cache_latents.py vs prepare_buckets_latents.py

I am trying to understand what is the usage difference between these two scripts.
If I understand correctly, then both are used to compute latents and store them to disk in advance, but I can't figure out from the readmes what the difference is and what to use when.

As for my use case, I want to do a full finetuning of SDXL and I'd like to compute and cache the latents on disk in advance because my dataset is quite large. All the images in the dataset have the same size of 1024x1024 px.

Read cache_latents.py, prepare_buckets_latents.py, and the README sections describing them; compare their documented inputs, outputs, and intended training cases. Document when to use each script, including the full- fine-tuning SDXL case with a 1024x1024 dataset, so a user can choose without further investigation.

Written by the indexing model from the issue text.