home bbs files messages ]

Forums before death by AOL, social media and spammers... "We can't have nice things"

   alt.os.development      Operating system development chatter      4,255 messages   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]

   Message 2,784 of 4,255   
   antispam@math.uni.wroc.pl to Joe Monk   
   Re: PDOS/86   
   12 Aug 21 22:59:01   
   
   Joe Monk  wrote:   
   >   
   > > Well, as I wrote. it is possible that 360 architects thought that   
   > > 360 does not need relocation register. But but practice showed   
   > > that base registers do not work for relocation.   
   > >   
   > > I am not sure what you wanted to say in second sentence: 360/67 and   
   > > 370 series suport paging, so they really have no need for relocation   
   > > register. But for other 360 models, there is no replacement and   
   > > relocation register would be quite useful.   
   > >   
   >   
   > The 360/370/390/z/arch memory model has been flat since day 1.   
   >   
   > The concept of relocation was handled by the way the 360 object format was   
   done. For programs that were too big, they would be written as overlays ...   
   pieces of code that would not be loaded unless needed.   
      
   360 format handled loading at aribtrary address.  Relocation register   
   is about moving program in real memory.   
   > Remember, for OS/360, there were two different modes of operation: MFT   
   (multitasking - fixed # of tasks) and MVT (multitasking - variable # of   
   tasks). In the IBM mainframe world, a program of any size can be run by what   
   is known as an initiator. The    
   initiator processes the JCL and looks at EXEC statements for a REGION=   
   parameter. That says how much memory a given job step needs.   
      
   That is fine if you run program from start to end.  But if you need   
   to suspend program (normal thing in interactive use), then it   
   in natural to swap such program to disc and reload in different   
   place.  But OS/360 did not support this.  TSS-360 tried, but   
   the project was failed.   
      
   Even in batch system relocation could be quite useful: MVT   
   region had to allow for _maximal_ use, and were frequently   
   much larger than strictly needed.  With relocation it would   
   be possible to enlarge regions, so much smaller initial   
   region would be OK.  IBM tests indicated that paging   
   (going from MVT to MVS) allowed similar load using 3-4   
   times less memory.  Flexible regions would not give that   
   much, but clearly would give large saving compared to   
   MVT.   Of course, once 370 got paging there was no   
   reason for relocation register, but it would make   
   sense for original 360.   
      
   --   
                                 Waldek Hebisch   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)   

[   << oldest   |   < older   |   list   |   newer >   |   newest >>   ]


(c) 1994,  bbs@darkrealms.ca