System Builders Speak

(Ed.note: These are comments from those who built or have been intimately
involved with the construction and/or operation of dozens to hundreds of machines. You can see that these folks have
had much different experiences.

Maybe it might be a good idea to start asking folks who build a lot of systems for
just how they go about installing, a sort of search for best practices. Ed)

We have been in business since 1996 supplying custom built PC’s and combos on a
large scale. I am personally an IT Director for a large company on my
regular job and been building and repairing PC’s since 1983. I feel I am
fairly competent when building a PC.

By far, there have been more problems with VIA based boards and AMD CPU’s
than Intel counterparts. I am NOT an Intel fan as everything I run is AMD.
We have had more DOA chips arrive direct from the distributor by far than
Intel chips. Customer returns are much higher on AMD than Intel ever was.
These are NOT chips with cracked cores, overheated or even overclocked
chips. The RMA rate on Via based boards (both Intel and AMD based) are much
higher than and Intel chipsets (other than the famous 820). This is across
all brands too. Not just brand specific.

I don’t place the blame 100% on VIA or AMD as the problems are probably
because of motherboard manufacturers trying to rush product to market and
not fully testing out these boards, I do believe Ed pointed this out in his
articles, but the fact remains these systems have a higher failure rate and
more stability issues than most anything recently. As an IT Director, I
would NOT even consider any of these systems to be used in our company as a
support technician, I base this on what I see in the RMA bins.

Well, well. I work as a computer technician at a secondary college where we have 75 VIA
based machines. They get an absolute flogging, they have worked flawlessly
since day one. 50 of these are Athlons. I think the name of the article sums it up nicely, you wanker!!!

Sorry to add to your ever-growing pile of opinionated
emails, but as a system builder I must comment:

A while ago, I was asked such a question, and had to
justify to a customer why we did not sell Athlon
systems (as we had in the past). The simple reason
was that they were not “bulletproof”. Basically, it
meant that the running odds on building a nice, stable
Athlon box was hovering at 85%. Some may say, “oh,
that isn’t very bad”.

Yes, it is, if you are the poor smuck that needs to
warranty these systems.

You see, we could build twenty Athlon systems. Some
would be identical down to the software package
installed. Yet, for NO GOOD REASON, 15% of those
systems we know we will see come back for some
hardware reason. Some failed right on the bench, for
no good reason, while its sibling was happily
churning Prime95 and 3D Mark. And when 15% of
computers come back, and they need to have mainboards
swapped out, or OSs reinstalled or some obscure patch
downloaded, that’s time lost, and by extension, money

Naturally, our BX based systems never had a single
peep come out of them. As it stands, there is a 2%-5%
failure rate on those. A far cry from Via’s chipsets.
It’s not that a majority were bad, it’s not even that
a large minority were faulty, but a statistically
significant portion of built systems were bad. An
enthusiast builds one system, and odds are likely his
experiences are good. Build two or three systems, and
you may get a bum board, no big deal. When a hundred
systems are built, its a monumental headache.

So, everytime someone requests an Athlon, I say they
are playing Russian Roulette. Unfortunately, in my
experience, this analogy fits far too well.

In the last 2 years since I convinced my
company that we could build and sell PC’s we’ve done about equal numbers of
Intel and AMD based systems. The total number of these is probably somewhere around 300-500.

When building PC’s for government bids and for
cheapskate clients, we usually need to make the ultimate prices as low
as possible and use Via based boards on most of the Intel boxes we build.
Primarily, Gigabyte boards of the GA-6VXC7 variety is the current board of
choice here with the occasional MSI and Asus thrown in upon request.

Outside of the obvious performance hits, the overall stability of these systems are
not really an issue, but then again these boxes are really not tweaked at
all. No overclocking, no real tweaking of any kind out side of playing in
the bios.

Recently I had a client call me up and ask for a bunch of P3-866
machines with intel boards… He asked for the intel BX chipset,
specifically Abit BE6-II boards. While the stability and ease of install was
again, not much of an issue, the quallity of the boards themselves varied
quite a bit. I had 3 out of 7 boards come to me DOA. This does not inspire
confidence. Eventually I was able to get working replacements and all was
well, but this is an example that just because it doesn’t say “VIA” on it
doesn’t mean it’s a perfect product.

