ACADEMIA
UK Tops List of EGEE CPU Hours
The UK has come out top in the listing of CPU hours contributed to EGEEII since the project's start. In the four months to the end of July, the UKI region contributed more than 5 million normalized CPU hours. Tony Doyle, GridPP project leader, commented, "This is an excellent milestone for the UK, showing how much effort has gone into establishing and maintaining the GridPP grid. Of the UKI hours, more than a third each went to ATLAS and LHCb, demonstrating how the UK particle physics Grid is being used extensively to prepare for the LHC". 
The chart below shows the normalized CPU hours for each of the EGEE regions, from 1 April 2006 to 31 July 2006. The UK's 5 million hours is more than a third of all hours EGEE-wide. CERN itself was the largest single site, contributing nearly 2 million hours. Showing the growth of Grid use, for the same period in 2005 UKI contributed 1.3 million normalised CPU hours - still the largest regional contribution, but demonstrating a growth of nearly 300% over the year. The three LHC experiments of LHCb, ATLAS and CMS account for more than 85% of CPU hours. As an example of how these CPU hours are being used we can look at ATLAS, much of whose Grid use is currently as part of their Monte Carlo production service.
The ATLAS production produces simulated data consisting of a large number (approx. 10**7) of simulated "events". Each "event" is the result of a collision between two protons, and the full simulation requires four steps: * Event Generation * Detector simulation * Digitization and Reconstruction. Gilbert Poulard, the ATLAS Resource Management Co-ordinator, explains, "This requires running a chain of different programme with different characteristics in terms of memory usage and CPU time consumption. Typically a simulation (long) job runs for around 24 hours while a digitization or reconstruction (short) job runs for 3 to 4 hours". ATLAS's production system can submit jobs to three different Grids: EGEE, Open Science Grid in the US, and ARC/NorduGrid. During July ATLAS has been running production as part of Service Challenge 4, aiming to produce 2 million Monte Carlo events/week, requiring 2700 KSi2K. The number of jobs run successfully on the ATLAS production system in each country during July is below. UK sites contributed 12% of the ATLAS central production jobs, the second largest next to the US, successfully running over 10,000 jobs and using over 4,000 days worth of CPU.
The chart below shows the normalized CPU hours for each of the EGEE regions, from 1 April 2006 to 31 July 2006. The UK's 5 million hours is more than a third of all hours EGEE-wide. CERN itself was the largest single site, contributing nearly 2 million hours. Showing the growth of Grid use, for the same period in 2005 UKI contributed 1.3 million normalised CPU hours - still the largest regional contribution, but demonstrating a growth of nearly 300% over the year. The three LHC experiments of LHCb, ATLAS and CMS account for more than 85% of CPU hours. As an example of how these CPU hours are being used we can look at ATLAS, much of whose Grid use is currently as part of their Monte Carlo production service. 