>I imagine that if MSUs are derived from some standard workload test, then the >answer is yes.
Representative benchmarks are an oxymoron. IBM runs LSPR, but NOT for all models in a processor family. Unless things have changed, both above, and the fact that they user linear relationships to estimate values, when all the data points to a curve, makes the figures suspect. I've been using LSPR, since it was still presented in a ZZ manual, in the early 1980's, and while it had flaws, it helped understand processor capacity. Thrice, it had major flaws: 1. When it was used to OEM-bash. 2. When they suddenly dropped TSO and added WebSphere, on the zBOX. There was no bridge to previous processors. IE: No TSO on the new box, and no WebSphere on the old box, so you could not compare, and there was absolutely no justification for a throughput statement for WebSphere, specifically. In scientific experiments, that's called missing a step. 3. The linearisation of exponential regression curves. All three make results suspect. >If the marketing people mess with MSUs *within* a single processor line then >all bets are off. They do. It's called the technilogical dividend. - I'm a SuperHero with neither powers, nor motivation! Kimota! ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html