Hi Charles,

Thank you for the question. ENVI software is single-threaded, and so processing large imagery and/or many vectors can slow down processing, even with a lot of RAM. Sentinel-2 imagery is pretty large, as you're also accounting for the number of bands (usually 13 for Sentinel-2), as well as the pixel values and pixel type. So, processing may take a bit of time.

Some questions for you:
1. How long does processing take when you mean slow performance?
2. When you say ENVI Crashes, when does this happen in terms of specific workflow, what happens, and are any errors thrown - if so, can you share these? Having a step by step understanding of your workflow and where you experience the issue is beneficial to reproducing the problem.
3. What specific type of Sentinel-2 imagery are you working with (e.g. Sentinel 2A, 2B)? For example, where did you obtain it (e.g. Copernicus, USGS Earth Explorer, AWS), and what image type is your Sentinel-2 imagery? (Image type: https://www.harrisgeospatial.com/docs/Image_Types.html)

4. Can you elaborate on what you mean when you say "I am trying to run band statistics subsetting with ROAs (14 classes)" - what tools are you using within ENVI?

Let me know!
Jenny