Best deep learning workstation under $1800
Buying your first machine learning workstation
As a machine learning researcher, I am glad I have large GPU clusters I can train my neural networks on, which allow me not to worry about computational resources. However, when it comes to my side projects, I am a Mac user (no NVIDIA GPUs) and until last week, I did not have a proper workstation to work on. While Google Colab is an amazing free tool, it does not scale and as an AI enthusiast, buying my own deep learning workstation will quickly pay off, as opposed to using GCP or AWS.
Therefore, I have recently decided to buy a deep learning workstation for personal use but surprisingly, I could not find useful recommendations on the web. Therefore, after having done my own researches, I have decided to share with you my findings and recommendations.
What I was looking for
- Storage: I do not need much storage. Of course, some datasets are heavy but as I work on short-term projects once a project is done I can delete the dataset from my computer. The ideal configuration would be a 500GB solid state drive (SSD) for storing the Operating System and current projects together with a 500GB (or 1TB) hard disk drive (HDD) for storing past projects.
- CPU: As the algorithms I use are optimized to run on GPUs and as they do not require much control flow, the central processing unit is not that important to me. The CPU will mainly be used to transfer data to the GPU.
- RAM: I am looking for at least 8GB of RAM. Anything above 16GB would be an overkill for my usage of the workstation.
- Budget: I am looking for something that costs between $1000 and $2000.
- Noise: The workstation will be in my home office and some scripts will be running while I am working on other things. Therefore, I want a calm workstation that do not make much noise.
- Operating system (bonus): I am looking for a machine operating on Linux. That would save me from paying Windows 10.
I think these are standard requirements that many beginners in machine learning are looking for.
My choice
My researches led me to the MSI Aegis R.
It is a great trade off with respect to my requirements and the brand is well-know which ensures me that the machine will last in time. The GPU (GeForce RTX 3060) has 12GB of VRAM which enables to do a large range of deep learning projects. If 8GB are enough for you, you may prefer to go for the GeForce RTX 3060 Ti which is more powerful and has 8GB or VRAM. In that case, I would rather recommend the Skytech Shadow 3.0.
Please let me know in the comments what do you think or if you have other recommendations.
I hope the story was useful to you. If you enjoyed it, please leave a clap to support my work, it will propel other similar stories.
Disclosure: I only recommend products I use myself, and all opinions expressed here are my own. This post may contain affiliate links that are at no additional cost to you, I may earn a small commission.
