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

[Johnny Billquist]

On 2015-06-22 17:29, David Froble wrote:
> Johnny Billquist wrote:
>> On 2015-06-15 15:36, Simon Clubley wrote:
>>> On 2015-06-15, Dirk Munk <munk at home.nl> wrote:
>>>> bill at server3.cs.uofs.edu (Bill Gunshannon) wrote:
>>>>>
>>>>> Of course, all of this still doesn't explain how the VAX was
>>>>> "a best-seller" when compared to the PDP-11 when the PDP-11
>>>>> sold 50% more systems.
>>>>>
>>>>
>>>> Because one VAX could do the work of several PDP-11's?
>>>
>>> In the early days of the VAX, wasn't there some literature floating
>>> around showing how the PDP-11 was faster at some things than the VAX ?
>>>
>>> This was before my time however so I may be wrongly remembering
>>> something I read in the past.
>>
>> Interrupt response times on the PDP-11 are way better than on VAXen,
>> which led to people not wanting to downgrade from a PDP-11 to a VAX...
>>
>>     Johnny
>>
>
> Swapping can be rather detrimental to observed interrupt times ....

You cannot have interrupt handling done by something that is swapped 
out. That is pretty much impossible... Interrupt vectors points to an 
address that is expected to handle the interrupt. Not being able to 
handle the interrupt until some paging have happened creates a 
chicken-and-egg problem, since until the interrupt have been handled, 
you cannot drop the IPL level. But doing paging implies doing disk I/O, 
which in turn causes interrupts, which you need to be at a low IPL level 
in order for those interrupts to come through...

(Hence interrupt handlers are always memory resident, and in RSX, if you 
have a task that actually handles interrupts, which you can, locks that 
task in memory as well.)

> When your program is sitting in memory, because it can, things are
> usually better.

That is always true.

> Ok, I'm not saying there weren't some things the PDP-11 did better, when
> it wasn't over burdened with more tasks than the memory could handle.

Agreed. But interrupt latencies were/are much better on the PDP-11. 
Partly because an interrupt is a very heavy thing for a VAX do deal 
with. Lots of stuff to push and set up to handle an interrupt on a VAX. 
The PDP-11 have a much more minimalistic approach, which makes it both 
faster, and way less complex from a hardware point of view.

	Johnny