Same code run multiple times gives wildly different timings

General Usage Performance
, ,
28

I mean, I removed all differences between the nothing case and the InvGamma case, and the timings look the same, which makes sense since the two cases became identical.

Tried @turbo (1 CPU core) instead of @tturbo:

  1. Run 1: 00:48 vs 00:49
  2. Run 2: 00:49 vs 01:17
  3. Run 3: 00:48 vs 03:24 (!)

CPU usage was around 100% out of 400% (because I have 4 CPU cores total) and never dropped while running the code.