New PC to Build

[Dave]

Any comments on the Motherboard (ASUS M2N-MX Socket AM2 NVIDIA GeForce
6100 Micro ATX AMD )? It's only half the price of the previous one I had
selected, Asus M2N SLI Deluxe socket AM2, and yet it has onboard video
where the M2N does not.

I don't like Asus, but I'm one of the few people in the universe who
actually has that opinion, and it is based on professional experience
supporting systems with asus mainboards. In other words, I don't just test
them on a bench for a few hours and write raving reviews... I see how they
perform over time, in real-world conditions, and because of that, I HATE
Asus with a passion. My personal and professional opinion is that Asus has
(generally) good designs, but their quality control sucks ass.

HOWEVER, If you have to use an Asus board, the one you chose is a GOOD
choice. I used to say to avoid micro-ATX boards, for good reason. They
have fewer slots (ram and expansion) usually, limiting their upgrade
potential greatly. But now? Just about everything you need in a computer
is BUILT INTO a mainboard, regardless of format. So having fewer expansion
slots is not such a big deal anymore. And the board you chose has FOUR RAM
slots, which is as much as most full-size ATX boards.

You really aren't sacrificing anything by switching to the lower-priced
M2N-MX over the full-size (and expensive) SLI board you chose earlier. The
larger boards have premium prices because they are in greater demand. And
if you were ever going to use SLI, you wouldn't be considering a micro-ATX
board. SLI boards (like the one you first chose) also have premium prices.
So your savings is (gave up SLI) plus (switched to a smaller board that is
less in demand). Overall, you haven't really given up anything.

You've got three open PCI slots (one PCI-Express X1) in addition to the
PCI-Express X16 graphics slot. That's a lot of expansion/upgrade potential.
With gigabit LAN built in, that means that you might use one PCI slot for a
premium sound card (if you want to), and ... another slot for a TV tuner
sometime down the line? Now you've still got an open slot not being used,
and you haven't even touched your (8 or 10, usually?) USB ports yet!!!

In terms of capabilities, that M2N-MX should be all you need for several
years. It is a good choice, just buy it and don't give it another thought.
-Dave

 

[Dave]

They're not "slower than expected", you ignored all factors.
Video card technology is too complex for a usenet post,
pipes and shaders and dedicated busses and latency and
amount of devoted special-purpose silicon, a video card is a
purpose specific device, designed from ground up to do what
it does.

If anything, video card benchmarks aren't hard to find. The
only time you might find a 6100 gaining ground is if newer
DirectX features are enabled and they bring both cards to
their knees. Run a benchmark that does acceptibly on both
and you can make a better comparison, perhaps 3Dmark 2001.
Geforce 6100 might score around 6000 (this is only a
guesstimation, might vary a few dozen %). In one of the
newer (but last) generation Athlon64 systems (as with
Geforce 6100), even a Geforce 3 would easily exceed 6000
3Dmarks.

Actually, I was looking at one of tom's hardware VGA charts to try to get
SOME idea how a ti4200 might compare to the built-in geforce 6100. On the
3DMark chart I read, the TI4200 scored less than 2000, and some later nvidia
chipsets (after TI4200, but prior to the 6100) scored around ~6000 on the
same chart. Even if we assume that the 6100 is slower because it is
"integrated", it's still going to kick the crap out of a TI4200, performance
wise.

I stand by my earlier prediction that, if someone were to run a direct
comparison, the integrated geforce 6100 video would probably be about 3
times as fast as a TI4200 video card. -Dave

 

[Jack Bruss]

I don't like Asus, but I'm one of the few people in the universe who
actually has that opinion, and it is based on professional experience
supporting systems with asus mainboards. In other words, I don't just
test them on a bench for a few hours and write raving reviews... I see how
they perform over time, in real-world conditions, and because of that, I
HATE Asus with a passion. My personal and professional opinion is that
Asus has (generally) good designs, but their quality control sucks ass.

HOWEVER, If you have to use an Asus board, the one you chose is a GOOD
choice. I used to say to avoid micro-ATX boards, for good reason. They
have fewer slots (ram and expansion) usually, limiting their upgrade
potential greatly. But now? Just about everything you need in a computer
is BUILT INTO a mainboard, regardless of format. So having fewer
expansion slots is not such a big deal anymore. And the board you chose
has FOUR RAM slots, which is as much as most full-size ATX boards.

