[TUHS] Bell Labs Holmdel site coming down

[Peter Jeremy]

On Wed, 2006-May-17 15:30:47 -0700, Stuart, Jon wrote:
>I also am very disappointed about the abandoning of the Alpha chip.

Likewise.  And on a related subject, FreeBSD dropped its Alpha support
(from the development branch) earlier this week - it had fallen below
critical mass - and a lot of this would have been the death of the
underlying chip.

>From it's start I was very impressed with it.  It was a very good RISC
>architecture, and the first to really do 64-bit computing, and do it
>well.  Before they decided to kill it, it was still the best
>architecture for 64-bit computing on the market.  

I can't think of any other architecture where the designers considered
what they needed to do to make the architecture future-proof.  The
normal architectural design criteria are a mixture of the number of
transistors they can fit on a chip today and backward compatibility.

>so presumably OSF1/Digital UNIX/Tru64 is either dead or end-of-lifed as

It's effectively end-of-life.  There will be no future releases, though
HP will support it until about 2011.

>It's unfortunate that it seems we must resign ourselves to a future of
>x86-based OSs, such as Linux, or even Open/Free/NetBSD, which aren't
>really UNIX (Linux definitely isn't, and the modern BSDs have changed
>enough that they also aren't IMO).

If Linux and *BSD aren't Unix, how do you define Unix?  (Other than
having paid TOG the trademark licensing fees).

>PowerPC, and VAX.  For x86 to win, really shows that the quality of
>technology in a product really has no bearing on how it will do in the
>market.

That is a surprise to you?

>  It's not about quality, it's about profitability, and they are
>very often not the same.

There's a bit of a feedback loop: Starting with the IBM-PC, x86 sold
in large volumes, so there were lots of profits and design costs could
be amortised over a larger volume, allowing more man-hours to be
invested in the next generation whilst still returning a profit.  This
makes the next generation of x86 outperform the competition at a lower
price - encouraging more people to use x86.

>While IA-64 is based on the PA-RISC, it's still Intel,

I suspect the IA-64 will quietly fade away.  It hasn't lived up to the
hype and even Intel seem to acknowledge this by licensing the amd64.

>and lots of RAM (the fastest machine SunOS 4.1.4 could run on -- and
>I've heard that 4.1.4 did have very alpha SMP support, similar to what

SMP support started earlier than 4.1.4.  The sun4m machines (SS470,
SS670) were the first SMP machines and ISTR they were supported in
the last 4.0.x or early 4.1.x releases.  (I was using them at the
time but the details have faded a bit after 15 years).

>Linux and the modern BSDs used for a long time, that being a "big giant
>lock" [mutex] around the kernel).

Most early SMP systems worked this way - it's relatively easy to
implement and gives good CPU utilisation on CPU-intensive tasks (that
don't need the kernel much).

-- 
Peter Jeremy