Best Computer For TensorFlow

TensorFlow is a deep learning framework based on the Python programming language. It is used for tasks like voice recognition and image recognition. The minimum requirements in a computer for TensorFlow are quite basic, but some high performance components are required to make the most of it.

In this article, we’ll be talking about the components you need to build the best workstation PC for TensorFlow.

Why choose us?

Parts like CPU cooler, Power Supply Unit, and chassis are often glossed over when putting together a system like this. Surprisingly, these parts have a massive effect on the lifespan and performance optimisation of the CPU, GPU and RAM. We only use parts that guarantee excellent cooling, airflow, and power delivery.

We’ve also got a lot of experience in putting these systems together, and our cable management is flawless. This not only means dust will be a lot less of an issue, but it will be much easier to work with when you eventually need an upgrade. This will only happen several years down the line, of course. Not to mention our cases just look fantastic.

Each Modena Computers Workstation is tailored to the personal needs of the user, but the features below would make for a great starting point for a computer for TensorFlow.

Need help choosing your ideal spec?

Please include as much information as possible regarding your requirements, including budget, programs used and any other information you think that might help us craft the perfect computer for you!

CPU – Core Count

The CPU is not the main priority in a system built for deep learning, but it can be extremely useful to help the graphics cards reach their full potential.

The CPU is mainly used for data preprocessing, and there are two common approaches to this task. If you prefer to preprocess your data while you train your models, additional cores can make a huge difference. In this case, you would get the best results with 4 CPU threads per GPU.

Alternatively, if you would prefer to preprocess your data before any training takes place, you would get great performance results from just 2 CPU threads per GPU.

CPU – Frequency

With the minimal calculation done by the CPU in deep learning, clock speed doesn’t make much of a difference. This means that multiple threads can be prioritized in order to eliminate bottlenecks on the GPUs. Even with this in mind, AMD’s current generation Ryzen CPUs feature both excellent clock speeds and high core counts. This would make them a great fit in a computer for TensorFlow.

GPU

Graphics Processing is by far the most important consideration for deep learning. A strong GPU is extremely important, and we would in fact recommend the use of multiple GPUs in a computer for TensorFlow.

For this reason, the Modena Computers Hydra XL Workstation would be the ultimate machine for this kind of work. This system has enough room for multiple high end graphics cards, and enough CPU threads to eliminate any potential bottlenecks.

Another useful note is that your display cable of choice can be plugged directly into the motherboard, rather than outputting display from one of the graphics cards. This “headless” approach allows the GPUs to work at their full potential without splitting their attention between TensorFlow and display output.

RAM

RAM does not directly influence deep learning performance, but you should have at least enough of it to manage your GPUs. A general rule of thumb is to have at least as much RAM as the video memory of your largest GPU. Note that matching only the largest GPU is necessary, and it is not necessary to match the total VRAM of all GPUs in the system.

This is a great starting point, but you may still want even more RAM for your system, especially if you plan on working with large datasets. Additional RAM is also useful to provide headroom for multitasking. Not to mention that you might not only be using your computer for TensorFlow.

Storage

Hard drives are generally not a bottleneck for deep learning, but we only use super-fast NVMe SSDs, which are much faster than more common SATA drives, and even regular M.2 storage. This is our recommendation from a quality of life perspective as well.

Conclusion and Sample Configurations

As we said before, each Modena Computers Vulcan Workstation is custom built to the workflow and budget requirements of its user. It’s important to consider other programs you may want to use with this workstation.

During our consultation process, you can let us know exactly which programs you want to use with your machine, and we’ll be happy to help you understand what you need. 

Below are three examples of specifications to match three different budgets.

Vulcan Workstation

Entry-Level
  • AMD Ryzen 5 5600X (6c/12t) CPU
  • AMD X570 Chipset Motherboard
  • 16GB 3200MHz DDR4 RAM
  • Nvidia RTX3060Ti 8GB Graphics Card
  • 480GB M.2 NVMe PCI-e SSD
  • Windows 10 Professional x64 Digital License
  • 3 Year Warranty and Lifetime Support
  • Keyboard and Mouse
  • Monitor

Vulcan Workstation

Mid-Range
  • AMD Ryzen 7 5800X (8/16t) CPU
  • AMD X570 Chipset Motherboard
  • 32GB 3200MHz DDR4 RAM
  • Nvidia RTX3090 8GB Graphics Card
  • 960GB M.2 NVMe PCI-e SSD
  • Windows 10 Professional x64 Digital License
  • 3 Year Warranty and Lifetime Support
  • Keyboard and Mouse
  • Monitor
Popular

Hydra XL Workstation

High-Spec
  • AMD Ryzen 9 5950X (16c/32t) CPU
  • AMD X570 Chipset Motherboard
  • 128GB 3200MHz DDR4 RAM
  • 4 x Nvidia RTX3080 10GB Graphics Card
  • 960GB M.2 NVMe PCI-e SSD
  • Windows 10 Professional x64 Digital License
  • 3 Year Warranty and Lifetime Support
  • Keyboard and Mouse
  • Monitor
Share on facebook
Facebook
Share on twitter
Twitter
Share on linkedin
LinkedIn

Leave a Comment

stay informed!

Subscribe to receive exclusive content, including specials and the latest industry news!