What's up with this code?

[Gregory Petrosyan]

Again, thanks a lot!
I'm optimising my polynomial module as hard as possible, because when
it used native floats/integers it was very fast, but in some specific
cases -- too inaccurate. So, I migrated to decimals, and module began
to be very slow. Current success is nearly 4x speedup, and now it seems
to me that decimals are the bottleneck. Is anybody working on C version
of decimals? Is it a really hard work? (I am thinking about
re-implementing decimals in C by myself)