Help me build ultimate LIDAR survey workstation price no object

[dinoarmas]

1. Budget & Location

Under $10,000 for sure, Maybe a $5,000 option. I do not need a video card. My Location is in Irvine, California.

 

2. Aim

Be able to adjust 1+ Terabyte Mobile terrestrial LIDAR scanning projects for land surveying purposes. The software does not use GPU for adjustment processing in anyway. I need as many Core/Threads as possible with the most bandwidth possible to fastest RAM possible. Software does not utilize more than 16GB of RAM. So 32GB of RAM is more than enough. I also need to have the fastest Read/Write speeds to Storage as possible. The software is reading and copying very large files to storage during the adjustment process of the LIDAR point clouds.

 

3. Monitors

No monitors needed, I will be using 2 monitors, but it might be nice to have a 3 monitor setup. I will be using a Video Card from a previous workstation setup.

 

4. Peripherals

None needed

 

5. Why are you upgrading?

Need to increase the speed at which I can process these point clouds.

 

I am spec'ing out a computer for my work desktop PC that my company will purchase. I am processing very large LIDAR point clouds for surveying purposes. Please advise on where I can go to compare motherboards. I would like to have a complete list of motherboards that have these boxes checked off.
 

• Has dual Xeon chip sockets
• M.2 NVMe gen3 SSD slot (that can utilize a 2TB+ if that is a concern)
• Intel Optane maybe?

If anyone could recommend which Xeon CPU and RAM would pair nicely with any motherboard that you recommend that would be appreciated as well.

I don't imagine that there is a long list of motherboards that have these requirements. Being that M.2 NVMe SSD's are fairly new.

The bottle neck I am worried about is the read & write speeds to storage from the CPU. I know I could do some type of RAID setup but would like to worry about that at a later time. Just want to get the fastest CPU to a single storage drive speeds as possible.

[Energycore]

~Moved to New Builds and Planning

[bimmerman]

Can't say I know much about motherboards for that configuration, only that more workstation-oriented boards now seem to have NVME slots.

 

Some more general questions though:

• does the LIDAR point cloud software actually take advantage of high core count? Do you know, roughly, how many cores and threads you need?
• Do you need ECC ram?
• Do you need much graphical horsepower (i.e. how nice of a Quadro) or is your workflow mostly computation and storage oriented?

 

If you really don't need dual sockets, I'd suggest looking at a Threadripper (because can use LOTS of ECC ram) or the new Xeon-WS chips (think Intel i9 x299 chips that support ECC). Threadripper or the higher end i9/Xeon-WS chips will get you 16+ cores and 32+ threads, if that's enough.

 

If you need GPU performance for rendering and manipulating the point clouds, I'd get a mid to high end Quadro or Firepro, because they have certified drivers (might not be relevant for your software) and because they have ECC memory.

[dinoarmas]

Company that makes the software is Riegl the name of software is RiProcess. The software does utilize multiple core and threads. I don't know if there is a limit to how many it can use. I am going with more the better.

 

ECC ram is NOT needed.

 

NVIDIA Quadro K5000 4GB of memory is what I will be using from my previous setup.

[dinoarmas]

Samsung 4TB 850 EVO            (probably would not utilize much but for storage of projects before I move them over to NVMe drive)

Samsung 2TB 960 Pro NVMe   (actual drive that would have the saved project that would be getting adjusted)

Intel Optane                               (would this 32GB of Cache storage even help?)

CPU                                           (Xeon? 1 or 2?)

32GB RAM                                 (what kind and what speed to maximize communication to CPU's?)

Motherboard                               (Which one would take the most advantage of all of this speed and also as much future up-gradable as possible)

 

Really looking for the exact Motherboard I will need to utilize all this stuff but also any advice for other hardware that I have not mentioned that might increase IOPS.