which to buy Chaintek nVidia or Abit VIA motherboard?

[GeneL]

Hello:

I have a choice of two motherboards for an AMD XP 2900, which is I
believe a Barton. The choices are an Abit KV7-V with the VIA
KT600/VT8237 chipset and a Chaintek 7NJL6 with the nForce2 Ultra 400
chipset. Both have support for a 400 MHz FSB.

Which would you buy? Two more questions, the AMD XP 2900, does it
have a built in heat sensor and will it shut down on its own if it
overheats like a PIII?

Thanks for your suggestions.

 

[Conor]

Hello:

I have a choice of two motherboards for an AMD XP 2900, which is I
believe a Barton. The choices are an Abit KV7-V with the VIA
KT600/VT8237 chipset and a Chaintek 7NJL6 with the nForce2 Ultra 400
chipset. Both have support for a 400 MHz FSB.

Which would you buy?

Chaintech. VIA still have issues, not least with the latest gen nVIDIA
cards and some games.

Two more questions, the AMD XP 2900, does it
have a built in heat sensor
No.

and will it shut down on its own if it
overheats like a PIII?

Later Socket A motherboards throttle downt he CPU and some have CPU fan
fail and overheat shutdown options in BIOS.

 

[kony]

Hello:

I have a choice of two motherboards for an AMD XP 2900, which is I
believe a Barton. The choices are an Abit KV7-V with the VIA
KT600/VT8237 chipset and a Chaintek 7NJL6 with the nForce2 Ultra 400
chipset. Both have support for a 400 MHz FSB.

The Abit is probably a better built board and will be better
supported (for example, bios bug patches). nForce2 is
faster than KT600 though. I'd lean towards the Chaintech if
forced to chose between the two AND you didn't plan on
keeping it as a primary system for a decade.

Which would you buy?

That's an entirely different topic.
Neither.
I'll pick one for free but no money towards either unless
they were disproportionately price-reduced relative to their
normal prices. In that case (and considering that buying
socket A today means you aren't looking for ultimate
performance), I might consider the feature sets on each,
whether one or the other better accomodates your needs.

However, if you're buying, there has to be somebody out
there willing to sell you a different board than just these
two. The world is not made up of only these two, you should
have no trouble finding a better board. If these are simply
the two cheapest boards some certain vendor has, then I
argue that just because it's no longer cutting-edge tech,
that does not make a board worth less during that era,
suddenly a better idea today.

Most popular enthusiast board for late-socket A was probably
one of the latter Abit nForce2 boards. AN7 or one of the
NF7 series. If you can find an MSI, Gigabyte, Asus or
Abit nForce2 board, that's what I'd get. Runner up brands
might be Shuttle, Biostar, Chaintech (but really, you should
not have to venture down the board quality chain too far,
there are socket A boards still at many vendors).

Two more questions, the AMD XP 2900, does it
have a built in heat sensor

Yes, every Athlon from XP onward does (anything faster than
~ 1.4GHz Thunderbird). "Some" older boards doesn' use this
sensor though, instead relying on an in-socke sensor. A
more modern KT600 or nForce2 board ought to use the sensor.

and will it shut down on its own if it
overheats like a PIII?

Not "on it's own", in that the CPU dictates when it shuts
off. Instead, AMD mandated this back in (might've been '02
or '03) that the board itself has the feature, that ALL
would (unless you came across some rogue board that had
extreme cost-cutting measures and disregarded the AMD specs.
In other words, don't buy brands you don't recognize at all
and PCChips/ECS are a gamble. Then again, any cheap
off-brand is more of a gamble. Best advice regarding
thermal shutdown is plan not to ever have to need it even
though it's there. IE- don't skimp on fans, use quality
(fan-manufacturer-brands, not PC gear brands) dual
ball-bearing thick fans or Papst / Panaflo if a
sleeve-bearing fan.

IMO, more PC parts fail due to fans dying than due to a CPU
that overheated till it failed, and the vast majority of the
failed fans were junk sleeve-bearing types.

 

[John Doe]

Yes, every Athlon from XP onward does (anything faster than
~ 1.4GHz Thunderbird). "Some" older boards doesn' use this
sensor though, instead relying on an in-socke sensor. A
more modern KT600 or nForce2 board ought to use the sensor.

