[vox-tech] Hardware Question
ME
vox-tech@lists.lugod.org
Thu, 28 Mar 2002 09:30:42 -0800 (PST)
There is truth to items mentioned on both sides of this issue.
You can build a stable Intel CPU based system just as you can build a
stable Athlon based system.
What happens in the marketplace, is any kind of cost saving that *can* be
made by using less expensive parts an still be able to quote similar
number (MHz speed, Memory, HD Size, MB Speed, etc) will be made for
less expensive models.
Imagine it like this:
If a company can save even $50 on an Athlon CPU over an Intel one (for the
same rated clock speed) or even over-clock a cheaper Athlon not rated for
higher speeds (by say 5010%) and save $100 and
Save $15 on a cheaper DVD/CDROM drive and
save $2 on a cheaper mouse and
save $8 on a cheaper keyboard and
save $120 on a cheaper 17" monitor (With 15.5" viewable, .28dp) and ...
They can offer what would appear to the casual observer to be an
"equal" system for less money.
People looking at CPU Speed, MB Speed, Monitor Size, memory, etc will see
"an equal system" without knowing what quality parts were used.
Histrically, companies like AMD (and Cyrix heh) produced chips that
claimed rated speeds that matched Intel's chips, but cost
less. Companies/individuals wishing to squeeze out more profits from each
sale could use inferior parts at a discount, and memory not within spec of
the MB, clock the MB outside the "company recommended ranges", and
overclock the CPU by 5%-10% - all to make their advertised system appear
to be on level ground with other systems costing $100-$600 more.
A tendancy for these shadier dealers to look for the less costly parts has
historically pushed them into non-intel CPU. As a result, sales of
"cheap" systems that use non-Intel chips give the non-Intel chips a bad
name (when the assembler is the more likely the person to blame along with
the consumer who was trying to get something for less.)
This frequently leads some techies to say, "Well, just go Intel" since on
the average, systems made with an Intel CPU have historically, mostly been
systems where the vendor is willing to spend more on higher quality parts
and pass the cost onto the consumer with a more expensive system, but with
higher quality parts.
This attitude has cause many vendors to catch onto this tred, and buy
Intel Celleron chips and still keep all the rest of the components
cheap. Seen this with some people who bought that $600 boxed, non-name
"intel based CPU" system from Costco between the Salami and the VHS
tapes.
As consumers become more educated in "what to avoid" as well as "what to
buy", the shadier vendors will catch onto these trends and try to make the
filtering system used by consumers become a boon to their sales instead of
a brake.
At one time "dot-pitch" for monitors was all-the-rage, and vendors of the
cheaper equipment found ways to claim lower dotpitch with horizontal
measurements instead of diagnal or vertical (with the more impressive
number in 72-point font and the less interresting ones at 6-point.)
With the addition of monitor sizes, vendors started selling monitors based
on the CRT size - not the viewable area, which is often in "()" or in
small print.
Tape drives that are sold based on an assumption of 2:1, 3:1 and sometimes
4:1 hardware compression!
One of the latest (within the past 2 years or so) items for consumers with
monitors has been "refresh rate". Mags say, "You want a high refresh rate
for your monitor as that signifies quality."
Of course vendors look to try to improve their monitors, but ony as much
as is necessary. The vendors have seen this and will claim "140Hz
refresh rate" and then in small print (at 640x480 or 320x440) while the
Mags that performed their side-by-side tests did all of their tests at
1024x768. (This is not the case now so much, but earlier comparisons used
higher resolutions so as to better filter out the monitors that were not
as good and have the differences be more obvious.) However, the dumb-ass
consumer just remembers "high refresh-rate, high refresh rate, high..."
Um. Ok. Where was I?
Oh yeah...
Sure, there are bad Intel systems and bad non-intel systems, but systems
that use the non-celleron intel chips, in larger cases, or from
main-stream vendors (Dell seems to use rather standard and stable hardware
but are frequently more expensive, while Gateway's systems seem to be made
from cheaper parts, but also frequently cost less) seem to work
better than the systems made at new computer shops with no history, and
people you do not know.
When you know the people who make the systems, and you know their
reputation, that can make a HUGE difference in the quality of the system
you purchase.
My last comment in this message on Intel vs. Athlon:
One feature that was true, but I am not so sure it is anymore, is an Intel
Pentium III based system can still function for the most part when the
heat sync+fan on the CPU fail. (Built-in themostat for throttling speed
when temperatures get to high in the newer Pentiums) while failure in the
fan+heat sync in the AMD chips leads to melt-down. This last item was true
when I bought the last batch of servers, and is another reason I had for
going with Intel.
There was a nifty video showing these effects.... let me find it.
