Why use a Unix workstation?

I am completely new to unix risc-based workstation. However, by
comparing the benchmarks between a 3.2Hz Xeon NT workstion and a HP
9000 (c3700, J6000 for example), I found that NT workstations have much
higher benchmarks(SPEC 2000, www.spec.org/osg/). So I am wondering why
people still using unix workstations. Could anyone give me some clues
please. thanks

yzhang

yzhang <> wrote:
> I am completely new to unix risc-based workstation. However, by
> comparing the benchmarks between a 3.2Hz Xeon NT workstion and a HP
> 9000 (c3700, J6000 for example), I found that NT workstations have much
> higher benchmarks(SPEC 2000, www.spec.org/osg/). So I am wondering why
> people still using unix workstations. Could anyone give me some clues
> please. thanks

Not all benchmarks fit in cache and so not all benchmarks benefit from
the much higher clock frequency. A j6000 runs the CPU at a whopping
550 MHz. IIRC the c3700 was a bit higher.

The c3700 and j6000 are/were 64-bit systems and so could go beyond 4GB
RAM limits that until "recently" one could not (at least not easily)
do with a Xeon system. They may have more aggregate I/O capacity.

Keep in mind that the j6000 is now a very old system and the c3700 is
an old system. I'm not sure when 3.2 GHz Xeon's hit the streets. So,
while you may be new to Unix RISC-based workstations, the systems you
are using are not at all new.

rick jones
--
Process shall set you free from the need for rational thought.
these opinions are mine, all mine; HP might not want them anyway...
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

Rick Jones wrote:
> yzhang <> wrote:
>
>>I am completely new to unix risc-based workstation. However, by
>>comparing the benchmarks between a 3.2Hz Xeon NT workstion and a HP
>>9000 (c3700, J6000 for example), I found that NT workstations have much
>>higher benchmarks(SPEC 2000, www.spec.org/osg/). So I am wondering why
>>people still using unix workstations. Could anyone give me some clues
>>please. thanks
>
>
> Not all benchmarks fit in cache and so not all benchmarks benefit from
> the much higher clock frequency. A j6000 runs the CPU at a whopping
> 550 MHz. IIRC the c3700 was a bit higher.
>
> The c3700 and j6000 are/were 64-bit systems and so could go beyond 4GB
> RAM limits that until "recently" one could not (at least not easily)
> do with a Xeon system. They may have more aggregate I/O capacity.
>
> Keep in mind that the j6000 is now a very old system and the c3700 is
> an old system. I'm not sure when 3.2 GHz Xeon's hit the streets. So,
> while you may be new to Unix RISC-based workstations, the systems you
> are using are not at all new.
>
> rick jones

Uh, the reason he's looking at old PA workstations is that there aren't any
new ones. HP pulled the plug on PA development years ago, touting Itanium
as the wave of the future. Home PC users were going to have Itaniums in
their machines, serious workstation users were going to Itanium, and of
course servers of all types were going to be Itanium.

However, as the Itanium program imploded over time, the target for the chips
shrank considerably. Intel is now saying that liquid-cooled mainframes is
where they'll find a home. HP officially dropped their Itanium hpux workstations
last September, effectively abandoning the workstation market. That better
explains why competitive unix workstations from HP are hard to find.

Itanium has always been the quintesential solution in search of a problem to
solve, and still struggles in that regard. I'd give it one more CEO change
at Intel before they flush this multi-billion-dollar boo-boo down the drain
totally.

yzhang wrote:

> I am completely new to unix risc-based workstation. However, by
> comparing the benchmarks between a 3.2Hz Xeon NT workstion and a HP
> 9000 (c3700, J6000 for example), I found that NT workstations have
> much higher benchmarks(SPEC 2000, www.spec.org/osg/). So I am
> wondering why people still using unix workstations. Could anyone give
> me some clues please. thanks

1.) The c3700 (PA-8700 750MHz) is around 3 years old, the J6000 is even
older (>5 years). Your 3.2GHz XEON is how old? The c3700 and the J6000
use 120MHz SDRAM memory which of course is much slower than the 800MHz
Dual Channel DDR-SDRAM the XEON uses. All in all You are comparing old
hardware with current hardware. And You really wonder why the newer
hardware is faster? You better compare the XEON system with something
more up2date like HP's c8000...

2.) SPEC is a synthetic benchmark which depending on the kind of
programs You want to run says much or simply nothing about the
real-world performance of these systems.

3.) Performance isn't everything. Some companies buy PA-RISC
workstations to have desktop computers that are binary compatible with
their big supercomputers, or simply want the reliability of HP-UX on
their desktop...

Benjamin