XPost: alt.comp.os.windows-11   
   From: robin_listas@es.invalid   
      
   On 2025-02-22 22:58, Mark Lloyd wrote:   
   > On Fri, 21 Feb 2025 22:37:50 +0100, Carlos E.R. wrote:   
   >   
   >> On 2025-02-21 19:40, Mark Lloyd wrote:   
   >>> On Fri, 21 Feb 2025 12:11:53 +0100, Carlos E.R. wrote:   
   >>>   
   >>> [snip]   
   >>>   
   >>>> Yes, an optional separate coprocesor was used initially. If it was not   
   >>>> inserted, an operation trying to use a coprocesor instruction would   
   >>>> raise an exception, which would be attended and the instruction   
   >>>> emulated in software by the program, which was significantly slower.   
   >>>   
   >>> IIRC, the 286 was the first to have that exception, although there were   
   >>> programs to emulate it on older systems. If I remember right, the turbo   
   >>> C compiler would replace every floating point instruction with a call   
   >>> to a routine that checked for a FPU. If it found one it would replace   
   >>> that call with the FPU instruction and return to a few bytes earlier to   
   >>> run that instruction, otherwise it would emulate it.   
   >>>   
   >>> [snip]   
   >>>   
   >>>   
   >> No, the 8087 also had that exception, it was used by Borland's Turbo   
   >> Pascal series.   
   >   
   > Did you mean 8086 or 8088? 8087 doesn't make sense here, since an   
   > exception would be for when that chip is NOT present.   
      
   Right, the 8086/8 generated the exception on finding an instruction for   
   the coprocesor and there was no coprocesor.   
      
   >   
   > I found my Intel data book for processors through the 286, and the   
   > exceptions are:   
   >   
   > 0 - Divide error exception (divide by 0)   
   > 1 - Single step interrupt   
   > 2 - NMI interrupt   
   > 3 - Breakpoint interrupt   
   > 4 - INTO detected overflow exception   
   > 5 - BOUND range exceeded exception (80186+)   
   > 6 - Invalid opcode exception (80286+)   
   > 7 - Processor extension not available exception (80286+)   
   >   
   > BTW, with the 8086/8088 Intel listed interrupt numbers 5-31 as reserved.   
   > Microsoft ignored that and used interrupt 5 for print screen. I don't   
   > remember how they dealt with that on the AT (maybe they just didn't use   
   > the new BOUND instruction).   
   >   
      
      
   --   
   Cheers, Carlos.   
      
   --- SoupGate-DOS v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)