I have been interested for some time now in integrating the strengths of three Operating Systems, MAC OSX, Windows 2000 and Linux Red Hat, into a “super computing” environment to enhance my approach to music composition. Being an independent composer I don’t have the resources to upgrade my computers as often as I would like so I also view this notion as a way to extend the life of my systems. Recently, I arrived at a way to accomplish two thirds of this approach, which is working very well and thought it might be of interest and use to some of you.
First of all let me detail the machines I am using. The Windows 2000 sp2 machine is a powerhouse if not a speed demon by today’s standards. It has a Super Micro server motherboard, dual 600 MHz Katamai processors, a dual channel SCSI controller, 500mb RAM, 60gb ATA hard drive, two 9gb Cheetah 10,000 rpm SCSI hard drives, SCSI CD-R and CDrom and a Card D plus w/digital I/O professional quality sound card. The main apps I use are Sonar Professional, Sound Forge, Red Roaster 24 bit, Excel XP and of course Csound and Cmask. As an example of the power of this system I can run 60 – 80 tracks of digital audio at one time in Sonar.
The Linux Red Hat 7.2 machine is very fast. It has an Abit motherboard, AMD 1200 processor, 500mb of RAM, 20gb ATA hard drive and a SoundBlaster Awe 64 sound card. The main apps I run on it are Csound and Cecilia.
I did an A – B comparison of these two systems to see the differences in the time it would take to render a complex Csound file. As expected just by virtue of the processor speed the Linux box blew away the Windows box. What was surprising was the amount of difference between the two. All things being equal one would expect the Linux box to be somewhere around twice as fast as the Windows box. Instead it was nearly four times as fast.
Seeing that, I knew I had to work with Linux when rendering my Csound files. The trouble was that I need professional sound quality to monitor the subtleties of my compositions as they evolve. I know that the RME Hammerfall sounds card are supposed to be very good and are Linux compatible but at the time I cannot afford to get a new sound card. Also, I use Excel to generate and manipulate my scores and did not want to give up working in that way. I did not want to be jumping between two systems while composing either. Here is what I came up with:
First of all I installed 100mbps NICs (network interface card) on both systems and plugged them both into a 100mbps hub. Next, on the Linux box, I set up Samba. This enables file sharing between the Windows box and the Linux box. (Incidentally, OSX is Samba capable too.) This let me see the Linux hard drive and all its files in Windows Explorer. Also I enabled SSH, which is secure telnet. On the Windows box I installed Putty, a freeware Telnet/SSH app that works very well.
Now when I am working I have the following windows open on my windows machine:
Excel – for Score file generation
Programmer’s File Editor – for orchestra generation
Windows Explorer – open to the directory where my sound files will be written to on the Linux box.
A SSH session connected to the Linux box.
Here is my process: I write my .sco and .orc files on the Windows box but save them to the Linux box. Then I run Csound from the command line on the Linux box using SSH from the Windows machine. When the sound file has been rendered on the Linux box I double click it in Windows Explorer and it plays across the network on my Card D plus sound card on the Windows box.
I now have the incredible speed of Linux to render my Csound files, saving much time waiting for the results of my work, but also have the applications and sound quality offered by my Windows box. This is truly, the best of both worlds. I have found it to be a very simple and beneficial way to work.
Now if I can just come up with a way to integrate the MAC into this process… I am expecting it to involve MAX/MSP in some way.
Take care and Happy Csounding!!