You really aren't sacrificing anything by switching to the lower-priced
M2N-MX over the full-size (and expensive) SLI board you chose earlier.
The larger boards have premium prices because they are in greater demand.
And if you were ever going to use SLI, you wouldn't be considering a
micro-ATX board. SLI boards (like the one you first chose) also have
premium prices. So your savings is (gave up SLI) plus (switched to a
smaller board that is less in demand). Overall, you haven't really given
up anything.

You've got three open PCI slots (one PCI-Express X1) in addition to the
PCI-Express X16 graphics slot. That's a lot of expansion/upgrade
potential. With gigabit LAN built in, that means that you might use one
PCI slot for a premium sound card (if you want to), and ... another slot
for a TV tuner sometime down the line? Now you've still got an open slot
not being used, and you haven't even touched your (8 or 10, usually?) USB
ports yet!!!

In terms of capabilities, that M2N-MX should be all you need for several
years. It is a good choice, just buy it and don't give it another
thought. -Dave

Thanks, you've answered my questions very well. I will be ordering this
stuff tomorrow. For what it's worth, I've always used Asus boards and never
had a problem, although I've only built about 5 or 6 systems total.

Jack

 

[kony]

Actually, I was looking at one of tom's hardware VGA charts to try to get
SOME idea how a ti4200 might compare to the built-in geforce 6100. On the
3DMark chart I read, the TI4200 scored less than 2000, and some later nvidia
chipsets (after TI4200, but prior to the 6100) scored around ~6000 on the
same chart. Even if we assume that the 6100 is slower because it is
"integrated", it's still going to kick the crap out of a TI4200, performance
wise.

How about less vague nonsense and more specific comparisons?
I already made one, 3DMark2001. As already stated, the only
way for the 6100 to gain ground is if the benchmark is so
new that both of them are so slow they're unusable.

TI4200 is faster, even GF3 is faster.

I stand by my earlier prediction that, if someone were to run a direct
comparison, the integrated geforce 6100 video would probably be about 3
times as fast as a TI4200 video card. -Dave

you can stand by anything you like, and you'd still be
wrong.

 

[kony]

Thanks, you've answered my questions very well. I will be ordering this
stuff tomorrow. For what it's worth, I've always used Asus boards and never
had a problem, although I've only built about 5 or 6 systems total.

Most of the world has used Asus fine, long term,
professionally supported, etc. Dave seems to be randomly
making up nonsense.

 

[dMn]

I get that there are specific technical issues, I was just hoping you
had a little information available to help clarify. After all we are
talking about comparing two video cards, regardless of how complex they
are both built on the same principles.
You made the claim that the card would perform better then the
integrated video, maybe you could have referenced a few of those readily
available benchmark scores.

How about less vague nonsense and more specific comparisons?
I already made one, 3DMark2001. As already stated, the only
way for the 6100 to gain ground is if the benchmark is so
new that both of them are so slow they're unusable.

TI4200 is faster, even GF3 is faster.


you can stand by anything you like, and you'd still be
wrong.

After looking at 42 different articles regarding various Ti4200 cards
and integrated 6100 boards, and looking at benchmarks galore, the whole
thing is a bit clearer to me. Unfortunately, the time difference in the
product releases means that there is little overlap in test versions.

Kony you were absolutely right that the integrated card takes a
performance hit over a separate card. The factors and level of impact
vary based on board implementation, the first 6100 board reviews I found
had pathetic comparitive results. But the ASUS AM2 performed better.

3DMark 2001SE test comparisons:
range 9400-16500 for various Ti4200 cards
range 5700-7000 for early 6100 boards by Foxconn and Biostar

Numbers are more normalized for 3DMark03
Ti4200 1450 - 1661
6100 boards: 1300 - 1450

The 4200 still outperforms the integrated board. Interestingly though
the 6100 boards cleanly outperform the 4200 on 2D tests. And even
better is that the Ti4200 will not run the new benchmarks which is what
the 6100 was optimized for, and performs well within the range of mid
level graphics cards (not integrated).

The quality and performance of the Ti4200 boards varied widely based on
the configuration. As you identify earlier, its a complicated thing
with a lot of factors that can't all be put in a usenet posting. But
the difference is so drastic that your claim that a Ti4200 will run
circles around the integrated 6100 is hard to support completely without
identifying which Ti4200 card you are referring to.

Since the 6100 is forward looking with optimizations for DX9 OpenGL 1.4
and the Ti4200 has topped out, I would go with the onboard controller
and save up for a better card later when you decide you need it and can
afford it.

