By the time that quads get into the 12 hour CPU time, we'll be up to 70-80Mb, same as in the beta and as per the project news/ research descriptions and FAQs... i.e. no surprises.

The upload bottleneck they'll have to address. A single result upload manages to get here about 90-95kBps at top. 3 uploads concurrent, part _4, gets me about 35-38kBps, thus making it seem part of the crawl is caused on my side. For now don't think that lowering the requirement is yet in order... let's see first which side buckles first when the windows app comes on-line, but then WCG/Harvard have to address the time outs. Still seeing the occasional, restart when near 100% and then resuming from near the beginning... that wastes every contributors bandwidth, whether limitless or budget. We're now in the 13-15MB result sizes so expect to see more as we progress, if that remains unaddressed... but knobbing bwdown... there's no reason. My quad shows 1.8mBps, so the bandwidth is there and that after a longer series of receiving only CEP2 tasks.

edit: added windows part.

Wow... no, I never saw any that large, and I had quite a few beta WUs.

Thanks for the quick reply.

For what it's worth, when the largest result files here got up to about 35MB, they dropped back down to 15MB and started growing again.
I don't know how many times they've done that - I just happened to notice it a couple days ago.