GC performances

[bearophile]

Craig Black:

> Here's an array-based C++ implementation from the shootout.  It performs the 
> best but was disqualified because it allocates everything in one big chunk 
> and then uses placement new to allocate each node.

Allocating huge chunks is slow on Win, look at my better version here:
http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D&article_id=62373

Bye,
bearophile