 |
 |
Hi SCCS, ATLAS has a piece of code that is very memory intensive. Depending on the run configuration, it uses anywhere from 4 GB to 30 GB. [The latter is a projection based on observed linear scaling.] There are intense efforts to reduce this memory requirement. In the meantime, we need to produce small samples to keep other activities alive. Would it be possible to use a small number of memfs machines for this purpose? We will start with the 4-GB configuration. There are two things. (1) Allow one user -- John Chapman who is just getting his SLAC account -- to run interactively, if needed, to debug issues that come up locally. He has already run this code at CERN, so we don't expect really basic problems. (2) Allow several memfs machines, say 5 for now, to be usable by jobs submitted through the Grid. There are no doubt issues like advertising the large memory so these jobs would be directed here, preventing "normal" Grid jobs from clogging these machines, etc. Thanks! Charlie -- Charles C. Young M.S. 43, Stanford Linear Accelerator Center P.O. Box 20450 Stanford, CA 94309 [log in to unmask] voice (650) 926 2669 fax (650) 926 2923 CERN GSM +41 76 487 2069