I had not looked into it >>>> lately. >>>> >>>> I am getting a crash if I set my --num-threads > 1, with the >>>> attached >>>> CSD. It uses a UDO, if I remove it, it works; but I am also getting >>>> a distorted result (say if I substitute the UDOs for simple >>>> oscillators). >>>> >>>> With a different CSD, if I have this code: >>>> >>>> if isrc == 1 goto noise >>>> asig diskin "flutec3.wav", 1, 0, 1 >>>> asig = asig/2000 >>>> goto next >>>> noise: asig rand 1 >>>> next: >>>> >>>> csound complains 'diskin not initialised' and hangs. >>>> If I make the if ...goto into if... igoto, it complains and crashes. >>>> >>>> If I remove the if ... goto, it works. >>>> >>>> I am not sure what the code does, so I can't really play with it; >>>> perhaps >>>> someone else could comment on this. >>>> >>>> Regards >>>> >>>> Victor >>>> >>>> >>>> >>>> >>>> >>> >>> >>> -------------------------------------------------------------------------------- >>> >>> >>>> ------------------------------------------------------------------------------ >>>> Stay on top of everything new and different, both inside and >>>> around Java (TM) technology - register by April 22, and save >>>> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >>>> 300 plus technical and hands-on sessions. Register today. >>>> Use priority code J9JMT32. http://p.sf.net/sfu/p >>> >>> >>> -------------------------------------------------------------------------------- >>> >>> >>>> _______________________________________________ >>>> Csound-devel mailing list >>>> Csound-devel@lists.sourceforge.net >>>> https://lists.sourceforge.net/lists/listinfo/csound-devel >>>> >>> >>> >>> ------------------------------------------------------------------------------ >>> Stay on top of everything new and different, both inside and >>> around Java (TM) technology - register by April 22, and save >>> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >>> 300 plus technical and hands-on sessions. Register today. >>> Use priority code J9JMT32. http://p.sf.net/sfu/p >>> _______________________________________________ >>> Csound-devel mailing list >>> Csound-devel@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/csound-devel >> < >> tests >> .csd >> > >> ------------------------------------------------------------------------------ >> Stay on top of everything new and different, both inside and >> around Java (TM) technology - register by April 22, and save >> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >> 300 plus technical and hands-on sessions. Register today. >> Use priority code J9JMT32. >> http://p.sf.net/sfu/p_______________________________________________ >> Csound-devel mailing list >> Csound-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/csound-devel > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 20:34
From	Anthony Kozar
Subject	Re: [Cs-dev] lpc on doubles is buggy
	I think there is a note in the section on "Csound 64", but that is perhaps not very noticeable. Regarding the lpread bug, I noticed today it was reported at least 4 years ago. The irony is that there was a comment in the code explaining exactly what the bad assumption was which implied the solution ... I believe that Richard Dobson has suggested in the past that we create new analysis formats with robust support for reading them regardless of float size or platform endianness. A good idea, but perhaps for a future date (Csound 6?). Anthony victor wrote on 4/21/09 3:07 PM: > yes, it seems to work to the same level of floats LPC. The main thing is > that files are version (and platform) dependen. I think this has confused > some users. We might need to put a note to that effect in the manual. > > Victor > > ----- Original Message ----- > From: "Steven Yi" > To: "Developer discussions" > Sent: Tuesday, April 21, 2009 7:48 PM > Subject: Re: [Cs-dev] lpc on doubles is buggy > > > Hi All, > > I know LPC did not work with doubles for a very long time. So are > they now fully functioning? If so, that'd be great! :) ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-21 22:21
From	victor
Subject	Re: [Cs-dev] lpc on doubles is buggy
