best to install /temp on my raid0 "partition" for video editing; space limit. for /?
 

I only have experience in audio editing/recording and want to learn basic video editing. I have to (re-)install Hardy U. Studio LTS and find that raid0 could be usefull for me.

I have 11 GiGs space left on fast enough SCSI (XP is still needed), plan on installing / there. A generous friend just gave me a 36 GIG 10k Ultrastar IBM SCSI (2001), that operates at slightly lower access/specs, which I wanted to keep 2 GIG for swap, and ~ 8 GIG for /home. I wanted to use 25 GIGS for my raid array and another 25 from an extended partition in my recent 500 GIG IDE drive. I boot from a separate primary partition on same drive, as well, I have other OS and storage partitions.

I heard that /temp files can swell depending on project, and if I don't want to create a LVD within my raid partition, can I send my rendered files to a link from within the /temp partition and play them from same partition and only send final results to my other non raid partitions on IDE afterwards, while I start to work on a new edit. Or do I have the whole process wrong!?


Does cinelerra work with temp files and then writes to the data target partition or does the file get processed straight to a data partition?


Here is an excerpt from the Cinelerra doc.:


"Cinelerra is demanding on all PC subsystems, as reading, decoding and playing video can be quite taxing. Thus, performance and usability of Cinelerra are directly proportional to the video format (SVCD/DV/HDV/HD/etc) used and the CPU and I/O bus speeds and video and memory bus architecture of your hardware. Therefore, it stands to reason that a less powerful system will be sufficient for users working with audio only or lower resolution video formats. However, that same system may slow down considerably when playing back a higher resolution format, such as DV video. Effects and several tracks of audio will compound these problems. Given these constraints, here are some suggestions for running Cinelerra:

* CPU speed
At least 500 MHz CPU speed, anything less would be useless. Dual-core and SMP processors greatly improve Cinelerra speed.
* Memory
When working with video, a large amount of free memory available can help speed up operations by avoiding unnecessary disk swaps and keeping material ready accessible. Have at least 256 Megabytes of memory. To really use Cinelerra for higher resolution video formats and larger projects, greater than 1 Gb memory space is suggested.
* Storage
Video editing can be quite I/O intensive. Storage requirements are based on your particular video editing needs. If you expect to produce long pieces in uncompressed or larger resolution formats, you should have large (>200 Gb) and fast (<10ms) disk drives. For example, DV uses about 3.5 Megs per second, or 12 Gigs per hour. For smaller projects you might get away with 1 Gb. RAID0 (stripe set), RAID1+0 (striped/mirrored) or RAID5 (stripe set with parity) will also speed playback
* Video adapters
Since version 2.1, Cinelerra benefits from OpenGL hardware acceleration. Make sure the video card you use supports OpenGL 2.0 in order to benefit from that acceleration. Nvidia series 7 (ie. 7600GS) are known to work well. Unfortunately, ATI's Linux drivers do not support a complete implementation of OpenGL 2.0. If you are going to send a composite signal directly to a TV or video recorder, make sure your video card supports it.
* Multiple monitors
You can use XFree86's Xinerama features to work on multiple monitor heads. This feature can be a very effective way of increasing productivity."

My video card does not even support open GL 2.0, only GL 1.0, so that means this will slow things down at the outset! My 2 2.4 GIG CPUs are ok, 1 GIG memory just ok. My nv quadro4 900XGL videocard enables 2 monitors at least! I have a 2002 HP workstation, with no more room or budget for extra drives or hardware...

Any ideas on better setup for video editing, will be much appreciated!
If anyone knows of any more linux video production (forum or other) links (besides Ubuntu), I'd appreciate...