PCPartPicker

  • Log In
  • Register

Build

Dual GTX 1080ti Deep Learning Build

by jjpkrohn

Description

PURPOSE

This is a headless box running Ubuntu 16.04 that was built to train Deep Learning model architectures, e.g., in TensorFlow and Torch. I based my build on this one by FourMoBro: https://pcpartpicker.com/b/NW4qqs

CASE

This was my first build so the large, two-partition case made assembly much easier. It further enabled us to tidy away the cables in the back partition alongside the power supply, creating a pleasing aesthetic in the windowed, front partition.

MOTHERBOARD

Another component that was tremendously helpful given that it was my first build was the motherboard. It provides feedback via multiple LEDs that allowed us to debug rapidly and precisely.

MEMORY

Our models are memory-intensive so I filled all eight RAM slots with 16GB cards for a total of 128 GB).

STORAGE

For storage, a 512 GB SSD was sufficient for our day-to-day modelling purposes. I bought a cheap 3TB HDD for local storage of larger data sets that we on occasion (say, monthly) sample from.

POWER SUPPLY

The estimated maximum wattage is 882W so I added a 100W buffer to be safe, purchasing a 1000W power supply.

BLOG POST

To read more detail on our experience putting the box together, our summer intern, Max, published a blog post here: https://insights.untapt.com/building-the-magical-data-science-box-e933656d625e

Comments Sorted by:

MjolnirThunder 7 points 4 months ago

Is the M.2 card supposed to be bent way down like that? I thought you were supposed to move the stand-off at the end to the correct hole.

Liera21 2 Builds 7 points 4 months ago

That's why he needs a "Deep learning" computer

MjolnirThunder 1 point 4 months ago

Top kek.

lelouch462 3 points 4 months ago

That poor M2 slot....

Velcade 2 Builds 3 points 4 months ago

You are supposed to move the end support. RIP m.2 drive.

jjpkrohn submitter 1 Build 1 point 4 months ago

Oops :(

spoirier 1 point 4 months ago

I think he simply needs to add the stand-off under the screw.

jjpkrohn submitter 1 Build 1 point 4 months ago

Yeah, the drive's been in there and humming along happily since July 2017 so it doesn't seem to have been a fatal error. Got lucky!

IceColdInferno 2 points 4 months ago

Cute dog +1

jjpkrohn submitter 1 Build 1 point 4 months ago

Haha thanks. His name is Oboe. The server is called Bassoon.

IceColdInferno 1 point 4 months ago

Wow cute dog with a cute name +1

Gooberdad 8 Builds 2 points 4 months ago

Puppy FPS power.

Nice job. Where is the other 1080 ti?

jjpkrohn submitter 1 Build 1 point 4 months ago

I named the puppy Oboe and the server Bassoon :)

And, I've updated the photos so you can see both now!

Gooberdad 8 Builds 1 point 4 months ago

Nice. That's a lot of processing power.

Stingrider907 1 Build 2 points 4 months ago

Where is the second 1080ti? Also r.i.p. your m.2 slot.

jjpkrohn submitter 1 Build 1 point 4 months ago

I've updated the photos so you can see the second card now! And, thanks for letting me know about the m.2 slot -- oops :)

Radox-0 5 Builds 1 point 4 months ago

Your dog ate the second 1080Ti :O

jjpkrohn submitter 1 Build 1 point 4 months ago

Haha, I added a couple of new photos that show the complete build with both cards :)

skygunner27 1 Build 1 point 4 months ago

Sweet build!

blizz7 1 point 4 months ago

Does the cooler come with thermal paste pre applied? or did you have to buy your own? asking for a friend

jjpkrohn submitter 1 Build 1 point 4 months ago

You can let your friend (wink, wink) know that the thermal paste was pre-applied

blizz7 1 point 4 months ago

wink

poc88 3 Builds 1 point 3 months ago

Beast! love the Cube/dual chamber case, cant go wrong with that. SLI FTW!

AleksR 3 Builds 1 point 2 months ago

Hi,

can you check please if both of your graphic card´s are running at x16/x16 with this chip. i have the same cpu with a msi godlike and two evga 1080´s but they are running only in x8/x16

thank you

ZIC 2 Builds 0 points 4 months ago

Just get a Titian V and an R7 1800X

jjpkrohn submitter 1 Build 1 point 4 months ago

Great tip! I train separate models on the separate GPUs in parallel so its helpful to have both cards.