[Info-vax] Development Tooling (was: Re: Opportunity for VSI?)

[Scott Dorsey]

Phillip Helbig (undress to reply) <helbig at asclothestro.multivax.de> wrote:
>In article <pv65gm$3um$1 at dont-email.me>, Stephen Hoffman
><seaohveh at hoffmanlabs.invalid> writes: 
>
>> Scientific computing was migrating off of VAX/VMS in the late 1980s and 
>> into the 1990s, as the VAX prices and the performance became 
>> uncompetitive.
>
>Right.  I was there.
>
>> The Alpha price and performance brought some of that activity back to 
>> DEC, though not to OpenVMS Alpha.
>
>Right, because DEC said that unix was the future.  People who moved to 
>DEC unix soon moved to Linux, and DEC is no more.

People in the scientific computing community were running code that was
mostly pretty portable, being written in fortran.  Consequently, they
bought whatever the fastest computer was that week.  The week that it was
a vax with an FPS array processor, they bought that.  The week that it was
an Alliant FX-8, they bought that.

For the most part these people didn't care about the programming environment,
they didn't care about the support environment, they were often willing to
put up with the most godawful of development tools as long as they could get
their fortran code to run a tiny bit faster.

These people are not brand-conscious and they are not os-centric, and they
have to be re-sold all over again on every product.  They are a hard sell.
This is why IBM totally abandoned that market after the 360/168. 

>With x86, performance is just as good as with other operating systems.  
>I remember when people who didn't know VMS at all wanted to work on my 
>machine because the compilers were better.  Perhaps VSI can revive that 
>tradition.

The scientific computing people aren't running on straight x86 systems
so much any more... GPUs pressed into coprocessors and massive arrays of
shared memory x86 systems are fairly common today.  So if you want to be
in that market, your OS needs to be able to support these things.  But
don't be surprised if you put a lot of money into supporting them to discover
that next week you need to be supporting something totally different because
something else has been invented to run fortran code 10% faster.  It is a 
fickle and very expensive market to be in.
--scott
-- 
"C'est un Nagra. C'est suisse, et tres, tres precis."