Deep PC
Deep PC
Computer Hardware
Entry-level Machine Learning / Deep Learning Build
Full specs on PCPartPicker
Highlights:
- CPU: Intel Core i5-7600K 3.8GHz Quad-Core
- RAM: 32GB (4 x 8GB) DDR4-3000
- SSD: Mushkin Reactor 1TB
- GPU: GeForce GTX 1080 8GB
Components: $1532.95
Labor and setup: $200
Total: $1732.95
$1,73295
Ultimate Machine Learning / Deep Learning Build
Parts for a high-end system suitable for deep learning POCs with container architectures, and running large number of instances on a single box.
Performance and quietness are primary. One could scale down various options (go with smaller disks etc.). In noise, the system is comparable to a quiet Macbook Pro, the fans are barely noticeable due to the sound insulated case and the quietest fans. The current system has one Titan X GPU card, more can be added. The workstation motherboard supports efficient data exchanges between the GPUs and m2 SSD disk and provides a room for growth.
In comparison with systems one could buy, at the time of purchase, it was over $2000 cheaper than the cheapest similar system, and about $7000 cheaper than a rather expensive one offered by NVidia. The quality of our components (case, fans) is much better than available offerings.
Some competing systems:
- https://developer.nvidia.com/d
evbox - https://exxactcorp.com/index.p
hp/solution/solu_detail/225 - https://www.amazon.com/Lambda-
Deep-Learning-DevBox-Preinstal led/dp/B01BCDK1KC
Highlights:
- CPU: Xeon E5-1650 @ 3.6GHz
- RAM: 128GB - Samsung DDR4-2400 4x32GB ECC
- GPU: NVIDIA Titan X Pascal 12GB
- SSD: Samsung 960 Pro 1TB M.2
- HDD: Western Digital Gold 10TB SATA 6Gb/s
We charge $800 on top of components, and we can also install deep learning software stack at additional cost.
$80000
Power Efficient Build
This machine is designed to consume very little power -- set it up as a dev machine that you can keep on as a server, or run it as a NAS.
$74500