Hardware suggestion


I am trying to develop a deep learning based model for automatic segmentation of medical data. I would be using the UNETR model for this purpose. I wanted to get some suggestions on the hardware that I can use to develop this model. Currently my laptop has:

  1. OS: Windows 10
  2. RAM: 32GB
  3. SSD: 1 TB
  4. GPU1: NVIDIA QUADRO T2000 with Max Q design with dedicated GPU memory of 4GB, and shared
    GPU memory of 16 GB.
  5. GPU2: INTEL(R) UHD Graphics with shared GPU memory of 16GB.

The training dataset would consist of around 300-500 CTs. I needed some suggestions on what should I go with: an external GPU or a cloud based service. And what external GPUs would be a good option.