-b9 has been reduced to 7% dev fee.  6.5% me, 0.5% jh00.  Windows build is 6% me, 0.5% jh00, 0.5% Clintar.
http://www.cs.cmu.edu/~dga/crypto/ric/

(same filename, but re-uploaded on top of the old one - nothing else has changed).  As before, "smaller" uses less memory, but might be a little slower with the increasing diff.  It sieves out to fewer primes - 200m vs 500m.

md5 checksums:
951b331d896e47c754c974700a20b686  xptminer-sse4-b9
83abf911e7efb1ac973b6501c454ae4b  xptminer-sse4-b9-small.exeat
3805bc3baba80a42f586543f340d517f  xptminer-sse4-b9-smaller
edd62ba2088389d94016e71910b2ddf9  xptminer-sse4-b9.exe

Thanks to everyone who's putting up this experiment in yet-another-model-of-funding an eventual open source release of fast mining code.

Plan remains unchanged but added to a bit:
- Friday: 5%
- Sunday: 3%
- Whenever the definition of share is changed to match what Gatra & jh00 have decided:  Open sourced, with 1% dev fee left sitting in the code that will be easy to remove if you feel strongly about it.

I'll throw 200 RIC out of dev fees to the Riecoin foundation if it's created within the next two months.

Happy prime cluster finding, everyone.  I noticed the diff was over 1600 for the first time today - congratulations!

Update:  By request, there's now one avx2 version.  This is a test - if people provide good comparative benchmarks at the same diff and find that it's faster, I'll get it out for windows and keep providing that one too.  For now, regular size / linux only.  If it's faster enough, I'll work on improving the build process to automate creating some CPU-specific variants.

  -Dave