Interesting. Would it be helpful if I updated the open-source miner? I don't have very much time right now in the semester, but might be able to sneak in a weekend. It's been two years - I bet I can find at least a 10% speed bump, though it's been a while since I've looked at my code.
A performance boost for the
open source miner would certainly help promote better hashrate distribution.
+1 Here ... I might even start mining again.

ok. I'm out of time for now, but as a start, I've:
(a) Merged the pull request for stratum support from the riecoin github account into mine, in branch "stratum".
(b) Started a little bit of code cleanup as I try to remember my way around the code. As a starting point, I
ripped out a little code in favor of using the primesieve library, fixed a bug with the choice of the number of
workers doing the sieve, and started moving the code a bit more to more-standard c++11.
this is all in github/dave-andersen/fastrie on branch "stratum".
is there any reason to keep the old ypool/xpt protocol version around, or should we just go all-stratum?
I'd be curious if anyone sees a performance difference or regression from that bugfix, but it won't
be more than a percent or three. Right now, my focus is on getting the code into a slightly
better place for further optimization. (Please don't expect miracles or be annoyed if they don't
manifest.

Sweet ... a good start ... looking forward for me open source. This is what makes a community stronger!