That is interesting. There is a temperature sensor (looks like a
thermistor) in the ZIF socket on my K7N2 Delta2-LSR

http://www.msicomputer.com/product/p_spec.asp?model=K7N2_Delta2-LSR&class=mb

I see no mention of an ability to measure temperature through the
Athlon XP CPU. Is that handled by Windows software?

The socket temperature probe is used for BIOS temperature reading?
Does Any BIOS reading use the on chip temperature sensor?

In other words, don't buy brands you don't recognize at all
and PCChips/ECS are a gamble.

Just from window (online) shopping over the years, I have always
been skeptical about PC-Chips mainboards. You get what you pay
for.

Thanks.

 

[kony]

That is interesting. There is a temperature sensor (looks like a
thermistor) in the ZIF socket on my K7N2 Delta2-LSR

http://www.msicomputer.com/product/p_spec.asp?model=K7N2_Delta2-LSR&class=mb

Many boards did have them because they supported the
pre-"XP" Athlons as well.

I see no mention of an ability to measure temperature through the
Athlon XP CPU.

You wouldn't see most details of the electrical
interworkings of a motherboard in it's product description
or manual. It's possible the CPU thermal sensor is
disregarded, that the board is only using the socket-well
sensor, but not necessarily and not as good a design
decision. Since I dont' have that board here to check, I
can't give a definite answer for it.

Is that handled by Windows software?

No temp reading is handled by software, never.
It is an entirely electrical hard-wired feature. However,
the hardware sensor stores the read-temp values and any
software or bios can read this data, so software is a
supported but unnecessary feature- AMD's spec for shutdown
relies on no software, only (at most) bios firmware if the
ability is there to vary what threshold temp shuts it down,
as is often a feature.

The socket temperature probe is used for BIOS temperature reading?

I dont' have your board. On a well-designed board, no it
would only be used when CPU has no on-die sensor.

Does Any BIOS reading use the on chip temperature sensor?

How does this question differ from the previous?
If MSI took the quicker or cheaper route, they could have
chosen to only use the socket-well sensor. That is a slower
sensor, but unless your heatsink falls off, (and assuming it
is properly calibrated), the differences may be minor or
irrelevant. The truth is that knowing the actual temp of
the CPU core to a high level of precision isnt' necessary
for the shut-down feature to work properly. Instead, if the
CPU ventures into an instable temp range there is still some
margin before it's hot enough to be damaged.

Just from window (online) shopping over the years, I have always
been skeptical about PC-Chips mainboards. You get what you pay
for.

The funny part is that PCChips even puts silkscreening on
boards for some parts and then uses jumpers instead of the
part (like keyboard or USB fuses), when they could've
conceiled their cheap design better by not marking where
these parts would go, if they're not even going to use them.

 

[John Doe]

http://www.msicomputer.com/product/p_spec.asp?model=K7N2_Delta2-LSR&class=mb

Many boards did have them because they supported the pre-"XP"
Athlons as well.

You wouldn't see most details of the electrical interworkings of
a motherboard in it's product description or manual. It's
possible the CPU thermal sensor is disregarded, that the board
is only using the socket-well sensor, but not necessarily and
not as good a design decision. Since I dont' have that board
here to check, I can't give a definite answer for it.

No temp reading is handled by software, never. It is an entirely
electrical hard-wired feature.

Bad semantics, in my opinion.

Of course a temperature sensor is a hardware device. But the
reading and displaying is done by software.

I probably should have used the word "read" instead of "measure".
The measuring is a hardware function.

However, the hardware sensor stores the read-temp values and any
software or bios can read this data, so software is a supported
but unnecessary feature-

Unnecessary for thermal shutdown. Necessary for a user interface,
unless you have a temperature readout on the front of your case or
wherever.

Sounds like you are very interested in this stuff.

I dont' have your board. On a well-designed board, no it would
only be used when CPU has no on-die sensor.

How does this question differ from the previous?

It elicited a longer reply.

But seriously. Sometimes rewording a question helps the reader
understand.

If MSI took the quicker or cheaper route, they could have chosen
to only use the socket-well sensor. That is a slower sensor,
but unless your heatsink falls off, (and assuming it is properly
calibrated), the differences may be minor or irrelevant. The
truth is that knowing the actual temp of the CPU core to a high
level of precision isnt' necessary for the shut-down feature to
work properly. Instead, if the CPU ventures into an instable
temp range there is still some margin before it's hot enough to
be damaged.

