OpenPNP and other stuff server requirement

[blueskull]

Hi gurus here,

I'm looking for a computer that can do the following things as a home server:

1. Control an OpenPNP-based PnP machine (OpenPlacer) with 2 cameras,
2. Control a GRBL-based CNC (Carbide 3D Nomad 883 Pro),
3. Serve files from a relatively slow HDD array (software RAID 1, up to 200MB/s) over Windows file sharing,
4. SMB also shares a Canon printer,
5. Run a Brother scanner,
6. Stream desktop to another computer over RDP remote connection.

What kind of hardware should I need?

Here are two candidates that I have:

1. Intel NUC with Haswell i3, 8GB RAM, 256GB SSD
2. Intel NUC with Kabylake i5, 16GB RAM, 512GB SSD

I have both, but I want to sell one for some cash. Since I don't have OpenPlacer set up already, I can test out for real.
The second one has a 28W CPU that's more powerful than most high end Ultrabooks, so I have no doubt for it.
However, I prefer to use the first one since I can't really sell the first one for any reasonable amount of money, so I prefer to get rid of the second one.

OpenPNP knowledge is much appreciated from PnP gurus here.

[Halcyon]

What you want to achieve is relatively low-end stuff. They don't take a lot of computing resources at all (even all at the same time).

Your i3 NUC would do the job fine, but I guess it depends on how desperate you are for the money? I'd keep i5 NUC for myself and sell the older one but as you've already stated, you already have a preference so go with that.

Although, if at all possible, avoid Windows 10, it's complete garbage. I actually find it slower than Windows 7 in many ways.

[NiHaoMike]

Does OpenPlacer actually use the cameras for machine vision during operation? That can be quite GPU intensive.

[NiHaoMike]

Forgive me if I'm too outdated, but last time I used OpenCV (back in 2014), it didn't support GPU well.

It has improved by a lot. In mid 2017, I worked on a test setup that used GTX 580s (what they had lying around) as part of a machine vision system for automated testing.
https://opencv.org/platforms/cuda.html