My understanding is that OS X/HFS+ already optimizes for reusing already allocated blocks. This means that fast write-space is conserved as much as possible at the expense of larger commit-loads from the OS's RAM cache when writing to already allocated blocks.
Also, what I've read seems to indicate that OS X behaves sub-optimally, but this is still good enough to support snappy behavior under normal workloads.
Not sure I follow.... the fast-write space in a non-TRIM device with GC is just some "extra" blocks (or whatever) that are kept aside and now and then swapped out for other in-use blocks (as far as the drive firmware is concerned - the OS has no idea this is going on) - and then the spare blocks are wiped for fast write again, then used when writes are requested.
Really appreciate if you could shed any more light on why OSX would be better at this.. not sure I can picture this.
Also, what I've read seems to indicate that OS X behaves sub-optimally, but this is still good enough to support snappy behavior under normal workloads.