Close
0%
0%

Building a Deep Learning PC

This project is about a (moderately) powerful PC I'm making for myself, mostly for studying deep learning.

Similar projects worth following
Over the last few months, I've been saving up for buying myself a desktop PC with a GPU and enough RAM to be able to run at least Kaggle-grade problems without resorting to AWS or other VPS services.

Here are the basic specs:
1. GPU: NVidia GTX 1080 Ti
2. CPU: Intel i7 6850k
3. Memory: Currently 32 gigs, expandable upto 128 gigs
4. Motherboard has 4 PCI slots, but only two can run at 16x simultaneously.

Inventory:

1. Motherboard: Gigabyte GA X99P SLI

I found that the biggest filter in motherboards is the amount of memory they can support. I was pretty certain that I wanted a motherboard that would support up to 128 Gb of RAM. The next biggest filter was the PCI slots. I needed to be able to plug in at least two GPUs at full capacity. For these criteria the price tag on this model was pretty reasonable.


2. Processor: Intel i7 6850k

This wasn't really my first choice, its also a little expensive but I got a pretty good deal on it. The biggest bottleneck in choosing a CPU was whether the motherboard would support it (you probably have seen the warning about BIOS and X99 chipsets on). The Gigabyte website says that the GA X99P SLI supports this CPU since the F22 bios. This particular motherboard was manufactured in November 2016, so I wasn't sure whether it had the right BIOS version. After going through many forums, it seems like I don't have anything to worry about. In any case, the motherboard supports Q - Flash. So, worst case, I can update the BIOS myself if need be.


3. GPU: Gigabyte Aorus GTX 1080 Ti

Decent video memory and its the most expensive I can afford. The benchmarks are pretty good. Apparently if a Titan is 1.0, this GPU is at 0.9

4. RAM: 2x 16 GB Corsair Vengeance LPX DDR4 3000 Mhz

5. Case: Corsair Graphite 780T (Black).

I wanted to buy the 760T, but when I saw this, it was love at first sight. The only downside is that it's too heavy, maybe a little over 10 kilograms. It's big enough an EATX motherboard.

6. CPU Cooler: Cooler Master Hyper 212x

That's just the simplest model I could find. Although in hindsight I think air coolers are a little ugly. I don't want to have a hunk of metal staring back at me when I look at the case. Maybe I will replace it with a water cooling kit in a couple of years.

7. Power Supply Unit: Cooler Master V1000

A 1kW PSU seems a bit like overkill for something like this setup. But I have every intention of expanding it whenever I can. It's a "modular" power supply. Apparently that means that the cables are detachable.

8. Keyboard: HyperX Alloy FPS (Cherry MX Brown)

I'm really spoiling myself rotten with this. But the sound of it is beautiful.

9. Monitor: Dell U2412M

I've never really spent too much time with an external monitor with my laptop. This model came highly recommended from a friend, especially its anti-glare screen and the unique 16:10 aspect ratio, which seems to be ideal for looking at a lot of code.

Apart from all this there's a bunch of adapters, cable and a couple of screwdrivers.

  • 1 × Corsair Graphite 780T (Black) The case
  • 1 × Gigabyte GA X99P SLI The motherbord
  • 1 × Aorus GTX 1080 Ti The GPU
  • 1 × Intel i7 6850k The processor
  • 2 × Corsair Vengeance LPX 16 Gb DDR4 3000Mhz The memory kit

View all 11 components

  • All components received

    Jaidev Deshpande07/26/2017 at 03:45 0 comments

    I got the components I was expecting from Bangalore (CPU, GPU, RAM and SSD). We're chasing a product release at work this week, so I won't have time to build the PC before the weekend. Over the next couple of days, I'm going to take detailed inventory and use this to check items off, once they're consumed in the build.

  • Practice Runs

    Jaidev Deshpande07/24/2017 at 13:39 0 comments

    I've never built a PC myself before, so I made it a point to practice installing components first.

    I don't have the CPU, GPU, RAM and the SSD yet, but I've done a little dry run, trying to install the CPU cooler and the power supply on the motherboard and the motherboard in the case. So far it seems quite easy, although the CPU cooler installation scares me a little bit. And I don't want to touch the CPU until it's the actual build itself. Everything else is a breeze. The manuals are very detailed, and there's always Reddit, YouTube and Google.

  • Waiting for components

    Jaidev Deshpande07/24/2017 at 13:07 0 comments

    I've finished all my purchases. My GPU, CPU, RAM and SSD are with a friend in Bangalore who's about to come down tomorrow. I'll probably have to time to assemble the machine next Saturday.

View all 3 project logs

Enjoy this project?

Share

Discussions

caBattista wrote 07/24/2017 at 14:51 point

Interesting... What software will you use?

  Are you sure? yes | no

Jaidev Deshpande wrote 07/25/2017 at 07:09 point

I'm planning to use Keras + (Theano / Tensorflow) and the usual scientific Python stack. Might even tamper with PyTorch or caffe for benchmarking. 

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates