1 / 12

CMS Software at RAL

CMS Software at RAL. Fortran Code Software is mirrored into RAL AFS cell every 24 hours  /afs/rl.ac.uk/cms/ Binary libraries available for: HP HP-UX 10.20 Linux Redhat 5.1/5.2 Sun Solaris 2.6 Includes: CMSIM versions 113-118 CMSJET PYTHIA

Download Presentation

CMS Software at RAL

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. CMS Software at RAL Fortran Code • Software is mirrored into RAL AFS cell every 24 hours  /afs/rl.ac.uk/cms/ • Binary libraries available for: HP HP-UX 10.20 Linux Redhat 5.1/5.2 Sun Solaris 2.6 • Includes: CMSIM versions 113-118 CMSJET PYTHIA Glenn Patrick, CMS(UK), 30.11.1999

  2. HP BATCH LINUX BATCH HP LINUX SUN FDDI NIS Datastore userids 1 Terabyte Disk Scratch /home AFS RAL Particle Physics Unix Services ? 100 Megabit Switched Network

  3. Platforms & Problems HP-UX • PA7000 and PA8000 chip architectures cause problems: • Updating is done with PA8000 code which only runs on 5 fastest (newest) CSF nodes. • Cannot target particular batch nodes with NQS. Require special queue or recompile according to node. • HP-UX deprecated within CMS. • HP-UX being phased out at CERN and RAL over the next year. • Far greater capacity on Linux & future expansion.  Conclusion …... migrate to Linux.

  4. Predicted RAL-CSF Linux Profile

  5. Linux CMSIM successfully ported onto CSF Linux nodes. BUT • Have to take care to set PATH to pick correct compilers. System compilers are default and these appear in /usr/bin which usually comes before /usr/local/bin which contains the CERN versions of the compilers. • Sometimes differences between the CERN compiler versions on csflnx(egcs -1.1.1) & cmslx( egcs -1.1.2). • Sometimes the CERN library src directories disappear due to ASIS updating - need for GEANT. • Problems with old versions of CMSIM due to different FORTRAN libraries - what do CERN do?

  6. Linux : ATHLON(AMD K7) 600 MHz CPU versus Pentium II 450 MHz CPU Benchmarks

  7. PYTHIA CMSIM116 GEANT3 EXISTS FORTRAN Pileup UNKNOWN Digitisation ORCA OO Objectivity Analysis

  8. OO Software • Investigated setting up RAL as remote developer site and implementing developer spaces for OO projects. • SCRAM installed. • ORCA environment downloaded - Web to AFS. • Some initial security problems. • Some conflicts with FORTRAN updating. • LHC++ software required to build the code!! Need HTL, Objectivity,OpenGL, ObjectSpace, Iris Explorer, Iris Explorer, etc

  9. ODBMS C++ standard library Extension toolkit Geant4 CLHEP HepExplorer HTL LHC++ CMS Software Components

  10. LHC++ • Of benefit to all experiments, so John Gordon took on main task of installing code in local AFS space: • Institute authority for commercial components • /afs/rl.ac.uk/sw/lhcxx • Various problems with operating systems. • Various problems with versions, Exceed, etc. • However, I got to the stage of setting up an LHC++ environment on Linux, defining an Objectivity federation and running Iris Explorer to produce histograms.

  11. Objectivity • Need to understand the technicalities of Objectivity. • Complicated because of number of experiments supported at RAL and possible conflicts over versions and locking mechanisms. • Lock servers, AMS servers, performance. • FDID identifiers (using spare BaBar numbers). • Mass storage and data migration - disk to tape. • BaBar estimate it took 1FTE for their UK work.

  12. Forwards? • Yet to emerge from the LHC++ diversion….. • What about GEANT4, OSCAR, IGUANA, etc? • What about user environment? • Only aiming for minimal system and capability. • Work exceeds available effort - not at all what I had in mind. • Need to see benefit by getting to stage of processing physics events beyond CMSIM. • JIF bid is needed if you want to do this seriously on a large-scale!

More Related