[Info-vax] Unix on A DEC Vax?

[David Froble]

Stanley F. Quayle wrote:
> Hoff, I don't understand what you mean when you say an emulator is
> slow.  The various commercial emulators are faster than their
> physical counterparts.  No, it's not a VAX running with a 3 GHz
> clock, but it is still faster.

That's a vague statement, and I'd have to see some details.  If the 
physical counterpart is a MicroVAX 2000, yeah, even snails win that 
race.  But if you're talking the last generation N-VAX, say a MicroVAX 
3100 Model 98, maybe not.  Also, as you state below, it depends upon 
what's being done.  Disk activity will be closer to hardware speeds, but 
as for number crunching, I think the emulators are toast.

If you're talking say an Alpha EV6 or EV7, I'm going to have to say 
"show me", and even then I might not believe what I see.

> The important part is how well the customer's application runs.

Oh, so very true!

>  A
> lot of the big wins are in improved disk speed.

Of course, once the hardware is activated, the CPU isn't much involved.

>  I have a customer
> (www.stanq.com/wf1) that reduced their backup time by a factor of 8.
> The only complaint was that the users thought the system was "too
> fast".

Come now, that's not a complaint.  :-)

> As for an emulator being a temporary fix, there are lots of customers
> that rely on corner-cases in VAX floating-point operations.  They're
> stuck on an emulator.
> 
> There are customers that have instruction-time dependence.  There are
> hardware emulators for them (at least for VAX and PDP-11).
> Expensive, but if they really need it, they need it.
> 
> And there's a big chunk of my customers that want to get new hardware
> but don't want to change anything.  Re-training users for some new
> application can be really expensive.  And some have tried to develop
> a Windows or Linux replacement, and failed horribly.

This is something so many want to ignore, or talk away as not 
significant.  Convert!  To what?  Re-write!  Are you nuts?  Sometimes 
the applications are very "right" for the users, and they are not a big 
enough business to consider the effort of porting / re-writing in 
another environment.

Another issue is support of new devices.  Disk drives for older systems 
just aren't manufactured any more, as far as I know.  Sooner or later 
that's going to be an issue for me.  I don't even want to discuss power 
supplies, as that problem has already taken a huge bite out of my butt.

Damn PCs!  That's all anyone builds for anymore.

However, it has been my experience that the itanic is a valid forward 
path for most VMS users.  Most things just need a re-compile, link, and 
they just work.  Yeah, there will be those who don't have source code. 
That's just a past mistake that is waiting to haunt them.

> The FDA, FAA, DOD, NRC, etc. have been accepting emulators as valid
> replacements for physical systems.  Re-qualifying an application can
> cost thousands, if not millions, of dollars.

I've got to ask, when you replace hardware with new software, how can 
you avoid re-qualifying?  I know how NASA does it when under pressure. 
Too many "do-not-fly" problems?  Easy, change the definitions.  The 
president wants this thing to launch today?  Temperature doesn't matter. 
RIP Challenger.

If a system may cost millions just to re-qualify, then what would that 
system cost to re-write?  How many of these would it take to make 
continuation of VMS and hardware the lesser cost?