My original question was how is it possible that the older board would
outperform a newer onboard controller, the answer comes down to
dedicated hardware. The 6100 shares resources with the system. The
other difference is the targeted audience for the cards. The Ti4200 was
developed as a high-end 3D graphics card with hardware optimizations
and the 6100 is a mid level card designed to be a fast 2D card and a
capable performer in 3D. It's optimizations are balanced between
hardware and software.

dMn

 

[dMn]

On Thu, 07 Dec 2006 22:35:13 GMT, "Jack Bruss"
<SNIP>

Most of the world has used Asus fine, long term,
professionally supported, etc. Dave seems to be randomly
making up nonsense.

For the record I agree with Jack regarding the ASUS boards. I've had
several with capacitor problems that were a real pain to troubleshoot.
I've burnt one out (a mishap that can't really be blamed on the board).
And had one pentium III board that behaved as if it were badly
grounded or shorted out intermittently. But my Bro in Law is a die-hard
ASUS customer, so I keep going back. Once the systems were built and
burned in, they were usually pretty solid.

dMn

 

[kony]

I get that there are specific technical issues, I was just hoping you
had a little information available to help clarify. After all we are
talking about comparing two video cards, regardless of how complex they
are both built on the same principles.

Why when a search engine can find the details?

You made the claim that the card would perform better then the
integrated video, maybe you could have referenced a few of those readily
available benchmark scores.

Frankly I'm amazed you didn't know it already. It's not a
"claim", it's common knowledge to most who know video cards.
To reargue and reprove that which was already solidly
establidshed fact is a waste of time.

After looking at 42 different articles regarding various Ti4200 cards
and integrated 6100 boards, and looking at benchmarks galore, the whole
thing is a bit clearer to me. Unfortunately, the time difference in the
product releases means that there is little overlap in test versions.

Kony you were absolutely right that the integrated card takes a
performance hit over a separate card. The factors and level of impact
vary based on board implementation, the first 6100 board reviews I found
had pathetic comparitive results. But the ASUS AM2 performed better.

3DMark 2001SE test comparisons:
range 9400-16500 for various Ti4200 cards
range 5700-7000 for early 6100 boards by Foxconn and Biostar

Numbers are more normalized for 3DMark03
Ti4200 1450 - 1661
6100 boards: 1300 - 1450

The 4200 still outperforms the integrated board. Interestingly though
the 6100 boards cleanly outperform the 4200 on 2D tests. And even
better is that the Ti4200 will not run the new benchmarks which is what
the 6100 was optimized for, and performs well within the range of mid
level graphics cards (not integrated).

Video is no bottleneck at 2D uses, a benchmark is only as
useful as the real-world apps it represents.

6100 doesn't "perform well" at anything relative to mid
level graphics cards. However, performance is only a factor
when video was the bottleneck and for many computer
applications it isn't.

The quality and performance of the Ti4200 boards varied widely based on
the configuration. As you identify earlier, its a complicated thing
with a lot of factors that can't all be put in a usenet posting. But
the difference is so drastic that your claim that a Ti4200 will run
circles around the integrated 6100 is hard to support completely without
identifying which Ti4200 card you are referring to.

TI4200 is significantly faster... not 10X or anything like
that, but if either can run an app it comes closer.

Since the 6100 is forward looking with optimizations for DX9 OpenGL 1.4
and the Ti4200 has topped out, I would go with the onboard controller
and save up for a better card later when you decide you need it and can
afford it.

Nonsense. Has nothing to do with forward-looking, it's only
a matter of new platforms not having an AGP slot. 6100
video is inferior. "IF" the performance difference matters
then it is the better option. If the performance difference
doesn't then there was nothing to discuss.

My original question was how is it possible that the older board would
outperform a newer onboard controller, the answer comes down to
dedicated hardware. The 6100 shares resources with the system. The
other difference is the targeted audience for the cards. The Ti4200 was
developed as a high-end 3D graphics card with hardware optimizations
and the 6100 is a mid level card designed to be a fast 2D card and a
capable performer in 3D. It's optimizations are balanced between
hardware and software.

The 6100 is not a mid-level anything, it's low end
integrated video. It's a little faster than some other
integrated video, but it's really splitting hairs
considering the disparity in performance between modern
integrated video and modern low-midrange cards.

It is not designed to be fast at 2D, if you want to compare
it to anything else, it's just fast *enough* to use for many
purposes, which is an acceptible tradeoff for lower cost,
heat, noise, depending on the user's needs.

