compaq fumbles
Terry C. Shannon
David A Froble <da...@tsoft-inc.com> wrote in message
news:37DC19F2...@tsoft-inc.com...
<snip>
>
> Microsoft, for whatever reason, (including dropping of support by the HW
mfg),
> has chosen to drop Alpha NT. I've not been everywhere, nor seen
everything,
> but I must say I've seen no criticism of MS, and I must ask if anyone
anywhere
> has seen anything at all of this nature? Please do not reply by saying
it's
> Compaq's fault, if you're going to reply, let it be to point out any
criticism
> of MS. That's my real question, has there been any?
Nowhere near enough! Note that CPQ was taking the lead on the Win64 port to
Alpha, even tho' Win64 belongs to Microsoft (how u$oft acquired NT is
another, Mica-flecked, story). From dealing with folks at DEQwest, this
created significant problems. For example, the Q demoed Win64 on Alpha back
in April, but couldn't say squat about when the product would ship, etc.
"That's for Microsoft to say," said the Compaqtians. Hell of a note,
developing and demoing a product and then not being able to talk about it.
Seems to me that Bill Gates read "Tom Sawyer" and did a good job convincing
Compaq (not to mention HP, who has its own Win64 porting effort) to
whitewash the fence.
Terry Shannon
Bill Todd
Terry C. Shannon <sha...@world.std.com> wrote in message
news:FHyyM...@world.std.com...
My counter-question would be "Who (save for disgruntled Compaq customers who
would rather find someone other than Compaq to blame) would have any reason
to criticize MS in this matter?"
MS stockholders? NT on Alpha was a miniscule (read: it makes even 2% look
*big*) percentage of MS revenues. Probably a negligible, though not quite
miniscule, percentage even of NT-specific revenues. Very likely, at best
close to a break-even product for MS, possibly not even that.
MS customers? Certainly none save those using NT on Alpha (who else would
care at all?), and since Compaq was the precipitating influence the fact tha
t MS merely followed Compaq through the exit makes it pretty clear where
this group's anger should be directed.
Contrast this with people who might be inclined to criticize Compaq:
NT on Alpha customers - as just mentioned.
Other Alpha customers: NT on Alpha gave Alpha a degree of industry
credibility far out of proportion to its numbers; now that it's gone,
existing customer Alpha investments - whether running OVMS, Tru64, or
Linux - look a lot more shakey.
Compaq stockholders, at least those who hoped Compaq's investment in DEC
would be something more than a gigantic write-off - which is exactly what
much of it will be if Alpha becomes perceived as a 'legacy' architecture,
regardless of what OSs may run on it.
Dave began his post by stating "Reality is rarely fair...", but then seemed
to equate 'fairness' with 'treating all platforms the same way, regardless
of their contribution to profit' when describing how MS devoted so much more
effort to the Intel platform than to Alpha. I don't understand why this
should be at all surprising, or deserving of criticism: MS in this
particular case was simply devoting its energies to its areas of maximum
profit, in a public and entirely defensible manner, which is what
publicly-held companies are expected to do.
NT on Alpha wasn't important to MS. At most it gave them modest bragging
rights to OS portability and better-than-PC hardware, but it would be really
hard to argue that the absence of an Alpha version would have had any
significant impact on NT's market penetration. Back when NT was created,
Gates and Cutler gave high priority to a portable implementation, and in the
earliest stages its presence on MIPS, PPC, and Alpha might have helped get
it off to a measurably better start, but once it got rolling the sole
significant platform was IA32 (though of course having the 64-bit Alpha code
as a stepping-stone will most likely make the IA64 port much easier).
NT on Alpha *was* important to Compaq, for (at least) the reasons I noted
above. That explains who paid for its development: it had nothing to do
with 'fairness', but rather was the result (at least in recent years) of DEC
(and later Compaq) *wanting* NT on Alpha and MS being willing - perhaps even
happy - to work with them to make it happen (while understandably retaining
fairly tight control over NT itself). You could criticize DEC/Compaq for
being foolhardy, I suppose, but criticizing MS for allowing them to be seems
a tad over the top.
And as for the Mica migration Terry mentions, one should remember that this
too was caused by DEC 'push' rather than MS 'pull' - and that if indeed
questionable technology transfer then took place, DEC not only failed to
take appropriate legal action but at least in some ways seemed to view it as
a partnership benefit. So once again criticizing MS for allowing DEC to be
so foolish is a bit hard to swallow.
I don't like a lot of MS business practices that have come to light over the
years, and have very little respect for most MS software (NT perhaps being
the best of the lot in terms of its stability/complexity ratio, but still
being far from admirable). But in this particular situation, I just don't
see anything that merits the criticism that Dave and Terry seem to feel is
indicated - so I too will be interested to find out what if any specific MS
misbehavior others might wish to point out.
