What hardware do I need to run 4CAT?

The question 'what kind of hardware do I need to run 4CAT' is difficult to give a definitive answer to. Generally these are the considerations for CPU, RAM and storage:

• Generally speaking, RAM will be the bottleneck, as data needs to be stored in memory for many types of analyses and especially with bigger datasets you may run into hard limits in that regard.
• The CPU is less important, as it is not the bottleneck for most types of analysis. 4CAT can only benefit from more CPU cores to some extent due to Python's Global Interpreter Lock. In terms of CPU speed more is as a rule of thumb better, but in our own use of 4CAT we have never had cause to pay too much attention to the speed of the CPU.
• For storage we recommend flash storage, i.e. SSDs (over NVMe ideally, or SATA) over spinning disks as 4CAT reads and writes data often in its process of capturing and analysing it.
• Disk Space is important insofar as it determines how much data you can capture. Especially when downloading images data can accumulate rapidly. 4CAT can be configured to automatically delete old datasets to help you manage this.

Capturing data with 4CAT and serving its front-end requires relatively little resources and a standard 8GB (virtual) server may suffice. When also running processors for analyses, as a rule of thumb we would recommend at least 16GB of RAM especially when using NLP processors (such as tokenisation, generating word embedding models or using topic modeling).