 

[kony]

For the record I agree with Jack regarding the ASUS boards. I've had
several with capacitor problems that were a real pain to troubleshoot.
I've burnt one out (a mishap that can't really be blamed on the board).
And had one pentium III board that behaved as if it were badly
grounded or shorted out intermittently. But my Bro in Law is a die-hard
ASUS customer, so I keep going back. Once the systems were built and
burned in, they were usually pretty solid.

They had a few boards in P3 with cap problems, but so did
practically everyone else on a few products in past years.
Unlike many of the 2nd tier makers they do (and have for
several years) used quality brands of caps in the more
critical roles. If Nichicon et al. has a bad batch, it
effects everyone using them, not just Asus, it is out of
their control.

What is less excusible is when a manufacturer knowingly uses
lower grade caps and those perpetually have lower life, and
more susceptible to overheating.

 

[dMn]

<SNIP>

Why when a search engine can find the details?

But that's so much more impersonal

Frankly I'm amazed you didn't know it already. It's not a
"claim", it's common knowledge to most who know video cards.
To reargue and reprove that which was already solidly
establidshed fact is a waste of time.

Be amazed, I'm still a little baffled at why a four year newer
technology with all the potential hardware advantages didn't show
better. I'm not trying to argue or prove anything, I'm trying to
understand.

<SNIP>

Video is no bottleneck at 2D uses, a benchmark is only as
useful as the real-world apps it represents.

6100 doesn't "perform well" at anything relative to mid
level graphics cards. However, performance is only a factor
when video was the bottleneck and for many computer
applications it isn't.

3 separate published reviewers would disagree with you.

TI4200 is significantly faster... not 10X or anything like
that, but if either can run an app it comes closer.

You missed my point here, I was referring to the fact that different
venders delivering cards based on the Ti4200 chip seemed to produce
radically different performing cards.

Nonsense. Has nothing to do with forward-looking, it's only
a matter of new platforms not having an AGP slot. 6100
video is inferior. "IF" the performance difference matters
then it is the better option. If the performance difference
doesn't then there was nothing to discuss.

Just call it as I see it. If the app requires DX9 or OpenGL 1.4 which I
presume the next generation of apps will begin to require, then being
able to run it slower beats not at all.

The 6100 is not a mid-level anything, it's low end
integrated video. It's a little faster than some other
integrated video, but it's really splitting hairs
considering the disparity in performance between modern
integrated video and modern low-midrange cards.

It is not designed to be fast at 2D, if you want to compare
it to anything else, it's just fast *enough* to use for many
purposes, which is an acceptible tradeoff for lower cost,
heat, noise, depending on the user's needs.

Again, here I'm just relating what I learned from other published
experts on the topic. If they're wrong, then so be it. But the fact
that three reputable sources made similar claims leads me to believe
they had some validity. I have no opinion one way or the other. You on
the other hand, have clearly made up your mind. IMHO whatever works go
with it.

But thanks for the discussion, it gave me an opportunity to review some
things I hadn't looked into for a while.

dMn

 

[dMn]

<SNIP>

What is less excusible is when a manufacturer knowingly uses
lower grade caps and those perpetually have lower life, and
more susceptible to overheating.

Agree.

dMn

 

[kony]

But that's so much more impersonal

Maybe but it's also more comprehensive than any one person
would write about.

Be amazed, I'm still a little baffled at why a four year newer
technology with all the potential hardware advantages didn't show
better. I'm not trying to argue or prove anything, I'm trying to
understand.

Because it's not a matter of "newer" always being better.
Same goes for supposed features and DirectX support, for any
of it to be useful the raw video processing power and
dedicated memory throughput has to be good already.

With integrated video, it isn't. Integrated video (these
days, opposed to years ago (or server boards) when it was
more common to have a discrete GPU, even discrete memory
chips soldered onto the motherboard) uses minimal silicon,
just what they can squeeze onto a small northbridge core.
The whole point is that it's cheaper, to buy, because it's
cheaper to make.

3 separate published reviewers would disagree with you.

This is vague unuseful, nonsensical. Post specific
scenarios, direct links, etc.

If you really want to pretend it's better, go right ahead
and buy or use it. Not my problem. It really IS better in
some circumstances, like when you don't need the extra
performance, just a video head on a server or internet
kiosk, whatever, where there's no benefit from higher cost,
power, heat, etc of a separate card.

