[TN]opyrt 2013-06-26 17:08:52 | Linac900Ext8Xc2 seems to have slowed down a lot around 3.65%. I believe we can still push it quite a bit further, so a couple of weeks ago, I made some custom samplefiles and distributed them to my team mates. After that, more team mates have joined in. Optimization 1 is between 3300 Mpts and 3450 Mpts and is currently at 3.1%. Optimization 2 is between 3200 Mpts and 3300 Mpts and is currently at 3.55% They should be quite easy to spot in the graph: http://stephenbrooks.org/muon1/plots/graph1/?optim=Linac900Ext8Xc2&coords= Both optimizations are seeded from BOINC results. It will be interesting to see if they will pass the current best! -opyrt |
Stephen Brooks 2013-06-27 01:09:25 | Looks good. Some time I should really have another look at this "getting stuck" behaviour and see what's so difficult about breaking out of the various plateaus. Currently a bit busy though. |
runesk 2013-06-29 14:03:43 | I had forgotten new lattices, and alas, the Linac900Ext8Xc2 were not grabbed from the start. As we started the alternative optimizations, I added the lattice to the grab list, and the video now clearly shows our mission to find a new better path. http://tools.1up.no/muon1/?lattice=linac900ext8xc2 .R [Edited by runesk at 2013-06-29 14:04:14] |
runesk 2013-07-01 14:27:11 | And the results are in. A new best optimization from the second "strain" by [TN]steinrar at 3.700289 / ~3294 Mpts. http://www.stephenbrooks.org/muon1/plots/graph1/?optim=Linac900Ext8Xc2&coords=?1120,100 .R |
[TN]opyrt 2013-07-01 18:28:04 | And then havardma decided to make it all interesting again by finding a better optimization on the old track! -opyrt |