Re: Fastcode MM B&V 0.41

Hello, everibody!
This is first time i post in this news group, but i read it very much before.
Especially topics on memory managers, because i'm very intersted in its implementations.
I have one little question about FasMM2:
Since version 0.30 (i'm not sure exactly) i came into strange timings of this MM.
It began to work much more slow than before (about 1.5 slower) on many tests.
For example:
Block size spread:
v0.30 = 2253 MTicks
v0.40 = 3300 MTicks
ReallocMem bench:
v0.30 = 1752 MTicks
v0.40 = 5300 MTicks

I can't find the reason and very puzzled about it, since source of FastMM wasn't changed
and source of this benches also wasn't changed. Can anybody biring a light in this question?
FastMM is my favorite Memory Manager and i very appreciate the work of Pierre le Riche.
.