Anyone here running Linux on dual core CPUs 2004

Douglas O'Neal

AIUI, vetting of results for software is never anywhere close to complete...programmers can not exhaustively test all possible test configurations and computing architectures. The best they can do is test a few, let their clients test the rest and deal with problems as they appear.

In the Appendix of LS Dyna's manual, it even says: "...slight differences in results may be observed as compared to the serial and SMP versions of LS Dyna."

Unfortunately, "slight" was not quantified and for my research, "slight" appears to be rather significant.

Yes I have. In fact in some simpler computational models, the MPP version utilising multiple nodes is actually *slower* than the single processor computation on a desktop. This is because additional computational and network overheads are buttociated with decomposing the model for a parallel processing system.

The idea, however, received a cool reception from the IT support staff. Something about incompatible Linux versions and CPU architectures. For the record, I have verified that ia32 Redhat 9.0 LS Dyna produces the same results as x8664 Suse9.2 LS Dyna on both a x8664 CPU running SuSE 9.2 and a Pentium 4 running Red Hat 9.0.

They are a strange lot who appear more eager to shoot down a suggestion than try one. It beggars belief that I might as well convince a jury that they should basically earn their pay, get off their rears and just install the bloody single processor version to see what happens instead of getting into hypothetical debates about Linux-CPU compatibilities...:-