What are the 'fast' LRU variants?

Both ConcurrentLru and ConcurrentTLru have 'fast' variants called FastConcurrentLru and FastConcurrentTLru.

Fast variants are functionally identical, but have the hit counting logic eliminated via meta programming to squeeze out the most performance possible. When benchmarked, this results in somewhere between a 10% and 15% speed up and a 15% to 20% increase in throughput.

If you don't use the hit count properties, use the fast versions and save some CPU cycles!