That's peculiar, esp. since the app caches frames into memory during prefetching, so it's usually less effort to play the same stream twice than it is to play two separate streams.
Can you provide any details about the source files you're using? Does it happen with only specific formats? Does it only happen on the first attempt to play that timeline area?
Care to comment on the Hack floating around the web for other using Nvidia cards for MPE that are NOT on the list of "official" cards?
I know, you would think that it would much easier but apparently thats not the case.
It does it with AVCHD files from both a Canon HG-20 and a Panasonic AG-HMC40 doesn't seem to matter the compression settings or frame rate. It does it whenever it gets to a part where the files overlap and it basically freezes the playback untill the CTI is moved off that section.
1 person found this helpful
I'm thinking it's probably an AVCHD specific issue then. I'll check around & see what I can find out.
@Chris - Gee, there's a simple text file that allows you to unlock non-supported CUDA cards? How cool is that...?
Arg, i didn't really want to have to convert all that footage into something else. oh well I will have to play with it some more.
Well don't I feel sucker punched, I purchased that 4800 yesterday. Oh well, for the moment Im not about to base my business off adobe not fixing the hack.
Don't feel ripped off - you've bought a solid card that has been thoroughly tested & is supported to work correctly with MPE. Anybody using a hack is doing just that - hacking. It's like people overclocking CPUs - if you burn out your hardware or have inexplicable rendering errors or lockups, you're on your own.
For instance, currently the Fermi chipsets have some rendering errors that actually require bug fixes in an upcoming dot before they'll work correctly. So, to reiterate, it's not as simple as 'CUDA works everywhere' with every chipset. There's a reason we're validating boards...
Yeah I know, but the cost of the card it still stings, especially when the same retail card is $209 vs $1400. Well I'll Try transcoding my files into a half a dozen different supported HD formats and see where I get. I'll post my findings.
1 person found this helpful
It looks like we do have a bug that's specific to long-GOP prefetching. There's been a recent bug fix to the source frame prefetcher, and the bug sounds like the same description (cross-dissolves on 7D H.264 files playing back choppy). So I'm betting this is the same issue, and it'll be addressed in our first upcoming dot.
I'm using the 4800 hundred and initially had the same problems.
Turned out it was the codec making MPE hang. As soon as I imported a supported format, it was smooth as silk with multiple layers and real-time effects.
Might be worth double checking the codec you're using.
Thanks for that report.
I guess I will just have to wait then.
no biggy, my project isn't shot till the end of the month.
It is cool - but I do not think it is cool if it turns out that the FX 3800 I have ordered does not perform better than my "softmodded" GTX 260 (using the txt-file mod). It seems to work - I can play back 4-5 layers smoothly with transitions, effects etc. AVCHD 720 50P. So could you please give me some info about excactly what else - besides from the txt-file - you have done to test and make the FX 3800 running more stable than GTX 260 - wich is - as I am told in this forum - almost the same GPU and specs inside - just looking different from the outside - but FX 3800 costs 5-6 times the GTX 260.
I feel a bit like a fool right now because - for a non-programmer like me - it looks as if the trick is just to change a line in a textfile and VOILA - you card is supported. Is it as simple as that? Or what else have you been doing inside PPro (in headlines) to make the FX/Quadro running better more stable? Will I gain more and better performance using the FX 3800 compared to the softmodded GTX 260? If the FX3800 will surely and with garantee will give me better and more stable performance compared to GTX 260 - I wil feel OK about the investment, but if it turns out that I might as well have kept the GTX 260 and softmodded it - I will feel like being cheated to buy a card that basicly are the same - gaining nothing but a empty pocket.
I´m sorry for my bad english and long explanations - I just try to clearify what I mean....hope you understand what I am asking about?
Thx in advance - all the best.
Let's not high-jack this thread to talk more about MPE & CUDA boards. I suggest that we move that part of the discussion to the already existing thread...