Need advice for ML Setup

Sep 5, 2018
2
0
10
So my current System is completely useless for Machine learning and for general usage. So I was thinking about buying the parts and setting up a pc.
This is the setup I created.
At first I wanted to pick the 1080TI, which turned out to be too expensive, so I wanted to change to the 1070, but after reading a few forum entries it seems that the 1080 is highly recommended over the 1080 for ML, because of the CUDA Cores etc. Which gives me still 200€ saving compared to the 1080 TI.

But then I found a finished setup on Amazon.

It basically contains everything I need for a little less money and is already build. Is there any reason not to pick the Amazon one compared to my own build? Is the own build enough for ML in general?

PS: I'm from Germany in case of suggestions etc.
 
Solution
Good afternoon.

I speak from a general point of view. I do not know the specific needs of a PC for ML.

The Amazon PC is very inferior to the one you have selected. The GPU and CPU can be the same, but the rest of the components are not up to par.

Your CPU allows you to do OC (if you want) if it is paired with a Mobo Z370 as you have chosen. The Amazon one is a H310 chipset mobo and does not allow overclock, only has 2 memory banks and only allows 2666 memory speed. It is MicroAtx with less expansion cards.

Your DDR4 are 3000. The others do not indicate it, count 2133 or 2400.

Your PSU is golden 90% efficiency. Other one is 82%, they say....and I do not believe it.

You choose a liquid cpu cooler. Other one is stock intel cooler...
Good afternoon.

I speak from a general point of view. I do not know the specific needs of a PC for ML.

The Amazon PC is very inferior to the one you have selected. The GPU and CPU can be the same, but the rest of the components are not up to par.

Your CPU allows you to do OC (if you want) if it is paired with a Mobo Z370 as you have chosen. The Amazon one is a H310 chipset mobo and does not allow overclock, only has 2 memory banks and only allows 2666 memory speed. It is MicroAtx with less expansion cards.

Your DDR4 are 3000. The others do not indicate it, count 2133 or 2400.

Your PSU is golden 90% efficiency. Other one is 82%, they say....and I do not believe it.

You choose a liquid cpu cooler. Other one is stock intel cooler.

The advantage is that it includes windows 10.

If you are never going to overclock, or will not put a second GPU it is possible that the Amazon team will compensate you.




Produktbeschreibung:


Gehäuse:AeroCool Cylon schwarz, Acrylfenster, Kabelmanagement, Staubfilter, integrierte RGB-LED-Beleuchtung
Netzteil: 550 Watt Super Silent Netzteil Gaming mit 120mm Lüfter sehr leise und 82% Effizienz
CPU:Intel Core i7 8700K / 6 x 3.7 GHz - SixCore mit Hyperthreading
Cooler: Intel zertifizierter Kühler
Mainboard:ASUS Prime H310M-A – Mainboard – MikroATX - LGA1151 Socket – H310 Chipsatz, USB 3.0, SATA3, Gigabit Ethernet, Onboard-Grafik, 7.1 (Realtek ALC887), 2x DDR4 Speicherbänke, 1x PCIe 3.0 x16, 2x PCIe 2.0 x1, 4x SATA 6Gb/s
Speicher: 16 GB DDR4-RAM
Grafikkarte:Nvidia Geforce GTX 1080 mit 8GB Speicher
Festplatte: 256 GB SSD Solid State Disk SanDisk Plus
OHNE Laufwerk, nur extern möglich
Netzwerk: 10/100/1000 MBit Netzwerkchip onboard
Anschlüsse vorn:1x USB-A 3.0, 2x USB-A 2.0, 1x Mikrofon, 1x Kopfhörer
Anschlüsse hinten:2 x USB 3.0, 2 x USB 2.0, 2x PS2, Sound, DVI, HDMI, VGA(D-Sub), 1x RJ-45 Netzwerk
Gewährleistung: 24 Monate mit Pick-Up & Return-Service
Lieferung: PC-System, Netzkabel, Treiber CDs, Windows 10 Pro 64bit vorinstalliert inkl. Treiber und aller Updates ohne Werbeprogramme
Qualität von Memory PC - fordern Sie uns -
 
Solution
Sep 5, 2018
2
0
10


I was actually thinking about using my current gtx 970 as support for other operations, while the 1080 is doing the learning, so I guess in that case my own build will make more sense. Thank you so far gingerrankin!
I think I will wait a little and hope for someone with a little ML expertise to give his 10 cents about this and make my decision eventually.