On Mon, 4 Feb 2008 15:36:29 -0600, Kelman, Tom wrote: >In run time, yes. In CPU time, I wouldn't think so.
In practice, there is seldom a large difference in CPU time utilization, but the OP asked about the maximum when there were different workloads. When an instruction references memory, that instruction executes a *lot* faster if the memory location is in cache. If the processor has to access main store it runs a lot longer. The time that it takes to execute that instruction is CPU time. In a worst case scenario, a large number of cache misses could increase CPU time by quite a bit. Imagine, for example, a very low priority job running in a very low priority LPAR on a CEC that has far more logical processors than physical processors. When the system is lightly loaded, it might run just fine. When the LPAR and the CEC are very heavily loaded, your job might not get to run long before being interrupted, and it might have to wait a long time to get redispatched. During that time, many of it's cache lines may have been used for other purposes, forcing it to make more references to main storage. -- Tom Marchant ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
