It’s been a busy year, all in all.
First of all, the changes. We started this blog, and a twitter feed. I felt (with Stephen’s support) that it would be one of the better ways to get some information about the nuts and bolts of the project, in a more understandable style. However, since it wasn’t going to be Stephen doing it, it really couldn’t go on the Muon1 site, not at first, so this ‘un’official blog was born. Along with it came the twitter feed, which helped with presence. Both are run by me, although Stephen has access if needed (like something seriously wrong gets posted)
Early in the year, a new client version was released. While it wasn’t any quicker, it was consistently more accurate, and had less random noise in the results, which had led to some unusual results on occasions. I think this image best describes the difference, when v4.44d and v4.45 are used on exactly the same simulations.
 |
| v4.44d is the darker, v4.45 is the lighter White is yield (reds are 5x runs), blue is mpts |
Later in the year, I gave a talk at Dragon*Con, which covered the basics of the project, and showed its aims, and goals, as well as some of the science behind it. This was reasonably well received.
There was also a project started in the summer, to cover all the past lattices. Giving a brief overview, some details about the work done on the lattice, as well as who got the best result, it also included a video of that design. It started in mid-July with SolenoidsOnly, and is currently scheduled to finish sometime in May 2012. It doesn’t include the high-resolution scan projects (Linac900Ext10d2_axial3, and Linac900Ext10d2_zoom) used to determine client noise, however, since they were not open for the general userbase, being a brute-force scan of a very limited design area.
Of course, the biggest story of the year has been with the sheer amount of work done. These are covered by the stats published weekly, on a Tuesday (with a few gaps at the start of the year), on the twitter feed, the facebook page and on google+. Coincidentally, stats were released on December 27th in both 2010 and 2011 and that lets us get a nice overview of the stats for the year.
21,885 Trillion Particle timesteps overalland
Linac900Ext10d2 is at 2.771697%, first found by a BOINC wrapper user - an increase of 0.000315% over the past week
Linac900Ext6tc2 is at 4.386301%, found by a BOINC wrapper user - an increase of 0.000998% over the past week
27 December 2010 at 18:04
29,969 Trillion Particle timesteps overall (+212T over the past week)
Linac900Ext10tc2 is at 3.771338%, found by a BOINC Wrapper (yoyo@home) user, no increase from last week.
Linac900Ext9X is at 1.654297%, found by a BOINC WRapper user, an increase of 0.117037% over the last week.
27 December 2011 at 18:43
There is just one major thing to note. In almost exactly 1 year, 8,084 Trillion particle timesteps were calculated. That’s 27% of all work done to that point had been done in the previous year. And 33% of work had been done over the previous 18 months.
This meant that in June we hit the 25 Quadrillion (or Million-Billion) particle-timestep mark, but the real eye-opener was when, just over 6 months later, we hit the 30 quadrillion mark. It took 9 years to do 20, another11 months to get to 25 and just 6 more months to get to 30, and it's looking like 35 won't take too long either.
The 70-Million results milestone was also reached, in early November, and now, two months later we're done another million more.
There’s no question that Muon1 is growing, and it looks like it’s got a bright future ahead of it.
No comments:
Post a Comment