wikiC/2.06
Main page
Alphabetic index
Recent Edits

jMax-english

jMax was intended to be a platform independent follow-up to the IRCAM workstation or ISPW, that ran on the Next machine with the DSP-boards. After Next stopped making the Cube in 1993/94 IRCAM decided not to be stuck on one platform. Also because general CPU's became more capable of doing the kind of number-crunching that were normally associated with DSP, they redesigned FTS and tried to do all the client stuff with a java gui (1). This also reflected more or less the approach taken previously with the Sogitec 4X and the ISPW.

During the period between 1998 to 2001 it was used for internal performances and productions, and mostly served as a platform for software development, like score following. It ran mainly on Irix/Silicon Graphics machines, because at that time these provided the features that were needed: superior processing, real time performance, reliability and stereo and multichannel digital audio I/O with the ADAT optical interface, using existing or additional hardware in a modular way [According to a press release Friday September 25th, 1998]. IRCAM wanted a 'virtual ISPW' and found this at that time in SGI machines with multiple cpu's and the OS (Irix) that offered, through the REACT extensions, an adequate realtime response (under 20 microseconds...). Therefore, for example the controlling host became a Java-gui on cpu 1 (or as in the picture above on a separate LinuxPC), the signal processing on cpu 2 and the spatialization on cpu 3. Under Irix these functions can be assigned to different processors, because REACT allows this clear separation of responses to work load requests:
http://www.sgi.com/products/software/react/features.html

It was used for concerts on SGI Octane workstations and on SGI Origin servers with SGI o2 and generic LinuxPC (PentiumPro/P2&P3) clients for studio-work. There is a nice description of a SGI Origin multi-processor setup running jMax at the University of York: http://music.york.ac.uk/rimm/tech.htm

and showing what it is/was capable of: http://music.york.ac.uk/rimm/spatmod.htm

Because of the people working and being taught at IRCAM, there were several users and developers involved with jMax. However increasingly it was abandoned in favor of Max/MSP during concerts. A major setback was the change from jMax 2.5 to version 4.0. Which meant that a lot of the applications had to significantly be reworked again:
http://jmax.sourceforge.net/packages.html#porting

Also the rather exuberant demands, even in its most humblest form, of one cpu for the graphic-client and one cpu for the dsp-server, placed on the computer by jMax in order to run smoothly, might have played an important role in its demise. Particularly with a graphic user interface that was rather crash prone and had a latency between 50/100 ms, which one can hardly call responsive. Still, it remains an impressive and ambitious design, with its different layers for User Interface Design in Java, Process Development with Scheme/Lisp and Unit Generators building in C. It also makes fully use of network technology for its functioning, which is now prevalent around many places in the world. As such it is a pity that it is disbanded by IRCAM.

Since 2002 all the development of libraries for jMax was transfered to Max/MSP. In particular the FTM-library (see:http://ftm.ircam.fr), which contains most of the significant elements of the Java editors and user interfaces and through the mxj object, is able to talk with the rest of Max/MSP (2). The multiple cpu-setup of jMax that was used for processing and spatialization, is more or less emulated by using network protocols for communication with other machines. jMax is now not supported anymore by IRCAM and peacefully laid to rest at: http://sourceforge.net/projects/jmax/

For historical purposes:
http://jmax.sourceforge.net/

Arie van Schutterhoef
(thanking Norbert Schnell, Laurent Ghys and Andrew Gerzso for the additional information)

See also: IRCAM-hardware.

REFERENCES:

jMax: a new JAVA-based editing and control system for real-time musical applications.-Francois Dechelle, Riccardo Borghesi, Maurizio De Cecco, Enzo Maggi, Butch Rovan et Norbert Schnell; ICMC:International Computer Music Conference, Octobre 1998.
FTM: Complex data structures for Max.-Norbert Schnell, Riccardo Borghesi, Diemo Schwarz, Frédéric Bevilacqua, Remy Müller; International Computer Music Conference, septembre 2005.