Ultra Low Latency Built

[LaSardine]

Hello all,

 

I am trying to built a machine with the lowest latency possible for a professional application.

I don't need an especially powerful rig, I am not going to game on it, I just need to make a selection of components that will be the fastest to carry an information coming from a fiber network connection straight to a display. 

I'd love to have your suggestions for each part, and ideas on where the latency bottlenecks are:

 

CPU=

MOTHERBOARD=

MEMORY=

SSD=

GRAPHICS CARD=

MONITOR=

NIC=

ROUTER=

NETWORK CABLES (fiber?)=

 

Thanks in advance for your help!

[emosun]

21 minutes ago, LaSardine said:

for a professional application

whats the application , stocks?

[LaSardine]

47 minutes ago, emosun said:

whats the application , stocks?

No it's an aerospace application (real time flight data analysis)

[emosun]

Just now, LaSardine said:

No it's an aerospace application (real time flight data analysis)

does the information get processed on the computer or is it already ready to go and just being displayed by the machine?

It would help to know more about what you're doing because (i don't mean this sarcastically) any computer can display information that it takes in from a network faster than it can take in the information itself (even with a fiber line). Essentially the lag from the infrastructure leading to your machine is greater than any lag from the machine to the display.

[LaSardine]

2 minutes ago, emosun said:

does the information get processed on the computer or is it already ready to go and just being displayed by the machine?

It would help to know more about what you're doing because (i don't mean this sarcastically) any computer can display information that it takes in from a network faster than it can take in the information itself (even with a fiber line). Essentially the lag from the infrastructure leading to your machine is greater than any lag from the machine to the display.

The information is already processed by another machine on the network, my application is very much like a web browser displaying multiple data streams coming from that server. There's only just 2D graphics rendering involved on my machine, that's it.

 

What I am trying to achieve here is to get the lowest latency possible between the data stream coming through the fiber and my eyes

[emosun]

Ok then essentially your bottleneck will be the data stream then. With your computer only displaying the data it can stream in the refresh rate of the data will only be limited by the server feeding your machine.

Basically , if you have a fiber connection and are just doing data streaming then a 300$ computer will update just as often as a 3000$ computer if there's no on site processing that has to be done. 

I only know very little about flight data but from what i've seen of simple position data the stream tends to only update a few times per second. Whist your monitor on a typical low end computer will update 60 times per second. So almost any computer you use should outrun the data it brings in by far.

[LaSardine]

My data stream has a refresh rate of 0.1ms, although not all data have the same refresh rate but everything is being pushed out at this rate.
I have seen massive latency differences in rendering between several machines I have tested this application with, just the display itself makes a huge difference, but also the video card, the connection from the video card and the display (hdmi/displayport/DVI), the Network Card, I have seen all these adding a lot of latency to the rendering. 

I am looking for the best theoretical latency for each part: like which network interface technology has the lowest latency, or is an 8-core cpu has more latency than a dual core, has DisplayPort a lower latency that DVI or HDMI, what about RAM, storage, etc.