[Info-vax] Dave Cutler, Prism, DEC, Microsoft, etc.

[Bob Koehler]

In article <hejq08$nf0$1 at usenet01.boi.hp.com>, "FredK" <fred.nospam at dec.com> writes:

> Nobody really 
> knows what a 3GHz VAX with modest changes to the architecture might be 
> capable of today.
> 
> But all that is beside the point.  I also said in a different point that I 
> believe that essentially the VAX running out of gas was inevitable.  Even if 
> it was made faster, the 32-bit VA space was a limitation for the servers - 
> and technical workstations.

   At the time Alpha was introduced, only a few applications were bound
   by 32 bit address space.  I know the C compiler supports 64 bit
   addressing on Alpha, and the Fortran compiler was promised (was it
   delivered), but I think most of the native language compilers still
   don't.

   But just as the 18 bit PDP-10 was updated to 23 bits, and the 32 bit 
   IA32 was extended to 64 bits, adding extended addressing to VAX would 
   have been a straightforward thing.  Not all single byte opcodes were 
   used up, and very few two byte opcodes were in use.  Vector instructions 
   were added using two byte opcodes, and 64 bit addressing could have been, 
   too.  Even longer opcodes were possible, extending the same scheme as
   two byte opcodes.

   But DEC insisted that customers didn't want a 64 bit VAX, and did
   need to listen to the wind which was clearly blowing RISC.  Only
   Intel seemed to resist that wind.