BFS stands for 'The Brain Fuck Scheduler' and is an alternative kernel scheduler. BFS was created by kernel programmer Con Kolivas in 2009 who thought the mainline kernel scheduler CFS was antiquated and too complex. The BFS scheduler is very simple compared to the default CFS scheduler present in the mainline kernel, which stands for 'Completely Fair Scheduler.' Most default Linux distributions use the CFS scheduler, however it isn’t thought to be great. The supposed benefits of BFS versus CFS have been debated, but BFS is generally regarded as faster and offering better performance on lower spec hardware and desktop machines while CFS is intended for higher scalability (e.g., multi-CPU servers and clusters.) You can check out some more great information as well as some benchmarks of a shiny new BFS 3.0 Kernel, which were performed by Phoronix here: http://www.phoronix.com/scan.php?page=article&item=bfs_two_years&num=1
I'm using the new 3.0.0-Fusion kernel and I admit I've noticed a nice little improvement in snappiness all around.
Though, I don't know if I'm falling victim to affirmation bias, considering I do have a six core processor, which would seem to benefit from the CFS scheduler versus BFS, at least according to the aforementioned benchmarks. I'll give this kernel a spin and see what I think after a couple weeks of use.