Message boards :
Number crunching :
LHC@home support added to KBoincSpy CVS (purty pics included)
Message board moderation
Author | Message |
---|---|
Send message Joined: 18 Sep 04 Posts: 3 Credit: 1,372 RAC: 0 |
Great news for GUI Linux crunchers! Using the SixTrack 3.1 User's Reference Manual, I was able to make sense of the LHC@home workunit and state files, and to add in-depth support for LHC@home to KBoincSpy. First of all, I've added a panel that shows some basic information, including the current number of turns (screenshot). This is very useful to me right now, since the Linux LHC@home doesn't report progress. But the real treat is the "Tracking Details" window, that allows you to see the position of all the 60 particles (screenshot1, screenshot2). These two still images don't make justice to the animation, which looks awesome (in my very biased opinion). Some technical info about the implementation: a) For each particle, I determine the position by taking the "x(mm)" and "y(mm)" quantities (from the SixTrack manual). This is real data from the workunit! It's my understanding that instead in the Windows screensaver the particles displayed have nothing to do with the workunit. b) Since data is written to the state file at intervals of (usually) 100 turns, interpolation is necessary to compute the position of the particles for the missing turns. I use piecewise cubic interpolation: the curve in the interval x[i]...x[i+1] is the interpolating polynomial of the interval x[i-1]...x[i+2]. It's no cubic spline, but it appears to be smooth enough for my purposes. c) The processor usage when playing the movie for all 60 particles (I'll soon add the possibility to reduce the number of displayed particles) is currently 2-3% on a 1.6 Pentium M with a Mobility Radeon 9000. I'm a total newbie at OpenGL, so this can be probably furtherly optimized. The rate of the animation is 10 turns per second. d) The current prototype will improve in the next few days. Although I interpolate also the particles' energy level, I don't use that information right now; the next step is to draw each particle brighter or darker using this data. Support for LHC@home is in the KBoincSpy CVS repository. If you know how to retrieve and build programs from CVS, you can experiment with these new exciting features right now. Otherwise, wait for the next release of KBoincSpy, which will include them. KDE 3.3 and the Mesa (OpenGL) libraries are required. - Roberto |
Send message Joined: 2 Sep 04 Posts: 378 Credit: 10,765 RAC: 0 |
Rock on, those are cool pics. I'm not the LHC Alex. Just a number cruncher like everyone else here. |
Send message Joined: 18 Sep 04 Posts: 3 Credit: 1,372 RAC: 0 |
Thanks! BTW, I forgot to include in the original post a link to the KBoincSpy page, as well as to a page containing some CVS instructions. I've added the code that draws the particles brighter/darker according to the energy data, but I noticed that the energy stays the same throughout the workunit (on all the workunits I've tried), so the pictures look exactly like before. If one of the LHC@home developers is reading this: how this can be? I don't understand much about particle accelerators, but intuitively it seems to me the particle energy should vary (decrease?). - Roberto |
Send message Joined: 1 Sep 04 Posts: 139 Credit: 2,579 RAC: 0 |
Roberto, many thanks for your great contribution. You will be contacted very soon by a Sixtrack developer for discussions on improvements, etc. In the meantime, you'll be relieved to hear that the particle energies should only oscillate very slightly around a mean value for the whole period of the calculation, so it looks like your system is correct. Ben Segal / LHC@home |
Send message Joined: 2 Sep 04 Posts: 39 Credit: 4,910 RAC: 0 |
Like I have told you in the last mail, good work! I'm waiting for a Windows release or that I manage to run BOINC on Linux (with WLan). |
Send message Joined: 18 Sep 04 Posts: 3 Credit: 1,372 RAC: 0 |
> Roberto, many thanks for your great contribution. You will be contacted very > soon by a Sixtrack developer for discussions on improvements, etc. In the > meantime, you'll be relieved to hear that the particle energies should only > oscillate very slightly around a mean value for the whole period of the > calculation, so it looks like your system is correct. > Ben, thank you very much for your reply. I'm always trying to improve support for the LHC@home in my program (in part to compensate for the lack of a screensaver in the Linux version, but also because I'm very curious about the science behind LHC@home), so I'd be very grateful for any feedback you guys could provide me. - Roberto |
Send message Joined: 30 Sep 04 Posts: 112 Credit: 104,059 RAC: 0 |
Excuse the n00bish question, but this program will only run on the *.nix platform correct? |
Send message Joined: 30 Sep 04 Posts: 112 Credit: 104,059 RAC: 0 |
> Excuse the n00bish question, but this program will only run on the *.nix > platform correct? > EDIT: Nevermind.... I got my answer All POSIX (Linux/BSD/UNIX) Operating Systems... :/ |
©2024 CERN