I’m of the opinion that there is more
to blame on the mobo designer/manufacurer’s side than of the chipset
company. ALL mobo companies have been having problems lately. I remember an
article not too long ago afout the quality of Abit’s KT7 and KT7-Raid
boards, I’ve had more than I feel is my share of DOA Asus boards in the last

Moving on to the AMD system side of things. lately the vast majority our
systems have been either Duron or Thunderbird based. In the 200 or so that
I’ve personally built, I HAVE NOT experienced the phenomena of AMD chips
dying mysteriously, nor have I crushed a core yet. In fact, the only
problems that I’ve had are 2 Durons 650’s and 1 T-bird 950 that were either
DOA from my distributor or were flaky when running.

Again for these systems, I prefer the Gigabyte boards, and I used many of the GA-7ZX-1’s. I began
using the original version of MSI’s K7T board but the supply of these dried
up rather quickly. Of course my building is not limited to both of these

I’ve used Asus’s A7V, Abit’s KT7 and KT7-Raid, one of IWill’s boards
(I forget the model). None of these boards has really had any major
stability issues. that I’ve had to go into the field and fight, and believe
me, my clients would yell very loud about it.

In summary, I’d like the stress that I feel the mobo companies are trying to
cut too many corners when designing the high-end boards. Maybe it’s
semi-justified, most people won’t pay the $$ for a board that has had enough
time put into it to make it stable for extreme tweaking. And also let us not
forget about human error. I think Cory had a valid point with the statement:
“People will lie when they did something stupid which broke their system.
I’ve seen it before. It’s human nature to try to cover up embarrassing
mistakes.” Clients do this to me all the time… And I’d guess that there is
more of this than you might think in the tweaking world.

I have quite a bit of experience with via boards, and I think the problem is not the motherboards themselves as far as reliability (well maybe to some degree). It seems that Via needs to enforce some quality control in their chips, because I get some boards that work great, another of the same board with issues, and my generic boards seem to have the same reliability as the good boards.

For example, in my personal system, I usually run a KT7a with a AXHA TBird at various overclocking levels. The first KT7A I got was flawless, made me think everyone that was having problems was just stupid. Then I sold that board with a system and get a new one. It was so-so in reliability. Sold that and I’m on the third and it is a pile. The hard drive seems to “freeze up” for a few seconds every now and then, but that’s only happened in this third system.

Same m-board, 3 completely different experiences.

I have had similar troubles with 2 A7V boards, and a Gigabyte. The most reliable boards i have ever used in Via socket A? IMHO it is the Amptron 807.(kt133, not “a”).

I also get compatability issues a lot more than I have ever before the KT133A. I cant get SBLive5 to work half the time, I have abandoned ATI Radeon whenever i use an AMD platform (except the RadeonVE, seems to be 100% compatible, must be the no T&L). I have had numerous problems with Western Digital drives on a KT7A, and I can only get a DVD player to play DVD video on 25% of the KT133A systems I build because of some incompatability (I use Toshiba DVD drives, PowerDVD3 or WinDVD2k1).

I also can’t seem to get the Detonator 10.80, the Via 4in1 .29s, and SBLiveXgamer to work together(windows protection eror when i install the Detonators).

All in all, I think I have about a 85-90% reliability rating out of the boards. If I build 20 computers in a week, I’ll have to fight the mboard on 2 or three, or something. The systems that work just plain work and are great, but this is how I make a living, and if I’m wasting a few hours per board trying to get one working properly, then I’m wasting a day’s work. Did i mention that I’m lazy??

I love the performance on VIA/AMD systems, but I can see why AMD is having trouble making it into the business market.

Hey, I have had an Athlon systems for almost 2 years, and after installing all the drivers (like any Intel chipset & graphics cards). Have had a rock stable system running 3D Studio MAX and TONS (Photoshop, Illustrator, Bryce, Office 2000, Painter, After Effects Ect.) of other graphics related programs running Windows 2000. Granted, there were problems initially (like more than a year ago) but have long since been worked out.

I don’t care about games at all, and professional software is all I use. I wouldn’t hesitate to refer AMD to ANYONE

P.S. I have built over 50 PCs for customers, ALL AMDs and no problems (besides them not knowing how to work Windows)

I know you have probably been taking a great deal of (doggy doo) over your ‘pebbles’
article. But I have to agree with you 100%. I work with AMD systems every
day in (a reseller’s store), by far, they are the most picky, pesty, fussy, HOT
RUNNING, petulant, grouchy machines I have to build, hands down. That is not
to say that I don’t like them a great deal once I get them running correctly
(the whole 686B/W2KPro thing has woken me in sweating fits at night),
because I do. When they run correctly they run FAST (some just seem to like
to stop running in the middle of whatever I am doing…). I personally have
two TBird systems running at home, my main gaming rig and my LAN party box.
However, I use an Intel solution for my Home Network-based File/MP3/UT/Q3A
server and for my own home-office workstation. They are easy to maintain,
easy to update, RUN COOL (even when OC’ed, which both are), and just don’t
cause me any problems.

