Starting off with some memory-bandwidth analysis, which provides an insight into real-world performance, it's safe to say that users should steer clear of DDR3-1,333 memory. DDR3-1,600MHz produces decent scores, though increasing the memory speed further does little to impact this benchmark.
The write test is memory-agnostic, at least as far as AIDA is concerned.
Memory latency is a product of frequency and timings. In an ideal world you'd want high frequency and tight timings. The best of the bunch is the DDR3-1,866 set, mainly because its speed is reinforced by relatively-tight timings. Again, DDR3-1,333 does the A10-5800K no favours at all.
A10-5800K's performance doesn't change a great deal when we swap out DDR3-1,333 for DDR3-2,133. The 1,866MHz set remains the best.