Was just reading the git wireless driver issues email list, and saw some comments on a fix for the wireless client latencies. I'll have to see that fix in actual firmware, but the Marvell guy sounded convinced that he had fixed this. However, he also said that he hadn't experienced the problem, so this is a theoretical fix on his part, a sketchy thing to hang one's hat on. I don't have a lot of confidence in theoretical fixes, myself, I have to see them actually work. A lot of times, when one fix is applied that's "sure to fix the problem" without the ability to test the fix, a secondary effect or an unanticipated addtional problem will come up. When fixing bugs, there's no substitute for being able to reproduce the problem in the lab.
So we'll see how things look when this theoretical fix comes out in some firmware.