This is unfair. We build performance systems and have
switched entirely over to Athlon from Intel on single CPU
systems. I have seen no bad AMD processors since the
K5 days and have had no (NONE) failures of either Athlon
or Duron CPUs. I wish I could say the same for Intel.
We’re NOT happy with Xeon reliability AT ALL.


As far as VIA:, VIA’s Athlon chipsets are better than
VIA’s Intel chipsets. They could be improved, but we’re
been pleased with the KT133a based boards. Be fair,
you don’t mention the i820 fiasco, or the crappy i810, or the
i815 that can only use 512MB memory. The BX is the last
good chipset Intel made and it’s dated. We’ve been starting
to use the AMD 760 and it looks VERY good.

The only reliability problem we’ve had is with ABIT and the
problems are worse with the Intel versions than with the
AMD. Neither are acceptable IMO. The KT7A is a nice
board, but bad QA from ABIT. We have no such problems
with EPoX or the AMD 760 based FIC AD11 so far.

To summarize, we’ve been building systems since the 8088
8/10MHz OC days, and have never seen systems as powerful
and reliable as the latest AMD based machines. If kids can’t
even put a hs/fan on properly, it’s not fair to blame the
hardware (and yes I hope AMD does add a thermal protection
circuit to protect the inept from themselves). On the whole in
our experience AMD quality has topped Intel for the last two
years, going away.

You are absolutely right as somebody who’s responsible for upgrading all the
hardware and rolling out windows 2000 to a 1000+ PC’s at work it not a
difficult decision to make.

I have both Asus P3V4X and KT7A-Raid systems at home and before this BX.

I have time to nurse these systems but at work forget it, it’s a shame but
it logical from a business priority, maximum up time with minimum support.

I hope VIA can grow and resolve the problems.

We’ve got 18 AMD cpus running here and not one has had a problem with
crashing. All our main db servers run on AMD cpus. All our IIS servers also
run on AMD cpus.

Currently our IIS servers have been running for 87 days and
61 days. Our two db servers have only been running for 28 and 14 days. The
only reason the db servers were taken down was to upgrade their software and
add more memory. None of these 4 computers have ever gone down because of a
crash (at least not that I can remember). Even more amazing than that is
the fact that these four computers are running on A7V Mbs with TBird 750s
overclocked to 1100 mhz. Maybe we just got lucky with these computers?
Maybe Durons have more problems?

Then again we have two Duron 750s running on IWILL KA266 mbs at 900mhz with an overclocked bus.
These two systems have been running almost 60 days without a burp, hiccup or crash. We also have 4
TBird 1.2 266s running on IWILL KK266 and IWILL KA266 mbs. All run between
1400 and 1440 mhz on overclocked busses. In the 2 months we’ve had them,
they’ve only been rebooted once. And even that was only because we upgraded
their memory. Are we just lucky? I wonder how well the Pentium III 1.13
would do here? Or maybe the i820 chipset – yeah thats it!

Actually now that I think about it, we did have one Asus A7V mb go bad.
Not the cpu mind you, just the mb. So we packed it up and sent it off to Asus under warranty of course.
However, I do remember two IWILL BD100+ mbs also dying. Only they both died
2-3 months after we got them. So statistically our Intel failures have been
roughly twice our AMD failures.

At our company performance is very important. Unfortunately a lot of large
companies seem to have the same impression of AMD that you have. Even
though they have NO statistics to back it up. Right now I’m looking at a
Compaq 1ghz Pentium III computer with an 18 gig raid scsi array running some
of our analysis software.

The company that owns the machine insisted on
building an Intel machine in order to get the performance and reliability it
insisted upon. So far the machine has been running for 12 hours with another
12 to go. The AMD machine here ran the same software on the same data in 3
hours and 20 minutes on a non raid scsi drive. OK the AMD machine is running
at 1440 mhz but still – 3 hrs to 24hrs is ridiculous. This is the seventh
time we have run this test. After the second test proved no better than the
first, they built yet another machine – same results. We even went so far as
to recompile our code specifically for the Pentium – no difference. And
heres the best kicker, the Pentium cost 3 times our AMD machine. So whats
up? Are we just special? How long do our machines hafta run in order to be
considered reliable? What’s the average? My overclocked desktop machine has
been running for 43 days now. How much longer till its considered reliable?

Email Ed

Be the first to comment

Leave a Reply