On Wed, 12 Sep 2007 12:48:31 -0500, McKown, John wrote: > >From what I understand, PRSM calls these things "logical CPs" and >assigns each LPAR a number of "logical CPs". It then dispatches the >"logical CP" on a physical CP. I don't know when PRSM assigns a logical >CP to a physical CP. It may be at LPAR activation, or may be dynamic.
A logical processor is not "assigned" to a physical processor. It is dispatched on a physical processor, and not necessarily the same one as it was dispatched on the last time. However, PR/SM does try to dispatch a logical processor on the same physical processor to improve the probability that there will still be useful L1 cache data. >But within a single LPAR one physical CP will map to at most one logical >CP. Therefore the number of logical CPs which may be assigned to an LPAR >cannot exceed the number of physical CPs. Sounds logical, but since there is no real mapping of logical to physical processors, I'm pretty sure it's not correct. Someone please correct me if I am wrong, but as I understand it, processors are dispatched individually, not in groups. I see no reason why PR/SM can't support more logical processors for a partition than there are physical processors, though it would not be desirable to do so. It sounds to me as if this is actually what was happening in this case. -- 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
