I'm sorry, Sir Alan, but there really were "AP" (Attached Processor) mode ls in which the base processor could do I/O and the attached processor could not. the IBM 3033 AP was one such model. See http://www-03.ibm.com/ibm/history/exhibits/3033/3033_TR03.html , for example.
"Operating system support for the 3033 Attached Processor Complex is prov ided by IBM's Operating System/Virtual Storage 2 Multiple Virtual Storage (OS/VS2 MVS) or Virtual Machine Facility/370 (VM/370)." Also supported by MVS/SEPP and VM/SEPP. No doubt this was before you were born. Sir Alan the Persevering On Thu, 10 Dec 2009 18:41:41 -0500, Alan Altmark <alan_altm...@us.ibm.com > wrote: The "base" CPU is the first CPU that is defined to your virtual machine. >What it really means is that you can't DETACH it, since every virtual >machine must have at least one virtual CPU. Otherwise the term is >meaningless. > >Each processor has always been able to do its own I/O. What changed (37 0 >-> XA) was the introduction of a "channel subsystem" that enabled any >processor to get to any device. Prior to that, each CPU had its own set >of channels. I/O to *that* device had to be scheduled on *that* CPU. N ow >we're all one happy family! > >Alan Altmark >z/VM Development >IBM Endicott >======================== ========================= ========== ==============