- bill
Terry C. Shannon
Bill Todd <bill...@foo.mv.com> wrote in message
news:7rhnpv$o0o$1...@pyrite.mv.net...
>
<snip>
>
> And as for the Mica migration Terry mentions, one should remember that
this
> too was caused by DEC 'push' rather than MS 'pull' - and that if indeed
> questionable technology transfer then took place, DEC not only failed to
> take appropriate legal action but at least in some ways seemed to view it
as
> a partnership benefit. So once again criticizing MS for allowing DEC to
be
> so foolish is a bit hard to swallow.
The only "push" was DEC's decision to terminate PRISM, which of course ended
Mica, and pushed Cutler to the Western Provinces. DEC of course was damned
foolish for not negotiating a better deal re: the "questionable technology
transfer" than what they ended up getting.
JF Mezei
> MS stockholders? NT on Alpha was a miniscule (read: it makes even 2% look
> *big*) percentage of MS revenues. Probably a negligible, though not quite
> miniscule, percentage even of NT-specific revenues.
At this pont in time correct. However remember that Microsoft wants its
operating system to rule everywhere, and the one place where they have yet to
succeed is in large enterprise applications (hence low percentage of Alpha NT sales).
Removing Alpha NT will slow down MS's attempt to penetrate that high end
market since it will have to wait for IA64 to be ready for prime time. Until
then, NT will be relegated to specific smaller applications or having a whole
fleet of PCs to support in order to acheive the scalability require to support
such a large app workload.
mat...@seqaxp.bio.caltech.edu
>NT on Alpha wasn't important to MS.
Ah, but it was. Not because they made any money on it directly but because
through it they could dance Digital around like a marionette on its
strings. By hooking Digital on this MS supplied crack (crock) they
effectively removed Digital as a possible competitor in the OS market and
also obtained who knows how many business and technology concessions. Not
the least of these was avoiding a lawsuit concerning how Mica turned into
NT.
Having Compaq exit the WNT/Alpha business does not free them from MS's
grasp though, as the Q has a Windows habit of epic proportions.
David Mathog
mat...@seqaxp.bio.caltech.edu
Manager, sequence analysis facility, biology division, Caltech
**************************************************************************
* RIP VMS *
**************************************************************************
Alan Greig
"Terry C. Shannon" <sha...@world.std.com> wrote:
> Seems to me that Bill Gates read "Tom Sawyer" and did a good job
convincing
> Compaq (not to mention HP, who has its own Win64 porting effort) to
> whitewash the fence.
If you read through Michael Capellas presentations on the Compaq web
site buried in the middle of one of the documents is the statement
that Compaq will continue to transfer key pieces of OpenVMS technology
to Microsoft for inclusion in NT. Also stated is that most OpenVMS
customers are only using VMS for historical reasons due to applications
availability. It does grudgingly admit that a very few VMS systems are
the first choice OS in a very limited set of circumstances. I was going
to cut and paste in tbe actual quotes but Compaq's ftp server seems to
be playing up on me at the moment.
I really don't think they have a clue. Do they really genuinely think
they're doing us a favour by dragging us luddites kicking and screaming
from VMS? Someone at Compaq needs a reality check.
> Terry Shannon
>
>
--
--
Alan Greig
Sent via Deja.com http://www.deja.com/
Share what you know. Learn what you don't.
Alan Greig
Alan Greig <agr...@my-deja.com> wrote:
> to cut and paste in tbe actual quotes but Compaq's ftp server seems to
> be playing up on me at the moment.
Sorry, the URL is:
ftp://ftp.compaq.com/pub/strategy/nonstop-ebusiness-strategy.pdf
Peter da Silva
JF Mezei <jfmezei...@videotron.ca> wrote:
> Removing Alpha NT will slow down MS's attempt to penetrate that high end
> market since it will have to wait for IA64 to be ready for prime time. Until
> then, NT will be relegated to specific smaller applications or having a whole
> fleet of PCs to support in order to acheive the scalability require to support
> such a large app workload.
This will possibly give them time to address the inherent scalability issues
in NT that make trying to run a large app workload on a single system image
impossible (things like drive letters, and the use of flat name spaces for
shared libraries, events, and the lack of support for multiple user instances
and multiple instances of any application, and the lack of a "superserver"
mechanism for assigning ports and named pipes to multiple providers of the
same resource...).
--
In hoc signo hack, Peter da Silva <pe...@baileynm.com>
`-_-' Ar rug tú barróg ar do mhactíre inniu?
