The top 100 user list has been updated for the first time in several years.

---

And the majority of the featured users have a large pie-slice of Collatz Conjecture processing. How accurate is it to describe that as floating point operations, when the conjecture itself concerns integer mathematics?

And the majority of the featured users have a large pie-slice of Collatz Conjecture processing. How accurate is it to describe that as floating point operations, when the conjecture itself concerns integer mathematics?

The collatz' issue is that it follows FAH, in assigning quick return bonuses.
On the other hand, it does rate the gflops PPD correctly, meaning a high end GPU should get between 100 to 1000x better performance than a CPU.

Again, my question was: floating point maths/operations, or integer ditto?

How do you know that "it rates the gflops PPD correctly"? What measure of floating point speed are you using? Might you be using the 'GFLOPS Peak' reported by BOINC, which is a theoretical maximum known not to be achievable in real-world scientific maths: its only value is for advertising.