[Info-vax] VAX vs. MV/8000 [was Re: Hard links on VMS ODS5 disks]

[Dan Cross]

In article <0f65bba5-93d1-469f-8bc9-9e9d91da6d76n at googlegroups.com>,
terry-... at glaver.org <terry-groups at glaver.org> wrote:
>On Sunday, August 27, 2023 at 8:46:01 AM UTC-4, Dan Cross wrote:
>> Again, this is almost exactly what x86 does today. x86 is 
>> otherwise a dumpster fire of bad design, but the approach they 
>> took with the 64-bit extensions is spiritually closer to what 
>> DG did with Nova, Eclipse, and MV than what DEC did. 
>
>Intel has proposed X86-S which jettisons most of the 32-bit and
>16-bit parts of the architecture. That might let them clean up their
>microarchitecture (or at least provide a larger microcode patch
>area 8-).

With some caveats... They will retain 32-bit userspace support,
but yeah, the kernel will be 64-bit only and real mode will go
away.  Note, however, that the 32-bit instruction encodings for
using 32-bit operands are not affected.  That is, the opcode for
`xorl %eax, %eax` is not changing.

>I expect there will be unpleasant surprises if S-mode processors
>make it to operating system developers (outside of "most favored
>nation" ones like Microsoft).

Outside of people still trying to run DOS programs (frankly,
those folks are better off with whole-machine emulators like
DOSBox anyway), most users won't notice.  People like me will
have to care a little bit, but the biggest source of changes
will be in very, very early boot.  A more rational startup
sequence will be kinda nice, though they are making noises
about ancilliary processors (e.g., to set up the initial page
tables that the BSP will beginning executing on) that need some
elaboration.

	- Dan C.