[Info-vax] OT: obscure PDP11 OSes (even more dinosaury)

[Johnny Billquist]

On 2015-06-16 13:32, Bill Gunshannon wrote:
> In article <a5da$557f3c62$5ed4324a$14524 at news.ziggo.nl>,
> 	Dirk Munk <munk at home.nl> writes:
>> The PDP-11 was a 16 bit architecture. Writing larger programs on a
>> PDP-11 always involved using overlay. I've spend many hours trying to
>> construct my Cobol programs in such a way that I could use memory
>> resident overlay. Otherwise it would become disk-resident overlay, and
>> that involved lots of extra disk IO as the name suggests. That would
>> make an application a lot slower. Using RMS in supervisor mode also was
>> a must to keep a somewhat fast application.
>>
>> The 32 bit VAX/VMS architecture didn't have those problems. The
>> application would fit in memory, no need for overlay.
>>
>> So perhaps the PDP-11/74 would have been faster in pure processing
>> power, for real world applications the much larger memory footprint a
>> VMS application could have was far more important. Anything that can be
>> done in memory is much better than having to rely on disk IO.
>
> The PDP-11's memory capability was expanded twice, to 18 and then 22 bits.
> I see no reason why it could not have been expanded again instead of being
> replaced by a totally different architecture.  The harder part would have
> been the size of registers but even I can see ways to deal with that and
> I am certainly not in the same class as people who came up with things
> like the Alpha.

The memory addressing expansions changed things for physical memory. 
Didn't do a bit of difference for applications.
It simply helped the OS, along with improving performance for a large 
system with lots of programs in memory at the same time.

Further extensions of the physical address space would not have yielded 
much improvement. Physical memory space was not a limiting resource at 
that point. Virtual address space was, which is what the VAX expanded.

	Johnny