From: muta...@gmail.com   
      
   On Saturday, July 17, 2021 at 11:42:53 AM UTC+10, Joe Monk wrote:   
      
   > > That processor with no visible registers sounds like a   
   > > pie-in-the-sky design to me. You may as well design   
   > > the x64 in 1970. You can do anything on paper.   
      
   > But it's not ... it was actually produced along with an OS, written in ADA.   
      
   Maybe I used the wrong term. You could have produced   
   an x64 processor in 2000 BC, implemented in the form   
   of Egyptian slaves. It would have been ridiculously slow,   
   but it would "work" for some value of "work".   
      
   You can do anything you want for fun, but I haven't   
   heard of anyone taking away registers. The most I've   
   seen anyone do is reduce the number of instructions   
   which managed to produce a better result. Anyone   
   can produce a worse result.   
      
   > > Explain to me why segmented memory is not the   
   > > right approach to solving this problem in an   
   > > environment with severe limits on memory, but   
   > > still more than a single register can address.   
      
   > The right approach, IMHO, has always been linear addressing.   
   > Think about how the mainframe does it ... linear address space,   
   > with an ASID. So you could have x^asid linear address spaces.   
      
   That's not the issue being addressed. What happens when a   
   single application wants to access more than 4 GiB of memory   
   on the mainframe, and the memory is indeed available, but   
   you only have 32-bit registers?   
      
   tiny memory model is great, but eventually you need to   
   move to the other memory models and buy yourself   
   3 segment registers. I don't think you need any more   
   than that. I don't think I have any code that uses more   
   than cs/ds/es.   
      
   > If you remember the z80, then you understand. The z80,   
   > as a 16-bit address bus processor, was limited to 64K of   
   > directly addressable memory. But with bank-switching,   
   > you could put lot of memory on a system, and switch it in   
   > and out. There were many minicomputers in the '80s that did that.   
      
   I think you're right - that would work too. But those   
   different banks are basically just segments themselves.   
   The model is still the same. Just a different way of   
   loading a segment register.   
      
   And that buys you what, compared to a proper segment   
   register that allows fine-grained segment shifts to   
   enable packing?   
      
   BFN. Paul.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)