I was considering the possibility that a BIOS update which seems
to cause a change in temperature reading might have something to
do with switching from the under CPU thermistor to the on CPU
temperature sensor. Just guessing, mostly curious. It is no big
deal.

Thanks.

 

[David Maynard]

kony wrote:

The funny part is that PCChips even puts silkscreening on
boards for some parts and then uses jumpers instead of the
part (like keyboard or USB fuses), when they could've
conceiled their cheap design better by not marking where
these parts would go, if they're not even going to use them.

That is actually standard engineering practice.

 

[kony]

kony wrote:


That is actually standard engineering practice.

I agree, when parts that are optional "might" be used. I too
see that all the time, but not when it's such a basic
component and not when none of them have it. More often
it's doen when there are optional features not implemented,
or to facilitate component/supply/sourcing changes- neither
of which apply.

 

[GeneL]

  >>>Two more questions, the AMD XP 2900, does it
  >>>have a built in heat sensor
 >>
 >> Yes, every Athlon from XP onward does (anything
faster than
 >> ~ 1.4GHz Thunderbird). "Some" older boards doesn'
use this
 >> sensor though, instead relying on an in-socke sensor.
A
 >> more modern KT600 or nForce2 board ought to use the
sensor.

Many boards did have them because they supported the
pre-"XP" Athlons as well.



You wouldn't see most details of the electrical
interworkings of a motherboard in it's product description
or manual. It's possible the CPU thermal sensor is
disregarded, that the board is only using the socket-well
sensor, but not necessarily and not as good a design
decision. Since I dont' have that board here to check, I
can't give a definite answer for it.



No temp reading is handled by software, never.
It is an entirely electrical hard-wired feature. However,
the hardware sensor stores the read-temp values and any
software or bios can read this data, so software is a
supported but unnecessary feature- AMD's spec for shutdown
relies on no software, only (at most) bios firmware if the
ability is there to vary what threshold temp shuts it down,
as is often a feature.


reading?

I dont' have your board. On a well-designed board, no it
would only be used when CPU has no on-die sensor.


How does this question differ from the previous?
If MSI took the quicker or cheaper route, they could have
chosen to only use the socket-well sensor. That is a slower
sensor, but unless your heatsink falls off, (and assuming it
is properly calibrated), the differences may be minor or
irrelevant. The truth is that knowing the actual temp of
the CPU core to a high level of precision isnt' necessary
for the shut-down feature to work properly. Instead, if the
CPU ventures into an instable temp range there is still some
margin before it's hot enough to be damaged.

 >> In other words, don't buy brands you don't recognize
at all
 >> and PCChips/ECS are a gamble.

The funny part is that PCChips even puts silkscreening on
boards for some parts and then uses jumpers instead of the
part (like keyboard or USB fuses), when they could've
conceiled their cheap design better by not marking where
these parts would go, if they're not even going to use them.

Thanks for all the comments and recommendations. The reason I
mentioned those two boards only is that Tiger Direct has them deeply
discounted in some bundle deals which makes them very attractive in
terms of cost.

I did go to the Abit forum and it appears from the various threads
that people are having problems with the Tiger Direct bundled Abit
with the VIA KT600 chipset, possibly because the memory being offered
in the bundled deal isn’t the best. I think I will consider the
nVidia chipset and look at some other "deals".

Thanks again

 

[David Maynard]

I agree, when parts that are optional "might" be used. I too
see that all the time, but not when it's such a basic
component and not when none of them have it. More often
it's doen when there are optional features not implemented,
or to facilitate component/supply/sourcing changes- neither
of which apply.

Well, there are other reasons too. Like a manufacturer making boards for
multiple OEMs that have differing requirements. And then there's the
initial design with specification/component changes later. You don't go to
the expense of redoing the artwork nor the silkscreen when an assembly BOM
change suffices.

I'm not saying this is the case but simply an example of how things can and
sometimes do transpire. Say the initial design uses a chip that needs fuse
protect. Artwork is made with a fuse. A later chip
revision/upgrade/replacement/whatever has a new current limit feature
obviating the need for an external fuse. BOM change and a wire is put in
place of the previous fuse.

That's assuming some engineer didn't simply over-design it to being with
and got countermanded after the artwork was done.

 

[John Doe]

I agree, when parts that are optional "might" be used. I too
see that all the time, but not when it's such a basic
component and not when none of them have it. More often
it's doen when there are optional features not implemented,

That was my understanding too.