After nearly four years of using the Disk Read/Write test in the AJA System Test app to test raid throughput I still don't entirely understand the paradigm AJA uses, which is different from other throughput tests and far more complex.
Why for instance does choosing DV as the Video Frame Size generate substantially lower read/write times than 10-bit uncompressed HD? It seems to defy logic. Can someone please explain?
David Roth Weiss
David Weiss Productions, Inc.
POST-PRODUCTION WITHOUT THE USUAL INSANITY ™
A forum host of Creative COW's Apple Final Cut Pro, Business & Marketing, and Indie Film & Documentary forums.
This is not the case when I use the program. On the G4 Powerbook I'm on, I get generally the same read/write speeds (between 18-20MB/s) regardless of codec or write size.
So I can't replicate your issue, but based on what you're describing, perhaps the reason is because the program writes 1 frame of DV and then moves on to the next, writes one frame and moves on, so it never needs to "stress" itself and in general writes fewer MB's per second because it's writing fewer lines of resolution per second.
Why wouldn't the write times be lower for writing out 525 lines vs. 1080 lines?
[Jesus Ali]" On the G4 Powerbook I'm on, I get generally the same read/write speeds (between 18-20MB/s) regardless of codec or write size."
Thats because your IDE drive maxes out at that read/write speed. ( in theory the max capable on your drive is 25MB/s- Sata drives at 10K and up can reach sustained speeds of about 80-85 Mb/s on a single drive spindle
Davids comments are accurate when you have a high speed array that has the ability to cache the data as fast as needed for the UC formats. On my fibre and sata arrays the uncompressed read / write can be as much as 20% faster than the compressed formats- and the faster the array - the greater the difference between the speeds.
[David Roth Weiss]"Why for instance does choosing DV as the Video Frame Size generate substantially lower read/write times than 10-bit uncompressed HD? It seems to defy logic. Can someone please explain? "
if you notice - the only compressed formats listed are for DV and DVCPROHD,
I am going to reach out on the limb but if you turn off the checkbox "disable file systems cache"
you see that the write speed for DV and DVCPROHD start looking the same - ( the reads seem pretty whacked though)
I am thinking that it has to do with the additional processing that it takes the compressed files to be expanded/opened/ extracted out to baseband for playback. I arrived at this since all of the uncompressed content does not require this at all, whereas the compressed formats do.
I pinged the AJA guys for something more than my guess.