both those machines have i7 processors and then not so good graphic cards. The 240 is the weak side of the higher end cards. The 250 isn’t that much better. But both will run vvvv more than happy with trippleHead2go and do loads of cool stuff with it. It’s when using shaders like the SoftShadow or DOF that you’ll start to notice issues or very complex scenes with loads of xfiles/textures/antialising etc.
If you want to do really crazy crazy stuff with you’re GFX then get a nvidia 400 series. They rock and price/power is better than the 200 series. But with a good GFX card you’ll need a good processor whatever you’re doing really so i7 if you can afford it.
it’s not always that easy to get the most out of a great GFX card though so maybe if ya not fully into and well experienced with vvvv, just get something that runs it and then you’ll start to get a feel for what you’re power requirements are. Often the same scene will run on weak machine as it would on a powerful machine, you’d just have to lower the resolutions loads, and reduce spread counts.
but yeah I would say the processors in those machines you’ve linked to seem to be a bit over spec for the GFX cards they have with them. Generally better with vvvv to have high spec GFX and lower CPU.
reading the article, i would go for a i7 quad core because vvvv is a single thread application and will run only on one core. so clockspeed is still really important here. the other cores will help to decode videos or run other stuff besides vvvv. but having 5 free cores seems a bit much…
Well it might be a bit off-topic but I am kind of a fan of ffdshow tryouts for video-playback as it allows for multi-threaded decoding. So more CPU cores (should at least in theory and as far as my experience goes) result in a larger quantity of videos being played back without performance-issues.
Does video-decoding on the GPU influence performance of other stuff like shaders? This interview on Resolumes GPU-powered DXV codec suggests otherwise, but what is your experience?
Do the plugins generally run in different threads than the vvvv core or do only special plugins like Particles (Spreads) take advantage of multi-core CPUs?
On my current setup I have 2 video textures at the same time (@720p). They appear to use both cores of my Intel Core 2 Duo. So if I want to have 6 video textures running at the same time a six-core system doesn’t seem like overkill, right?
I started off decoding video via GPU but it was too limited and too slow. The amount of video memory + RAM required to decode just 100 frames of 720p video is ridiculous. The best solution I have found is to use a normal video texture and use a dedicated SSD drive for each video texture.
I’ve never noticed any performance decrease in shaders because of decoding video on the GPU. I’ve never really done any testing of this though, no need because I’ve always been happy. One example patch I had running in a club once; 3 x 1024x768, geometry corrected, soft edge blended, 4 layers of DVD quality MPEG-2 decoding on GPU, 1000 x fairly simple phong shaded xFiles using spreads, quite a few passes through various pixel shaders (mainly Blend.fx and blur) = 40-50fps.
multi-threaded decoding sounds like a good idea though if you have un-used cores. I’ll give it a go for sure one day.
What video card where you using when you’ve seen issues?
Dedicated SSD drive for each video texture? seems extreme. DVD quality is max 9.8 Mbit/s and 720 around double that, a SATAII drive can do 3000 Mbit/s so should easily handle a good few layers of video. I appreciate seek time needs to be included in any calculation and will have the biggest effect on this. I’ve seen no problem with 4 layers, maybe beyond this get another SATA drive.
Thinkpad T60p with ATI Mobility FireGL V5200. This card makes me never want to buy ATI again, BTW, because it causes constant BSOD (well-known ati2dvag problem) and vertical refresh tearing (despite “wait for vertical refresh” being enabled in CCC).
I have a lot of 20-200 frame video clips encoded with Indeo codec (key frame every frame etc) @ 720p. The clips get loaded in randomly, two are running at any given time. One problem with using the GPU was that loading the clip into texture memory caused a slight delay which happened every time a new clip was loaded.
You might be right that SSD is not so much faster than SATA for reading video but at only $85 for a 32gb drive it’s a no-brainer. And my laptop is getting hot enough another SATA drive might just fry it. Regardless of SATA or SSD, having separate drives makes a big difference because reading two videos simultaneously from the same drive is a huge slow down. (As far as I know there’s no way to background-read the compressed video file into RAM first.)
Can anyone tell me the best way to hook up a monitor and three projectors to a rig with a GTX 470? The specs list VGA, DVI, and HDMI ports separately from the video card and then two DVI ports on the video card. Nvidia’s website lists 2 dvi ports and one hdmi port for the card…
Could I hook up a VGA monitor to one of the ports and 3 projectors via DVI or HDMI using a TH2Go ?.. My concern with HDMI to projector is converting the signal from DVI… My concern with DVI to projector is the approximately 75 ft of distance from the PC to the projector.
Yes shure, but if I didn´t misunderstand there is still the 75ft/23 meter distance between the computer and the beamers. DVI only guarantees 5 meters without “signal degradation”, so I think there is a need to somehow boost the signal or use some sort of extender.