'U` << <KH> you did technical support for Hell ?
<susan> Didn't we all, in our youth? >:) >>
mat...@seqaxp.bio.caltech.edu
>In article <7rianc$6h3$1...@nnrp1.deja.com>,
> Alan Greig <agr...@my-deja.com> wrote:
>
>
>> to cut and paste in tbe actual quotes but Compaq's ftp server seems to
>> be playing up on me at the moment.
>
>Sorry, the URL is:
>
>ftp://ftp.compaq.com/pub/strategy/nonstop-ebusiness-strategy.pdf
>
Aargh. I think this is what you are referring to. It's enough to give you
a stroke.
Page 22, "Compaq NonStop e-Business", prepared by "Nonstope e-Business core
team."
"For OpenVMS customers, Commpaq will continue investing to ensure long-term
commitment to millions of users and to address requirements for ultra-high
availability in target markets. OpenVMS customers can depend on a new
release very 12 to 18 months. We will engage with and support select ISFs
for targeted growth markets. And for OpenVMS customers desiring an
alternative, we will provide excellent interoperability with Windows NT,
based upon our sharing of key OpenVMS capabilities with Microsoft."
Like we care that there be new releases on some arbitrary schedule, maybe
we care more what is in those releases???
Since the Q is not sharing OpenVMS capabilities with IBM, HP, Sun, etc., it
is safe to say that either MS has them by the balls or the Q thinks they
are getting something in return. Note however that the date on this
document is June 1999, and one must wonder if the recent NT unpleasantness
has decreased the Q's intent to give away the family jewels. Also, how
many of the core team still work for the Q?
If I owned Q stock I think I'd be demanding that they compensate me with MS
stock for each such technology transfer.
OpenVMS is only mentioned on pages 16 and 22 of this 24 page tome.
Timothy J. Bogart
Now now, first you have to give them time to actually deal with more
than a handfull of network connections thru the TCP/IP stack...in
retrospect, that is far newer-fangled technology than the concept
of actually having more than one user on a machine....
Chuck Ray
Alpha a bit telling. As I recall Intel shipped actual IA64 chips (to OEMs)
just before Compaq/Microsoft made the announcement to drop Alpha. Perhaps
Compaq saw the numbers of the real chip and realized that NT on Alpha was in
trouble.
Chuck
> Seems to me that Bill Gates read "Tom Sawyer" and did a good job
convincing
> Compaq (not to mention HP, who has its own Win64 porting effort) to
> whitewash the fence.
>
> Terry Shannon
>
>
>
>
>
>
Terry C. Shannon
Chuck Ray <chuc...@erols.com> wrote in message
news:7rlrb5$icl$1...@winter.news.rcn.net...
> Am I the only one that finds the timing of the announcement to drop NT on
> Alpha a bit telling. As I recall Intel shipped actual IA64 chips (to
OEMs)
> just before Compaq/Microsoft made the announcement to drop Alpha. Perhaps
> Compaq saw the numbers of the real chip and realized that NT on Alpha was
in
> trouble.
>
I don't think so... what Compaq saw was the sales numbers of NT on Alpha and
realized that this OS-hardware combination was in trouble. I seriously doubt
that IA-64 will, at least during the next several years, put Alpha at a
performance disadvantage.
Terry Shannon
JF Mezei
> As I recall Intel shipped actual IA64 chips (to OEMs)
> just before Compaq/Microsoft made the announcement to drop Alpha. Perhaps
> Compaq saw the numbers of the real chip and realized that NT on Alpha was in
> trouble.
I seriously doubt that IA64 will have any performance advantage over Alpha. If
Compaq gave up because IA64 wasn't as slow as expected, it means that Compaq,
the company that was able to displace IBM as the "de-facto PC standard" back
in the 80s, has given up on its ability to market Alpha.
I would therefore mean that Compaq has caught the marketing-inability virus
from Digital and is hence doomed.
Sheila&bob/Herbal Gypsy & the Tinker
Well said!
Then they also have to address the issue of more than one application on
a server.
Truly, in a large shop - say more than 25 users (smirk), this is a
problem.
John
<Paul....@compaq.nospam.com> wrote:
> JF Mezei wrote:
> > I seriously doubt that IA64 will have any performance advantage over Alpha.
>
> A report by the Register (http://www.theregister.co.uk) claimed that the
> first Merced silicon is running at 700Mhz with the performance of AMD
> K7.
> Not bad for the first pass, but not too impessive.
>
> Intel has already admitted that when Merced is released, IA-32 code will
> likely run faster on the current generation Pentium follow-on chip.
