Building a home server for CUDA development

Are any user here CUDA developers? What do you think of the following build?

PCPartPicker part list: pcpartpicker.com/list/w66bm8
Price breakdown by merchant: pcpartpicker.com/list/w66bm8/by_merchant/

CPU: Intel - Xeon E5-1620 V4 3.5GHz Quad-Core Processor ($286.99 @ SuperBiiz)
CPU Cooler: Scythe - Kotetsu 79.0 CFM CPU Cooler ($32.95 @ Newegg Marketplace)
Motherboard: Asus - X99-M WS Micro ATX LGA2011-3 Motherboard ($399.99 @ Newegg Marketplace)
Memory: Crucial - 16GB (1 x 16GB) Registered DDR4-2133 Memory ($139.00 @ Newegg Marketplace)
Storage: ADATA - Premier Pro SP600 128GB 2.5" Solid State Drive ($59.99 @ Newegg Marketplace)
Storage: Seagate - Desktop HDD 4TB 3.5" 5900RPM Internal Hard Drive ($107.94 @ Amazon)
Storage: Seagate - Desktop HDD 4TB 3.5" 5900RPM Internal Hard Drive ($107.94 @ Amazon)
Storage: Seagate - Desktop HDD 4TB 3.5" 5900RPM Internal Hard Drive ($107.94 @ Amazon)
Storage: Seagate - Desktop HDD 4TB 3.5" 5900RPM Internal Hard Drive ($107.94 @ Amazon)
Video Card: EVGA - GeForce GTX 1080 8GB FTW Gaming ACX 3.0 Video Card ($539.89 @ Amazon)
Case: Silverstone - TJ08B-E MicroATX Mini Tower Case ($98.99 @ SuperBiiz)
Power Supply: EVGA - SuperNOVA G2 750W 80+ Gold Certified Fully-Modular ATX Power Supply ($93.98 @ Newegg)
Total: $2083.54

>nvidia
meme

I would honestly say get 2 GPUs. That way you can work with one with out affecting your desktop performance.

More "enterprise" companies prefer CUDA over OpenCL. AMD only has them selves to blame for releasing the same CPU 4 years in a row.

>CUDA
Not worth it unless you're doing meme learning.

Not him but CUDA is better and more popular than OpenCL

>better
Questionable.
>more popular
Aka more libs aka the only way CUDA is "better".

Oh please, OpenCL is not a standard

Get a skhynix ssd for that price

OpenCL is the standard, but shitty implementation and tools.

it's free

>get 2 GPUs
Easily the best advice in this thread. Sure, get a 1080 if you want, but you definitely want another card to drive your display, even if it's something cheap. A 1050 may be a good choice for that.

It may sound strange, but there are plenty of graphics applications (rendering, video processing) that graphics processors are great for.

>be nvidia
>own cuda
>want update new shit
>add it in the next version without problem

>opencl
>want add new shit
>you cant/or need wait because no standard

CUDA will alway be superior over OpenCl, it's a fact

t. Pajeet engineer

OpenCL 2.2 looks amazing but ... AMD will implement to next navi.

I have a headless server where I do Vulkan development but when I tried to set up my 770 gtx with the Nvidia drivers that seem to require me to set up an x server to access any of the gpu features?
Nearly every gpu command line tool complains about there not being an x server when this is clearly a headless server.

Also fucking arch removed Vulkan.hpp from the Vulkan headers repo.

Go on reddit, people on Sup Forums are too much autistic and anti nvidia.

That's strange, I do performance measurements with CUDA all the time, and always kill my X server before running measurements. CUDA runs just fine without X for me. Also, you *have* to kill X in order to install the drivers. It sounds more like an issue with the Vulkan tools to me, but that being said I've never tried installing on the drivers that didn't even have X installed.

You'll find only neet amdrones and shills here.

>Video Card: EVGA - GeForce GTX 1080 8GB FTW Gaming ACX 3.0 Video Card ($539.89 @ Amazon)
>EVGA
Enjoy your future denied RMA faggot.

Sadly this

>Video Card: EVGA - GeForce GTX 1080 8GB FTW Gaming ACX 3.0 Video Card
>Gigabyte GV-N1080G1 GAMING-8GD
Same price, and speed. I remember that the evga acx 3.0 coolers had issues with the gtx 1080 ftw, so you might be better off with this gigabyte.

Same thing user, I eventually gave up and just let Debian install a gui and I never attach a monitor to it.

Is it not cheaper to just rent?