Re: OpenVMS Itanium port progressing well says Gorham!

From: Dave Weatherall (djweath_at_attglobal.net)
Date: 04/29/03 

Date: 29 Apr 2003 05:18:51 GMT

On Sun, 27 Apr 2003 06:52:44 UTC, "Bill Todd" <billtodd@metrocast.net>
wrote:

>
> "Dave Weatherall" <djweath@attglobal.net> wrote in message
> news:DTiotGxQ0bj6-pn2-MuUErke9ylPS@localhost...
> > On Thu, 24 Apr 2003 18:07:57 UTC, JF Mezei
> > <jfmezei.spamnot@vl.videotron.ca> wrote:

> > > Consider this: Had they not killed Alpha, how much further ahead would
> VMS
> > > have been today, and how much further ahead VMS would have bveen 1 and 2
> years
> > > down the road with the resources allocated to improvements instead of a
> port ?
> > >
> > > The port should be viewed as a necessary evil caused by a mistaken
> decision to
> > > kill alpha. It should not be viewed as something good for VMS. There is
> > > nothing outstanding about IA64, nothing that will give VMS a
> technological
> > > edge over competitors such as HP-UX or Tandem, something it had with
> Alpha.

> > Hang on JF. Porting VMS to another platform is _not_ a/the mistake.

> I'm afraid JF's vision is closer to reality than yours is here: you are
> responding with abstractions without evaluating actual trade-offs.

Well you're right, I was not defending the specifics of the chosen
platform and I certainly was not defending the killing of Alpha.
 
> > It would be good for VMS.

> Abstractions such as that one. In some ways, even a port to a completely
> irrelevant platform 'would be good for VMS', in the sense that it would help
> clean up the code to make additional ports easier.

Point taken about Abstractions. I missed the structure on first
reading. Yes that is the/my point. How could one get VMS onto another
platform without moving the architecture dependencies down the code
chain. The port to Alpha did a lot in that area. Moving to IA32 or
IA64 or Opteron involves more.
 
> But would it be a good use of resources? No way: there are far more
> important things to do, and as JF noted if it weren't for the port (and in
> particular the immediacy of its need, because the Alpha boats *are* burning
> and the sooner VMS can stop depending on them the safer it will be) some of
> those more important things could be being done (e.g., material OS advances
> that would both help VMS compete and demonstrate its owner's commitment to
> its long-term future), while porting work could proceed more leisurely in
> the background just in case supporting additional platforms became important
> down the road.

Again I don't disagree, absolutely, with that point of view. I do (try
to) keep the principle (abstraction) and the actualities separate
though. On the other hand, I'm not a good person to comment on that
area (new developments to match the competition) because I've only
just been able to drop VMS 5.5-2 as my lowest common denominator
platform. I am not on the cutting edge and don't miss much from VMS.
Crikey, I still, by choice, have a 3100/M76 on my desk instead of an
AlphaStation.

> Dave D. and others have been asking for it for
> > a long time now.
 
> Not quite. What people have been asking for is a port to a commodity
> platform - in particular, IA32. That is not what you're getting: whatever
> market share Itanic may slowly acquire, its volumes won't be a drop in the
> bucket compared with IA32's (or AMD64's) - and consequently its pricing will
> be closer to Alpha's (remember, a DS10 cost only about $6K, and API proved
> that it could be considerably less) than to a PC's.

Well while I would accept that a port of 64-bit Alpha/VMS to a 32-bit
IA-32 would be a waste _now, a port from VAX 32 bit to IA32 _then,
would have been completely different. If one then accepts the argument
that a low cost 64-bit workstation running VMS is desirable, Compaq/HP
would today have had the choice between producing low-cost (low-speed)
Alphas, IA-64 and now Opterons.

> I would like to be using at home but then I'm a VMS
> > bigot so we can't map that desire onto the Home computing population
> > at large. As you point out below, the mistake is setting alight to the
> > boats before you've reached the opposite shore.
>
> Again, not quite on target. It's not sufficient merely to reach shore, you
> then have to determine that said shore is so attractive that there's no need
> for all the things you were used to back home.

Absolutely, the route chosen by Compaq, with or without HP connivance,
is just plain dumb.

> If Compaq had kept the Alpha team and actually backed Alpha, that never
> would have happened: Intel would have been stuck with a power-inefficient
> architecture that it had no idea how to push forward and that therefore
> would never have looked attractive compared with Alpha (though AMD64 still
> might, for the low end), and EV8 would be coming next year to solidify
> Alpha's high-end leadership. The current situation with the engineering
> forced-march port is no favor to customers.

Again no disagreement. However, one of Fred's arguments for IA64 is
that it provides CHumPaq with a route to provide Reasonable, if not
Low, cost workstations. (Yeah, I too feel that he has contradicted
himself on occasion in this area). Now you yourself have accepted that
EV7 and above are aimed more at the multi-CPU Server market and
perhaps too expensive for the workstation roll. The low cost w/s is a
reasonable/sensible aim.

In the end, unless HP do somthing to market VMS, our debate is
somewhat moot. I've just come back from the reading room where I was
perusing the April edition of Dr. Dobbs. It contains 6 pages of HP
advertising about its support for Linux, not to mention the back page
advert for its Itanium-2 workstations.

Gotta go to work. 

-- 
Cheers - Dave.
PS. While writing this I was think about the post from Alan Greig 
where he argued, before alphacide, that a port to Itanium was a 
'must'. He's not been around for a while. I hope he's OK.

Relevant Pages

  • Re: Rookie on open VMS operating system
    ... The Alpha is just a microprocessor architecture: ... TRU64 unix and it supports Windows NT4. ... most current drives are supported by VMS. ... VMS also runs on a fourth platform: ...
    (comp.sys.dec)
  • Re: If Not OpenVMS / Alpha Than What ???
    ... Heck, as a hobbyist facing a vax to alpha migration, the problem isn't ... Moving to a platform which has about 2 years left in its active life ... VELO's system disk, today at 7.3 on a 4000-600 ... VMS used to be known as a very rich operating system because it came ...
    (comp.os.vms)
  • Re: Intel prepares to kill off the Pentium 4
    ... port VMS to industry standard platform. ... This would not contradict your above statement :-) And once HoffmanLabs announces the existance of this project, your sale of VMS licences would instantly exceed those of HP. ... will Intel/Amd at one point start to "enforce" UEFI for the X86 platform? ... but the work involved in this particular area of an operating system port is a rounding error in the context of the effort involved in a porting project for something of the scale of OpenVMS. ...
    (comp.os.vms)
  • Re: Anyone interested in building a vms-like OS?
    ... And then get sued by HP if any of the code looks anything like that in VMS ... since the Engineers you poached had access to the VMS sources. ... abandoned the OpenVMS platform. ... Microsoft doing the port then HP would probably be bought off for a pittance ...
    (comp.os.vms)
  • Re: If Not OpenVMS / Alpha Than What ???
    ... Alpha, implemeting it on alpha and then designing a structure that can ... If you move from VMS to anything else, the effort will be many times any VMS to VMS port. ... be done to fully move from one platform to another. ... When creating a totally new system disk on VMS, ...
    (comp.os.vms)