You missed my point here, I was referring to the fact that different
venders delivering cards based on the Ti4200 chip seemed to produce
radically different performing cards.

In general, no. Most used what is essentially an nVidia
reference design and they all performed very similarly, as
there was a specific GPU clock, bus width, memory speed
spec. What will vary results more is what the rest of the
system is like, notice in a past post I'd mentioned the
detail of running one on a semi-modern equivalent system,
not taking benchmarks of one running on a P3 or early
Athlon.

Perhaps you were looking at GF4 MX scores, that is a
different card than a TI4200. Some TI4200 did have 128MB of
memory, which was slightly slower (but mostly interfered
with overclocking efforts) at stock on old games, but more
than regained an advantage on newer games or with higher
game detail settings, set.

Just call it as I see it. If the app requires DX9 or OpenGL 1.4 which I
presume the next generation of apps will begin to require, then being
able to run it slower beats not at all.

If an app is demanding enough that it "requires" these
things, you will not be liking the low performance you'd get
with integrated video. You are thinking even of the future,
while TODAY 6100 can't do well even at 1024x768 resolution
with minimal eyecandy in any games, even older games like
Doom3 it can't play at that resolution acceptibly. Ti4200
isn't great at them either, but better...

There's just no point in pretending 6100 integrated video
has an advantage, it's too slow. No theory about paper
specs overcomes it's inability to perform at a usable level.

Again, here I'm just relating what I learned from other published
experts on the topic.

Experts? Realize that many of them were still in diapers
when I was overclocking video cards. Being a web author is
not a sign of expertise. Many get a great opportunity to
play with neat newer technology but playing, understanding,
and then writing a review that is correctly interpreted by a
reader are not always sync'd.

If they're wrong, then so be it. But the fact
that three reputable sources made similar claims leads me to believe
they had some validity. I have no opinion one way or the other. You on
the other hand, have clearly made up your mind. IMHO whatever works go
with it.

IMO, the thing to learn from this is that sometimes reviews
don't tell the whole story, and some reviews/reviewers are
better than others. If you just want some quick way to
compare different video cards then you ought to go to one of
the web forums like the video forum at anandtech.com, there
are plenty of people there that will answer your question
then begin arguing about which $400 ATI or nVidia card you
should buy instead.

 

[OSbandito]

Kony took the bull by the horns with:

Frankly I think you're lucky to have gotten the 250W PSU to
run those parts, it might have had a lot to do with the
A7N8X running the CPU power subcircuit off the 5V PSU rail.

I bought a few of those cases with the 250, 300, 340W, (then
"maybe" 360, don't recall if exact same case on this 360)
and anything under 340 wasn't a very good PSU. IIRC I even
had a system in recently that had one of the 340 with a
blown transistor so my recollection of it is a little better
than most PSU that old.

It (and especially your lower 250W unit) doesn't have the
12V current capacity required for a modern system. The case
you can keep using (though it's ventilation is borderline by
today's standards, personally I'd take out the front bottom
plastic fan mount/card-guide, then cut out the stamped-in
metal fan grills on that bottom front and mid-rear fan
hole(s). It also doesn't (IIRC) force any of the front fan
air past the hard drive rack which is unfortunate, I seem to
recall taking a drive cage cannibalized from another case
and mounting it to the floor of one of those cases so the
drives were sitting directly behind that lower front fan,
better cooled by it.

Otherwise, the case is standard ATX and plenty deep enough
for even a full width motherboard so anything but the
oddball Intel BTX boards will work fine in it... You'll
just need to replace the PSU, preferribly something rated
for a minimum of 400W in a good name-brand rated for minimum
of 18A on 12V rail, or more with a decent video card
installed and/or more than a couple drives. In other words,
ideally the PSU would match the expansion capabilities of
the case including # of drive bays that could have drives,
and supply >= 24A of 12V power.

Righto. Listen to Mr. Kony. Power supplies are meant to meet 3
functions: come as close as possible to battery-like performance with
clean, steady DC voltages (have to look into sq.wave, steppers, etc but
I'm speaking about general purpose), the capacity to provide needed amps
without straining the PS components or overheating them and, finally, to
provide protection in case of fault or out of spec operation. A good,
oversized, brand-name pwr.supply with big quiet fan is a thing of
beauty..well worth paying extra for. It will have beautiful, fat,
heat-tolerant capacitors. It will protect your precious electronics,
save you headaches and be ready for re-use on your next build. Nothing
in the computer can run in a healthy way without this solid foundation.

There, I feel better now.