Multichannel, still an expert request for an optimal setting

[sisoffi]

hello, I still open a post on the multichannel so that someone can really indicate a solution.
It is unthinkable to have to invest thousands of dollars for a system where so few sterile indications in the learning center are such that they can not afford to continue with a system that is free from stuttering.

I have done research in months and I wrote also post about it to find solutions but to date, there is not yet a system that allows you to fly and not have problems of stuttering on the visuals of the clients.

First of all:

1. minimum system requirements for a multichannel network.

In a domestic system, current computers allow performance. I last i7 or Ryzen generation CPU with 8 and more GB graphics cards such that even in medium-high setup conditions the system should run really well. Let's leave out the Lan network that has been tested and no problems and the Windows10 64bit operating systems are minimal and well optimized. In the learning center is well indicated that:

"For best performance, the client machines should be used to synchronize the frame of the display with the appropriate display. may fall out-of-sync with another display adjoining This is especially probable more than another client renderer a series of the same work. the frame flip may be delayed If hardware synchronization is not possible ng noticeably out of sync. CPU bound, or to limit the frame rate of the Host to provide enough time for all clients to finish GPU work. "

Browsing in other posts, investigating the genlock cards, however, this answer, to my own doubt:

"If the multichannel system is configured for network synch, the host should wait to start the next frame until all slaves are done with the current frame. In this case, channels should never be more than 1 frame off from each other, but each screen will still refresh as soon as that system is done rendering its frame. Gen-lock would help guarantee that all monitors flip the frame buffer at the same time. That would reduce visual tearing but would not reduce stutters. If your systems are prone to occasional stutters with a single channel, the problem will be worse as you add channels because every time anyone channel stutters, they will all stutter together. If your system is going to be running significantly lower than your output's refresh rate and/or you plan to make rapid camera movements, gen-lock would be preferable."

and here my first doubt is not knowing whether to invest in unconventional video cards such as GeForce or AMD, but I should adopt GPU genlock Quadro or ATI pro. Which model suggests? need a very boost Quadro or ATI pro card?

(There is no video comparison from Prepar3D for a hypothetical example of Multichannel setup with also Hardware used?)

2. FPS Syncronization

I tried all kinds of experiments: first with normal monitors, then fullHD and now with 4k TV, which by the way, I lowered the hertz to 30 that would be the perfect thing for every client PC blocking and always keeping 30fps in vysnc and TB ON I would have perfect fluidity. perfect!
The speech falls if the host that hosts my panels for avionics and only one external view in minimal settings and fps unlimited with no vsync with monitors are at 60hertz.

The conclusion that all the system's visuals are at the same update frequency either at 30Hz or 60Hz (but even in this case with normal systems and not with genlock cards I have not yet found the perfect smooth).
The Network.cfg from host there is : FrameSyncRate=30 that I saw is the only thing that can be changed on these files. If I try to put the fps sync to 0 the systems are not synchronized. I noted momentary low fps to 15 on clients (half from 30) while host running in unlimited frames.

3. AllowSharedAircraft=1
What is the purpoe of this line? shared for why? to show the effects on clients like landing lights?
in my case I have enabled and installed the same payware plane for both clients in such a way as to have the landing lights effect as well. The problem to share aircraft is if I open gauges and panels from cockpits the same info will send to clients and have the gauge or panel open on clients :(

4. Purpose of the multichannel
Is it really necessary to use a multichannel system with a host PC with minimal settings where it could be used as an external view? in home cockpits, usually programs such as prosim or magenta project turn the avionics into a client without having another prepar3D in use at minimum settings.

What would then be the optimal choice for having an ideal multichannel setup?

many thank
Simone

[sisoffi]

update 21 August 2018:

Given that I have not the faintest idea yet of the type of computers to adopt because i dont understand if server host pc need to have cpu more power than clients or not. Currently I have in test:

SERVER (HOST) I7 6700k - 16gbRAM - GPU R9 390 + 2MONITOR 23 "1920x1080
CLIENT1 (NETWORK) I7 7700k - 16gbRAM - R580 -> now in test with PRO WX5100 genlock + 2 TV 4k
CLIENT2 (NETWORK) I7 700k - 16GB RAM - R580 + 2TV4k

I tested a genlock video card (Radeon PRO WX5100 8gb) on a client with only one test on a 4k TV.
Given that, from specifications:

https://www.amd.com/Documents/radeon-pr ... asheet.pdf
 the card in question supports up to 4 displays in 5k:

"Drive up to two, 5K (5120x2880 pixel resolution) dual-cable displays @ 60 Hz, or one, single-cable 5K display @ 60 Hz"


first problem:
1. the TV does not go over 1080p at 60hz.
With a normal video card (gtx1070 or Rx580) I got in the two TV views 4k to 30hz. So my tv in uhd 4k scale to 30hz? the specifications of the card indicate:

does it take a particular cable? if the 4k to 60hz is supported, it should be logical to support even lower hertz rollouts, so a cable problem?


Test results:

with a genlock card in a client finally the fluidity is almost total on gorund and in flight. Made a quick test with a scenario like Palermo jetstream that is very light and well optimized, perfect. But in a flight test Palermo Rome, approaching Rome returns the stutters.
Ok, could it be a scenario? therefore implies the fact that every addons must be optimized?