On Thu, 17 Aug 2006 10:03:51 -0400, Alan Altmark <[EMAIL PROTECTED] > wrote:
>On Thursday, 08/17/2006 at 08:36 EST, Brian Nielsen ><[EMAIL PROTECTED]> wrote: >> That sounds like your HSA size was based on the actual defined devices , >> not the maximum that can be defined. >> >> This makes me question the utility of the MAXDEVS parameter. Why woul d >I >> bother limiting the number of devices if it has no effect on the HSA >size? > >It certainly affects the size of the HSA *IF* that particular IOCDS is >marked to allow dynamic updates. If you don't specify MAXDEV=, it >defaults to the maximum number of devices allowed by the processor. Tha t >means your HSA will be as large as it can possibly be. Does that mean that for a dynamic IOCDS I can change the size of the HSA by means of the MAXDEV parameter? For example, if I have only have 4K devices can I lower MAXDEV from 63K to 20K and thus recover some memory from HSA and still leave lots of room for dynamic I/O changes? >In fact, you get a >warning message from IOCP if you don't specify MAXDEV= on a dynamic IO CDS. > >If the IOCDS is not dynamic, then (a) you get a warning if you specify >MAXDEV= and (2) the I/O portion of the HSA will be only as large as ne eded >to hold the I/O config you coded, with no room for anything else. Yes, I knew that MAXDEV is only valid for a dynamic IOCDS. Brian Nielsen
