With 6.3, the AP radio is going to scan more aggressively compared to 6.1.3.9, there's a significant difference in terms of
the channel scan rate. If clients are PS enabled, it is only going to make things worse. Could you tell us more about what exactly is the measurement factor? It'd enable us to respond precisely.
i.e., TCP iperf/jperf test run from one WLAN client to another, where both clients are associated to same BSS
or TCP based download from wired source?
Also, are the 6.1.3.9 and 6.3 codes running across two different platforms which are benchmarked side by side (with default config, of course) or are they iterations on the same test bed?