View Full Version : just installed new hard drives now video jumpy

07-11-2003, 08:36 AM
after installing new scsi hd's when i capture video then play it back it very jumpy. any ideas.

07-12-2003, 07:24 PM
Could you supply a little more detail about your system, particulalry the configuration and type of the new drives, what sort of speeds they are providing, whether other drive (sets?) are providing the same performance they did previously, what the relevant settings in T[2]'s Prefs panel are set at, and whether the stuttering is always present, only with a certain number of layers, etc.

07-14-2003, 08:04 AM
i am very new on this system the hd's are set up as a raid 5. this is the first time i have used scsi hard drives.

07-14-2003, 08:07 AM
you wont be able to get the speed you need out of your drives with raid 5 you neeed to setup them up as a stripe set which is raid 0

07-14-2003, 09:05 AM
I agree with that. Add-on RAID boards generally haven't been up to par speed-wise by all accounts, where the much simpler and less expensive RAID level 0 stripeset performs admirably, albeit at the expense of a little security. If not your entire problem, this seems likely to be a major factor.

07-20-2003, 11:52 AM
RAID 5 writes data recovery info on each drive. As such, it is not useful for video applications. If you are using a model of RAID card that supports RAID 0 (striped) that's the solution.

I have used Adaptec RAID cards with RAID 0 succesfully. The advantage of using hardware RAID 0 rather than the OS is that system resources are not consummed to support the RAID function.

07-20-2003, 01:28 PM
I'm glad that's working for you. Strangely, though, over time a number of dealers have reported that add-on RAID cards, though as you have discovered *capable* of adequate speeds, don't seem to perform as well as a simple stripeset. I'd be quite interested in hearing your experience if you have a chance to experiment with the same driveset on a given system - with and without the hardware RAID.

07-21-2003, 02:50 PM
To add my 2cents worth, we've been running an Escalade 7850 IDE controller with 6 75g drives on a Supermicro P4DCE+ mobo. The Escalade is running in the 64bit pci slot.

Due to some problems with one of the drives possibly going bad, I have recently (meaning today) reconfigured the drives as all independant and striped them from Win2K. Previously, the drives were set-up as an array from the board level and then partitioned and formatted from Windows.

Running Disktest right at this moment, but the read speeds are averaging out somewhere around 150mb/s - comparable in speed to the board-level array config.

My reason for trying it this way has to do with feedback I've seen here in the message forums and wanted to see firsthand if it would give better speeds.

By-the-way, does anyone know how to config/run disktest without having it take forever to finish? Is there a way to configure it to run less numbers of cycles?

07-21-2003, 06:10 PM
If you just want a little quick feedback, simply double click on it - you don't have to capture to a file (if that's what you meant). You can exit it at any time.

BTW, my own limited experience with EIDE stripesets mirrors your own.

07-31-2003, 01:16 PM
Originally posted by SBowie
If you just want a little quick feedback, simply double click on it - you don't have to capture to a file (if that's what you meant). You can exit it at any time.

BTW, my own limited experience with EIDE stripesets mirrors your own.
Just to add my bit. The theory of hardware RAID is great; in practice, it never measured up. This was using the SCSI RAID boards, mostly Adaptec such as the 2100s. The problem seemed to be that these boards were primarily designed for RAID 1 and / or RAID 5 and switched between streams by one chip on the board. This chip turned out to be the weak link in the mix. I could never get the sustained throughput under load from the hardware RAID that I could from software. I'm now revisiting this entire issue with Serial ATA RAID. I will post results soon.
EIDE based RAID can give artifically high Drive test scores. Because of the load it puts on the CPU its results, in real life were always lower than the results from disktest. The problem is amplified by the lower 7200 RPM drive speed. EVen 8 meg CACHEs can't compensate when the drive becomes slightly fragmented. This is part of the reason IDE drives have higher Disk SEEK times than SCSI. I'm hoping that serial ATA with it's higher throughput and 4 independent data channels with dual 2GHZ or better processors will be able to overcome the disk seek issue and 7200 RPM disk speeds.