HEADSUP: Alpha support is being retired in 7.0
Paul Allen
nospam at ugcs.caltech.edu
Fri May 12 17:29:59 UTC 2006
>From Peter Jeremy <peterjeremy at optushome.com.au>, Fri, May 12, 2006 at 07:38:04PM +1000:
> On Thu, 2006-May-11 15:30:35 -0700, Kevin Oberman wrote:
> >exists and its creators (designers) are very sad at its demise. It was
> >an amazing design for its time and could have been a powerful force in
> >hardware with anything that resembled reasonable marketing.
>
> My sentiments as well. When I first used it (in late 1998), I thought
> the architecture looked weird but after using it for a while and
> reading more about the design decisions, I came to the conclusion that
> it was one of the better designed architectures around. There were a
> few warts (requiring software assistance to fully support IEEE FP but
> not supporting precise exceptions was the biggest IMHO) but DEC
> actually considered the likely impact of future changes to technology,
> rather than just band-aiding an existing architecture to meet the
> current technology limitations/requirements.
Realize that the original Alpha did not even do integer divide. It
was one of the cleanest RISC architectures available and to this day
is venerated within the VLSI community for its superb implementation.
The Alphas were the last high-performance processors using mostly
custom logic--the designers worked directly at the level of transistor
geometry.
Sadly this was also it's downfall.
You can tell how good the EV7 was when you realize how long it took for
other architectures to catch-up.
Ye Olde Hudson Design Team, we miss you.
Paul
More information about the freebsd-alpha
mailing list