>
>
>
> +---------------------------------------------------------------------+
> | Paul A. Jacobi Phone: (603) 884-1948 |
> | Compaq Computer Corporation FAX : (603) 884-0189 |
> | OpenVMS Systems Group, ZKO3-4/U14 Email: Paul....@compaq.com |
> | 110 Spitbrook Road |
> | Nashua, NH 03062-2698 |
> | |
> | To reply, remove "nospam-" from the return address. |
> +---------------------------------------------------------------------+
Probably good news for VMS customers who had been advised to migrate to NT
on Alpha (Affinity?) and also have large databases requiring 64bit. Now at
least they can anticipate high end Intel chips running NT to migrate to?
JF Mezei
> Intel has already admitted that when Merced is released, IA-32 code will
> likely run faster on the current generation Pentium follow-on chip.
I am curious. When Digital released the Alpha, did the VAX have a performance
advantage initially? I was under the impression that Alpha gained an instant
performance advantage over VAX.
Chris Morgan
Anecdotally at least, yes. I remember a demo of the 3000 Alpha model
circa '92/'93 (can't remember the proper name) at a time when we had
several giant VAX clusters (one was numerically maxed out in terms of
members) and things like VAXStation 4000 model 90s. The general
reaction watching it run a 3d graphics demo (the Ada demo of a Rubik's
cube spinning and solving - lots of cpu) was "Holy Crap!". Nice build
quality too in those days.
--
Chris Morgan <cm at mihalis.net> http://mihalis.net
Nowhere Man
who is migrating to NT should have his head examined. It is simply the
wrong option at EVERY technical level.
John <JG...@ummc.edu> wrote in message
news:JGore-22099...@207-172-154-126.s63.as23.anp.md.dialup.rcn.com..
.
> In article <37E16490...@compaq.nospam.com>, "Paul A. Jacobi"
> <Paul....@compaq.nospam.com> wrote:
>
> > JF Mezei wrote:
> > > I seriously doubt that IA64 will have any performance advantage over
Alpha.
> >
> > A report by the Register (http://www.theregister.co.uk) claimed that the
> > first Merced silicon is running at 700Mhz with the performance of AMD
> > K7.
> > Not bad for the first pass, but not too impessive.
> >
> > Intel has already admitted that when Merced is released, IA-32 code will
> > likely run faster on the current generation Pentium follow-on chip.
> >
> >
> >
Jerry Leichter
| > will likely run faster on the current generation Pentium follow-on
| > chip.
|
| performance advantage initially? I was under the impression that Alpha
| gained an instant performance advantage over VAX.
If you want to do a fair comparison, you'd have to compare VEST'ed code
running on the first Alpha to the original native code running on the
same VAX. If I remember right, the result was usually close to a draw,
though in some cases you could get significant speedup.
You could probably construct fairly natural examples - heavy string
processing would be a good place to look - where the translation
wouldn't be able to match the original (though a recompiled version
probably would).
Even then, there's a complication: The little-discussed fact that some
code (COBOL-generated code is the usual example) ran more slowly on
nominally faster VAX chips than on early "full" machines. The problem
was that the VAX chips omitted some of the string-oriented instructions.
When the compilers generated code using those instructions, the cost
could be quite large. Even when the compilers learned to avoid those
instructions that were emulated on all later VAX chips, they still
didn't quite match what the older hardware did. In some contexts,
"CISCiness" wins! Such would probably do (relatively) even more poorly
on an Alpha.
-- Jerry
Arne Vajhøj
> | I am curious. When Digital released the Alpha, did the VAX have a
> | performance advantage initially? I was under the impression that Alpha
> | gained an instant performance advantage over VAX.
>
> If you want to do a fair comparison, you'd have to compare VEST'ed code
> running on the first Alpha to the original native code running on the
> same VAX. If I remember right, the result was usually close to a draw,
> though in some cases you could get significant speedup.
>
> You could probably construct fairly natural examples - heavy string
> processing would be a good place to look - where the translation
> wouldn't be able to match the original (though a recompiled version
> probably would).
Back then I wrote a note about porting from VAX to Alpha and
among many other things made a small test about VEST.
The results were:
----------------------------
Efficiency chart:
VAX 4200 DEC 3400
(rated at 5 VUPS) (rated at 110 Specmark)
VESTed native
integer 13240 2260 430
calculations
floating point 36510 2470 420
calculations
character 25420 8500 2250
moves
Numbers=milliseconds of CPU usage on specified operation on small data.
----------------------------
The whole note including the test-program can be found at:
http://www.levitte.org/~ava/aftp/VMS/VAX2AXP/VAX2AXP_REVISED.TXT
Arne