[Info-vax] OT: news from the trenches (re: Solaris)

[johnwallace4 at yahoo.co.uk]

On Friday, 13 March 2015 14:20:05 UTC, li... at openmailbox.org  wrote:
> On Fri, 13 Mar 2015 06:28:17 -0700 (PDT)
> johnwallace4--- via Info-vax <info-vax at rbnsn.com> wrote:
> 
> > What is it *specifically* about "other stuff" that makes "other stuff"
> > high end and AMD64 (and compatibles) not high end? [Preferably without
> > including IBM-style mainframes, but...]
> 
> I read the last part and forget the first part.
> 
> Because POWER is made by IBM and they are known to put out good everything.
> SPARC was made by Sun and TI who also put out good stuff. Intel servers can
> be made with anything, anyone can buy the chips and build a whitebox server.
> 
> Premium designs have cleanliness of architecture (and I can't overstate
> the importance of this as a trickle down cause of quality through and
> through), top build quality, better management features (maybe commodity
> servers have caught up, I don't have any new ones to look at) and attention
> to upward compatability.
> 
> For example SPARC32 code runs just fine on SPARC64 because the 32 bit
> version had enough registers from the beginning (proper design) and can use
> full length 64 bit regs in 32 bit mode on 64 bit hardware (also well
> thought-out). You only need to go 64 bit SPARC for extended addressing
> range. Of course it's running UNIX so it still suffers from the same ABI
> bleedover issues as other commodity OS on Intel and the same brain-dead
> limitations of not being able to mix 32 and 64 bit code mentioned in my
> previous email. The ABI difference is less radical on SPARC than Intel but
> it is still not insignificant.
> 
> A lot of this is only meaningful for people who care because they write
> system code or work on the boxes. If you're just running end user
> applications and never see the hardware or pop it open then I agree Intel
> is good enough. But it will never be premium.
> 
> -- 
> Please DO NOT COPY ME on mailing list replies. I read the mailing list.
> RSA 4096 fingerprint 7940 3F02 16D3 AFEE F2F8  ACAA 557C 4B36 98E4 4D49

OK, thanks for that. Rather terse answer here as I need to be elsewhere...

1) I do see where you're coming from. 

2) Much x86 carp is, as you acknowledge, hidden from most users, admins,
programmers. But not all. Depends on hardware and depends on OS, and
depends on your area of interest.


3) You're new to VMS. Welcome. Before drawing conclusions on how much
of this legacy-x86 baggage will be visible in a nuVMS world, maybe get
yourself a bit more familiar with how the architectural differences
have been hidden (or not) in the previous VMS ports? Maybe? VMS is not
UNIX. Compatibility has been a long term goal (though sometimes long
term compatibility brings challenges too; a change of platform could
be a good time to look at some of those things).

And then once you know how much platform-specific stuff has historically
been (in)visible, we can start guessing how much it will *matter* to the
people with the chequebooks.

A decent HPQ Proliant seems to be "good enough" for a lot of people with chequebooks. And in comparison with generic x86 stuff, Proliant kit
would be considered "premium" by many (especially in terms of manageability
and so on). All it needs is a decent OS.

Whether Proliant will be the badge on VSI-supported boxes is a different question. But I'd be very surprised if the innards didn't have a lot in
common.

Must go.

Have a lot of fun.