	yes, I think this is definitely a good idea. There might be a format for LPC already around (or perhaps SDIF does it?) ----- Original Message ----- From: "Anthony Kozar" To: "New Csound Developer list" Sent: Tuesday, April 21, 2009 8:34 PM Subject: Re: [Cs-dev] lpc on doubles is buggy >I think there is a note in the section on "Csound 64", but that is perhaps > not very noticeable. > > Regarding the lpread bug, I noticed today it was reported at least 4 years > ago. The irony is that there was a comment in the code explaining exactly > what the bad assumption was which implied the solution ... > > I believe that Richard Dobson has suggested in the past that we create new > analysis formats with robust support for reading them regardless of float > size or platform endianness. A good idea, but perhaps for a future date > (Csound 6?). > > Anthony > > victor wrote on 4/21/09 3:07 PM: > >> yes, it seems to work to the same level of floats LPC. The main thing is >> that files are version (and platform) dependen. I think this has confused >> some users. We might need to put a note to that effect in the manual. >> >> Victor >> >> ----- Original Message ----- >> From: "Steven Yi" >> To: "Developer discussions" >> Sent: Tuesday, April 21, 2009 7:48 PM >> Subject: Re: [Cs-dev] lpc on doubles is buggy >> >> >> Hi All, >> >> I know LPC did not work with doubles for a very long time. So are >> they now fully functioning? If so, that'd be great! :) > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net

Date	2009-04-22 15:22
From	Andres Cabrera
Subject	Re: [Cs-dev] Wrong lines in error message report
	Hmm... strange. I'll make a clean checkout when I get back and try again... Cheers, Andrés On Mon, Apr 20, 2009 at 5:27 PM, joachim heintz wrote: > I'm getting the same as Anthony: > > 14 lines read > error: no legal opcode, line 11: > error > instr 1 > error: no legal opcode, line 17: > fgfds > 2 syntax errors in orchestra. compilation invalid > > in Qutecsound and in the Terminal. > > joachim > > > > Am 20.04.2009 um 03:13 schrieb Andres Cabrera: > >> I just updated from cvs and rebuilt, but get the same problem.... I'm >> stumped... Any one else see this? >> >> >> Cheers, >> Andrés >> >> On Sun, Apr 19, 2009 at 12:34 PM, Anthony Kozar >> wrote: >>> I get the correct line numbers with QuteCsound too. I am using the >>> latest >>> CVS code for Csound. >>> >>> Anthony >>> >>> Andres Cabrera wrote on 4/18/09 8:58 PM: >>> >>>> Odd... I get: >>>> >>>> 14 lines read >>>> error: no legal opcode, line 12: >>>> error >>>> instr 1 >>>> error: no legal opcode, line 18: >>>> fgfds >>>> >>>> On the command line and: >>>> error: no legal opcode, line 13: >>>> error >>>> instr 1 >>>> error: no legal opcode, line 19: >>>> fgfds >>>> 2 syntax errors in orchestra. compilation invalid >>>> >>>> in QuteCsound.... >>>> Any one get the wrong line number as well? >>>> I tried copy pasting from my email in case it was some newline >>>> weirdness, but it's not... >>>> >>>> Cheers, >>>> Andrés >>>> >>>> On Sat, Apr 18, 2009 at 3:44 PM, Anthony Kozar >>>> wrote: >>>>> I am getting the correct line numbers with your example >>>>> >>>>> -------------- >>>>> orch compiler: >>>>> 14 lines read >>>>> error: no legal opcode, line 11: >>>>> error >>>>> instr 1 >>>>> error: no legal opcode, line 17: >>>>> fgfds >>>>> 2 syntax errors in orchestra. compilation invalid >>>>> -------------- >>> >>> >>> ------------------------------------------------------------------------------ >>> Stay on top of everything new and different, both inside and >>> around Java (TM) technology - register by April 22, and save >>> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >>> 300 plus technical and hands-on sessions. Register today. >>> Use priority code J9JMT32. http://p.sf.net/sfu/p >>> _______________________________________________ >>> Csound-devel mailing list >>> Csound-devel@lists.sourceforge.net >>> https://lists.sourceforge.net/lists/listinfo/csound-devel >>> >> >> >> >> -- >> >> >> Andrés >> >> ------------------------------------------------------------------------------ >> Stay on top of everything new and different, both inside and >> around Java (TM) technology - register by April 22, and save >> $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. >> 300 plus technical and hands-on sessions. Register today. >> Use priority code J9JMT32. http://p.sf.net/sfu/p >> _______________________________________________ >> Csound-devel mailing list >> Csound-devel@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/csound-devel > > > ------------------------------------------------------------------------------ > Stay on top of everything new and different, both inside and > around Java (TM) technology - register by April 22, and save > $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. > 300 plus technical and hands-on sessions. Register today. > Use priority code J9JMT32. http://p.sf.net/sfu/p > _______________________________________________ > Csound-devel mailing list > Csound-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/csound-devel > -- Andrés ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Csound-devel mailing list Csound-devel@lists.sourceforge.net