Patrick Flanagan

I've built a prototype traveling immersive space in a trailer that simulates submarine voyages for people with limited/no access to the ocean (www.oceanlab.org). It features seven 4K TVs as "digital portholes" playing custom-recorded video from coral reefs, the deep sea, etc. I'm looking for recommendations on hardware to help it run smoothly.

For the initial build, I've been playing pre-recorded video on 5-minute loops, powered by one central PC. I'm new to Max, so for the moment I've been playing the video in multiple VLC windows while I learn to program it all in Jitter. I've found that the video starts to get laggy/choppy once I open more than four or five streams, and that's only playing 5-minute loops of 1080p video.

As I'd love to eventually get seven synchronized 4K streams running smoothly, I'm looking for advice on what kind of hardware I'll need. For ease of programming/wiring, I'd prefer to run everything from one computer instead of networking several together.

My current PC is a PowerSpec B728, with an i7-6700 3.40Ghz processor, a 480GB SSD, and 32GB DDR4-2133 RAM (upgraded from 16GB) - this motherboard maxes out at 32GB, so if the answer is more RAM, it'll have to be with a whole new computer. I've added an NVidia NVS810 video card with 8 mini-DP outputs and 4GB of VRAM.

I've built a prototype traveling immersive space in a trailer that simulates submarine voyages for people with limited/no access to the ocean (www.oceanlab.org).  It features seven 4K TVs as "digital portholes" playing custom-recorded video from coral reefs, the deep sea, etc. I'm looking for recommendations on hardware to help it run smoothly.

For the initial build, I've been playing pre-recorded video on 5-minute loops,  powered by one central PC. I'm new to Max, so for the moment I've been playing the video in multiple VLC windows while I learn to program it all in Jitter.  I've found that the video starts to get laggy/choppy once I open more than four or five streams, and that's only playing 5-minute loops of 1080p video. 

I've built a prototype traveling immersive space in a trailer that simulates submarine voyages for people with limited/no access to the ocea

hardware-recommendation-for-running-seven-4k-displays

as you begin your patching, first of all make sure you follow these best practices article: https://cycling74.com/tutorials/best-practices-in-jitter-part-1

I'd also suggest you convert your videos with the HAP codec and use the hap engine in your jit.movie objects, along with the @output_texture 1 attribute, so you'll be using OpenGL in the whole process (which is faster).

Also, the main output will be one big window spread across your seven screens, so you'll have to create 7 jit.videoplane objects which you'll locate in the different screens using the @position attribute.

Once you have this up and running, you can test if your system can handle this setup (one possible bottleneck could be also the reading speed from disk.

Hi, 
as you begin your patching, first of all make sure you follow these best practices article: https://cycling74.com/tutorials/best-practices-in-jitter-part-1

I'd also suggest you convert your videos with the HAP codec and use the hap engine in your jit.movie objects, along with the @output_texture 1 attribute, so you'll be using OpenGL in the whole process (which is faster).

Also, the main output will be one big window spread across your seven screens, so you'll have to create 7 jit.videoplane objects which you'll locate in the different screens using the @position attribute.

Once you have this up and running, you can test if your system can handle this setup (one possible bottleneck could be also the reading speed from disk.





I was able to put together a working patcher using your suggestions, and indeed, following this puts a lot more of the load on the GPU than the CPU. This seems to work smoothly with five, and I'll try moving up to 2.7K and 4K soon.

However, when the video files are read into the jit.world window, they come out squished into a more square format (they should each be 16:9). I've tried adding @dim or related attributes into various places, but they don't seem to work. What would you suggest?

Thanks LSKA!
I was able to put together a working patcher using your suggestions, and indeed, following this puts a lot more of the load on the GPU than the CPU. This seems to work smoothly with five, and I'll try moving up to 2.7K and 4K soon.

However, when the video files are read into the jit.world window, they come out squished into a more square format (they should each be 16:9). I've tried adding @dim or related attributes into various places, but they don't seem to work. What would you suggest?

https://www.matrox.com/graphics/de/products/graphics_cards/c-series/c900/

https://www.matrox.com/graphics/de/products/graphics_cards/c-series/c680/

9x WUXGA
https://www.matrox.com/graphics/de/products/graphics_cards/c-series/c900/
windows only

6x 4k
https://www.matrox.com/graphics/de/products/graphics_cards/c-series/c680/
windows & linux


@Roman the NVS 810 I’ve installed powers 8x 4K, I’m wondering more about the processor/RAM/disks

@LSKA Thank you! I was able to put together a working patcher using your suggestions, and indeed, following this puts a lot more of the load on the GPU than the CPU.

FS.PNG

Got it playing 7 simultaneous hap-encoded videos (from 5 1080p video files - two are duplicates), but now each video plays at about 2 fps. Is there a way to optimize my patcher/files more than this?

I think you have to use '@engine hap' attribute?

Hardware recommendation for running seven 4K displays