-ME
-----BEGIN GEEK CODE BLOCK-----
Version: 3.12
GCS/CM$/IT$/LS$/S/O$ !d--(++) !s !a+++(-----) C++$(++++) U++++$(+$) P+$>+++
L+++$(++) E W+++$(+) N+ o K w+$>++>+++ O-@ M+$ V-$>- !PS !PE Y+ !PGP
t@-(++) 5+@ X@ R- tv- b++ DI+++ D+ G--@ e+>++>++++ h(++)>+ r*>? z?
------END GEEK CODE BLOCK------
decode: http://www.ebb.org/ungeek/ about: http://www.geekcode.com/geek.html
Systems Department Operating Systems Analyst for the SSU Library
On Thu, 28 Mar 2002, Rod Roark wrote:
> LUGOD's web and mailing list server is an Athlon system built
> by little ol' me, and has been running reliably for the past
> 4 1/2 months now.
>
> -- Rod
> http://www.sunsetsystems.com/
>
> On Wednesday 27 March 2002 09:25 pm, Bill Broadley wrote:
> > > In my limited experience, Athlon systems build by small outfits or
> > > individuals are unstable. I would not use such machines for an important
> > > server.
> >
> > In my experience athlons are fine.
> >
> > Dual athlon (appro/small outfit):
> > sine:~> uptime
> > 9:16pm up 44 days, 11:42, 20 users, load average: 2.00, 2.31, 2.60
> > cosine:~> uptime
> > 9:17pm up 44 days, 11:37, 8 users, load average: 0.00, 0.00, 0.00
> > tangent:~> uptime
> > 9:15pm up 29 days, 10:10, 2 users, load average: 0.17, 0.04, 0.01
> >
> > Dual athlon (lnxi/small outfit):
> > beowulf:~> uptime
> > 9:18pm up 26 days, 4:53, 6 users, load average: 2.28, 2.19, 2.11
> >
> > Homebuilt desktop (single cpu):
> > [bill@bilbo bill]$ uptime
> > 9:18pm up 19 days, 22:04, 19 users, load average: 1.06, 1.06, 1.03
> >
> > Some random cluster nodes (lnxi/dual):
> > [n1]
> > 9:19pm up 21 days, 7:19, 0 users, load average: 2.00, 2.00, 2.00
> > [n2]
> > 9:19pm up 21 days, 7:20, 0 users, load average: 2.00, 2.00, 2.00
> > [n3]
> > 9:19pm up 21 days, 7:19, 0 users, load average: 1.00, 1.00, 1.00
> > [n4]
> > 9:19pm up 21 days, 7:19, 0 users, load average: 1.00, 1.00, 1.00
> > [n5]
> > 9:19pm up 21 days, 11:53, 0 users, load average: 1.99, 1.98, 1.99
> > [n6]
> > 9:19pm up 26 days, 2:12, 0 users, load average: 1.99, 1.98, 1.99
> > [n8]
> > 9:19pm up 21 days, 7:19, 0 users, load average: 2.00, 2.00, 2.00
> > [n9]
> > 9:19pm up 21 days, 7:19, 0 users, load average: 1.99, 1.98, 1.99
> > [n11]
> > 9:19pm up 21 days, 7:18, 0 users, load average: 2.00, 2.00, 2.00
> > [n13]
> > 9:19pm up 21 days, 7:19, 0 users, load average: 2.00, 2.00, 2.00
> > [n15]
> > 9:19pm up 26 days, 2:16, 0 users, load average: 2.00, 2.00, 2.00
> > [n16]
> > 9:19pm up 26 days, 2:15, 0 users, load average: 2.00, 2.00, 2.00
> > [n17]
> > 9:19pm up 26 days, 2:15, 0 users, load average: 2.00, 2.00, 2.00
> > [n18]
> > 9:19pm up 26 days, 2:14, 0 users, load average: 2.00, 2.00, 2.00
> > [n19]
> > 9:19pm up 26 days, 2:15, 0 users, load average: 2.00, 2.00, 2.00
> >
> > Of course there are bad motherboards out there, and of course people
> > often ignore a vendors recommendation on dimms and instead of getting
> > certified memory they get generic memory from a random vendor, or a
> > non-athlon certified power supply etc.
> >
> > Buy quality stuff the chances are better that it will work (amd or intel),
> > buy crap and your likely to have more problems.
> >
> > The last big vendor athlon I bought was a micron, worked fine, had a
> > ibm deskstar death, but otherwise worked. Unfortunately it was sold into
> > microsoft slavery so I can't run an uptime on it to see how long it's
> > been up 8-(.
> _______________________________________________
> vox-tech mailing list
> vox-tech@lists.lugod.org
> http://lists.lugod.org/mailman/listinfo/vox-tech
>