Ihor Radchenko <yantar92@posteo.net> writes:

> I will now try to record the data on my sessions and share it later.
> We can then see what useful information can actually be extracted.

I am attaching some stats I recorded by varying gc-cons-threshold and
using Emacs.

Some observations:
1. GC time increases over time
2. Total time spend in GC decreases with increases gc-cons-threshold
3. Avg time spend in a single GC is roughly the same, except for very
   high gc-cons-thresholds
4. Memory limit has no obvious correlation with gc-cons-threshold

Of course, we need data from more users to draw meaningful conclusions.