Physics Analysis Hardware Project Status

Physics Analysis Hardware Project Status

Ruth Pordes, April 10th

RFPs

The FY99 RFPs for the Physics Analysis Computers for CDF and D0 have been completed and delivered to CDF, D0 and CD management and are on their way through the purchasing system. We much appreciate all the work done by the procurement teams: Mark Leininger and Dave Fagan for both; Liz Buckley-Geer and Stephan Lammel for CDF; Mike Diesburg and Lee Lueking for D0.

Benchmarks

A lot of work was done on benchmarks and moving from the traditional TINY to use of SPEC95 and comparing the results to runs of D0 C++ code TRF++. We learnt that compiler flags have 20-30% effect on benchmark results; that TRF++ scales with SPECINT95 more or less; that using TINY which takes 30 seconds to compile is much easier than using SPEC95 which takes 6 hours to compiler; that we do not completely understand the difference in results between the KAI C++ and Native C++ results for TRF++. Information on the runs is at http://home.fnal.gov/~wike/benchmark/

Disk Purchase

The next phase of the project is to purchase the disk for the analysis servers for FY99. We expect to have a bid for this.

Operations

We will talk about operations of the Run II Systems in the Physics Analysis Hardware Project. There are several complementary efforts starting in the Run II Joint Projects and Computing Division to address operations for Run II and "Next Generation". These will keep in contact but it is expected will have enough separate focus to warrant being distinct efforts at this point.

Trailor and Global Computing

We plan to have some discussions on these issues before the end of the 99 Fiscal Year