Csound Csound-dev Csound-tekno Search About

[Csnd] Version 5.14 now tagged

Date2011-10-12 11:11
Fromjohn ffitch
Subject[Csnd] Version 5.14 now tagged
and sources (as 5.14.1) being transferred to Sourceforge as I type.
release Notes below
==John ffitch
------------------------------------------------------------------------
Notes for 5.14
==============

It has been quite a delay since the last release.  As well as many
bug-fixes this release has many new opcodes etc.

However the BIG change is that the new parser is now built for normal
systems and is believed to be complete.  The old parser is still
enabled by default but more experienced users might like to try the
--new-parser option so we can nail any remaining problems.

The other big but still somewhat experimental change is that if the
new parser is used the multi-core mechanism is available.  Use of a
"-j 4" in the options will use 4 threads.  This is semantically
correct as long as there is no use of zak or channel operations
between instruments, or dynamic writing of tables. These will be fixed
soon (we understand and have a scheme, but it will take time).
Multi-core should give improvements in performance when ksamps is
large, and/or when instruments are computationally heavy.  The
developers will be addressing ways of improving it over the next
period, but it should e used with care -- but we need feedback.

Also, still somewhat experimental, there are now t-variables which are
vectors of k-vars.   They are allocated space via init
        t1 init 10
allocates t1[0] through t1[10], initialised to zero.  
        t2 init 20, 1
allocates t2[0] through t2[20], initialised to 1
Individual elements can be assigned via
        t1[kvar] = ...expression...
and all expressions can contain t1[kexpr]
It i spossible to copy a table with
        t1 = t2
At present there are no global t variables (use ftables)

New Opcodes:
    fareyseq opcode was missing in some builds
    mp3len to give length of an mp3 file
    qnan and qinf to check if values NaN (not a number) or infinite
    exprandi opcodes for interpolating version of exprnd
    cauchyi opcodes for interpolating version of cauchy
    gaussi opcodes for interpolating version of gauss
    cpumeter opcode to read process load
    linsegb, expsegb, transegb, expsegba (like XXXseg but with absolute points)
    pvsgain
    pvsbufread2
    serial
    lua opcodes
    plustab, multtab, maxtab, mintab, sumtab, scalet -- for t-variables

New Gen and Macros:
    fareygen table gen was missing in some builds

Modified Opcodes and Gens:
    loris opcodes updated to latest version
    date opcode has new base in some platforms to avoid overflow
    pvsblur now works over reinit    
    diskin, diskin2 and soundin now can read up to 40 channels
    prints behaves better with rounding
    fmpercfl now has working vibrato
    atreson now has gain parameter at k-rate
    comb opcode safe if in and out arguments the same
    better accuracy in line and expon
    OSCsend recovers space previously lost
    OSCsend can send a table as a blob with the T tag -- experimental
        and untested.
    lpf18 now has an optional iskip argument
    i() will also accept an i-rate value in which case it is a no-op


Utilities
    makecsd revised and extended to have options for MIDI and score 
            processing and licenses
    beats processor renamed to csbeats and distributed
    lpanal reworked to remove bugs and oddities
    mkdb utility to provide a catalogue of plugin libraries/opcodes


Bugs fixed:
    reading values to fill tables was broken wrt comments
    internal error in wii_data fixed
    pvsshift fixed
    jacko fixed
    gen23 minor fixes
    wiimote fixed
    atsaadd fixed
    compress fixed to work with 0dbfs
    pvsbufread corrected wrt position counting
    tempo opcode fixed
     had a bug, now fixed
    deferred gen01 tables could have wrong size
    vbap_zak made to work(!)
    fixed memory issue in ATSsinoi
    various fixes to cscore
    various fixes to partials and tradsyn
    transegr could crash

System Changes:
    ladspa library build in default system
    macros are now expanded inside string in the score
    an issue with noise in alsa fixed
        and a click in portaudio fixed    
    there in an until .. do .. od looping syntax (in the new parser only)
    SIGPIPE signals are ignored rather than causing Csound to exit
    portaudio driver changed to be more robust on stop/exit

API:
   None

Internal:
   As ever there have been many tweaks and code improvements; most of
   the other internal changes have been related to the new parser and
   multicore activities.






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"

Date2011-10-12 11:47
Fromluis jure
SubjectRe: [Csnd] Version 5.14 now tagged

congratulations john and all the developers for this new release! i'll try
to build and try it as soon as possible. i really want to test the
multi-core mode!


>Multi-core should give improvements in performance when ksamps is
>large, 

i guess you mean "when ksamps is small" (that is, when kr is large)? that
would make more sense to me.


>        t1 init 10
>allocates t1[0] through t1[10], initialised to zero.  

this means that the syntax init 10 creates a vector of 11 values?


best,

lj


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"

Date2011-10-12 12:51
FromVictor Lazzarini
SubjectRe: [Csnd] Version 5.14 now tagged
No, he means when ksmps is large. This is because ksmps affect  
granularity and improvements increase with granularity.

Victor
On 12 Oct 2011, at 11:47, luis jure wrote:

>
>
> congratulations john and all the developers for this new release!  
> i'll try
> to build and try it as soon as possible. i really want to test the
> multi-core mode!
>
>
>> Multi-core should give improvements in performance when ksamps is
>> large,
>
> i guess you mean "when ksamps is small" (that is, when kr is large)?  
> that
> would make more sense to me.
>
>
>>       t1 init 10
>> allocates t1[0] through t1[10], initialised to zero.
>
> this means that the syntax init 10 creates a vector of 11 values?
>
>
> best,
>
> lj
>
>
> 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"
>

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





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"

Date2011-10-12 13:18
Fromluis jure
SubjectRe: [Csnd] Version 5.14 now tagged
on 2011-10-12 at 12:51 Victor Lazzarini wrote:

>No, he means when ksmps is large. This is because ksmps affect  
>granularity and improvements increase with granularity.

ah, i see... it seems pretty obvious now :-)


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"

Date2011-10-15 16:38
FromLouis Cohen
SubjectRe: [Csnd] Version 5.14 now tagged
John,

This is great news for me. Thanks so much for implementing this  
feature! I look forward to being able to use global versions of t- 
variables, and - dare I mention - multi-dimensional t variables such  
as t4[k1, k2, k3]

best,
Lou Cohen


On Oct 12, 2011, at 6:11 AM, john ffitch wrote:

> Also, still somewhat experimental, there are now t-variables which are
> vectors of k-vars.   They are allocated space via init
>        t1 init 10
> allocates t1[0] through t1[10], initialised to zero.
>        t2 init 20, 1
> allocates t2[0] through t2[20], initialised to 1
> Individual elements can be assigned via
>        t1[kvar] = ...expression...
> and all expressions can contain t1[kexpr]
> It i spossible to copy a table with
>        t1 = t2
> At present there are no global t variables (use ftables)



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"