Printable View
Brand new system from Highend. Mac Pro 2.66 with SCSI drive etc....
QuickTime clips about a minute or two long, 1024 x 576, Photo JPEG 75% or 50%, with 48 khz audio.
And in the middle of several clips, the frame rate stutters. And we are talking one clip playing on one layer...
Why?
I´m in another city now, so I don´t have the system in front of me anymore. But it´s a brand new system running version 4 point something. Probably not the latest build, but that should not matter.
there is a lot of difference between photjpeg 75 and 50 and whenever you do anything that isnt standard def - the whole game changes. things are much more critical.Quote:
Originally Posted by ping141
if you dont have any history with doing anything with digital video you will be in for a sharp learning experience- you will need to learn what works and what doesnt for whatever your system is.
most things do not work. you have to find something that works for what you need to do.
there are only a few things that do. its not the other way round.
what did you use photojpeg 75 or 50?
and are you trying to play audio at the same time?
I have 10 years video expericene from high end post production to 360 x 288 VDMX vj´ing, so digital video and codecs are not new to me.
It seems that in the middle of certain clips that are a minute or two long, the frame rate will stutter a bit and then play smooth again. The clips that I have noticed this with were originally in 75% Photo JPEG with 48khz stereo audio. Then I lowered them to 50% in order see if it helped and it didn´t. It´s not a major thing, this show is up and running every night, and I´m probably the only person noticing this problem, but still. On a 2.66 Mac Pro with a SCSI drive and an x1900 card, anything should in reality play fine, at least one stream on these resolutions. They play flawlessly on my laptop in QTP from the internal drive btw. It might be something funny with my clips, but I can´t see what it could be.
And in order to track what plays well and what doen not it would be great to have a FPS history in the Catalyst where you could see and track when the frame rate dips. Also, the FPS indicator in the Layers window does not work when playing with audio.
when you play with audio - playback is handled totally by quicktime.
so whatever happens with what you see if something quicktime is doing.
and audio takes priority.
quicktime player doesnt have to handle lots of layers and lots of fx...so it doesnt have so much to do. nor does it have to handle interfaces to lots of equipment like video input simulaneously.
---
what size and frequency are all your display screens set to?
what is your screen setup?
do you see this when you dont have audio?
can you send me your pdf profile
Using both outputs at 1024 x 768, 75 hz. I don´t use video input or anything else.
On one output the files are 1024 x 576, on the other 1024 x 768. This due to one screen being 16:9 and the beamer is squeezing the frame.
Problem files have been on both these outputs.
I´ll see if I can get the profile sent over.
so you are playing back 2 files at the same time?
No. Only one file at a time. But I have several troublesome clips.
It could be that the jump happens when the drives have there caches flushed by the system. I know with the Xserve RAID there is an option to disable this so the cache is manged by the hardware controller and not the Operating system. The difference is quite amazing. I don't know if this is an option with your SCSI drive (I don't use one) but I have noticed this same effect when playing back longer files off Raptors.
Unfortunately I think Quicktime controls all this stuff so it is out of our mere mortals control.
This is just a guess so it could still be another issue. I am not that familiar with SCSI set ups.
Cheers
Toby
That could be an explanation, actually. But QuickTime itself plays everything very nice, at least in the QTP, so it´s a little strange that it will make trouble here. But it seems to happen to my long clips.
This is a spanking new system from Highend and it came with an ATTO PCI card and a SCSI drive for contents.
The thing is that Quicktime player and any editing package knows what is coming next so it can sort out it's caching and optimise playback.
Catalyst is doing things far more on the fly and at random so requires raw horsepower to do the tricks.
Again, an educated guess. I am sure others could help clarify.
Cheers
Toby
Then again, if 2.66 Ghz and an x1900 is not raw horsepower, then what is? :)
a 100 gig Ram Disk:rolleyes:
To get back to how you are playing things.... Are you selecting the same video clip on two different layers to get them to come out of the two different outputs?
Seriously though, what I mean is that when the gang of people who write the operating system and quicktime do there thing, they are trying to optimise it for 95 percent of the users. There is a lot of components in both hardware and software that have to interact flawlessly each time to make it all work.
I think that despite how powerful the current hardware has become there are still little tricks and compromises being made to make things look smooth. My guess would be that Catalyst playback performance is not high on their list of priorities. In the ideal world quicktime is tailored just to our needs and they work out those little tricks in our favour with regard to high speed random drive playback. However in the overall scheme of things we are a tiny user base doing something outside the norm.
I guess my point is this, there is so much going on between Catalyst and the hardware that is out of Richard's hands and the hardware manufacturers that just having the latest kit doesn't necessarily mean it will all work smoothly all of the time. We are probably the only group testing it under these conditions and some of it is just unknown territory. Software developers and Hardware manufacturers just don't get the time to extensively test every scenario. They just have to go for what keeps most of the people happy most of the time.
Cheers
Toby
... People who are really serious about software should make their own hardware. ...
Alan Kay .
but in that case because of the little very little lighting industry a Catalyst will cost a huge amount of money .
I think there a not other application that play 12 full resolutions movies at the same time in the market .
I agree with Toby
E
the computer isnt the main limit - its the disc.Quote:
Originally Posted by ping141
--
can you send me your files?Quote:
Originally Posted by ping141
Nope. I play one clip at a time on one layer only, and I have had several clips that don´t play smooth.Quote:
Originally Posted by FxDrew
Quote:
Originally Posted by samsc
But in this case it is a Highend delivered SCSI drive delivering (according to the Decklink test software) about 113 mb/s both in and out.
And the clips I´ve had problems with are 146 mb over 1 minute and 8 seconds. The other I have before me is 60 mb over 2 minutes and 9 seconds.
To my understanding, this cannot push any limits in the world...
decklink software doesnt test performance of catalyst.
it only tests bulk block read and writing- not the critical random access reading and writing.
headline data transfer rate is useless for benchmarking.
You can only test catalyst with random read tests, where file can be read randomly from any sector on the disc.
as tests on barefeats ( other web sites ) and other systems show time and time again -
http://www.barefeats.com/quad07.html
they show very fast headline data transfer rates - in this case 200MBytes/second.
but lamentable random access transfer.
and i quote directly -
A raid is NOT faster than a single disc for random access- unless you have huge files.Quote:
The remaining performance issue is slow small random read speeds for one, two, three or four drives. No matter how many drives you configure in RAID 0 sets, the average random read speed for combined block sizes from 64K to 1024K is less than 30MB/s!
What is important for catalyst is the disc seek time - not the data transfer rate.
So a single fast scsi disc ( and really a 15k drive is essential for lots of layers ) -will be faster than any standard raid made up from slow ATA or SATA disc - ( unless they use aggressive caching like an xserve raid )
you need to check out the random readwrite tests on storagereview.com for any kindof performance evaluation.