Asus A8N32-SLI (AMD) and P5N32-SLI (Intel) motherboards are coming very soon

[John Lewis]

See:-

http://www.asus.com/news_show.aspx?id=1304

These motherboards use the new nForce4 X16 chipsets, giving x16 PCIe
channels to BOTH video cards ( or any other card occupying the extra
"video" slot if not operating in dual-SLI mode --- x16/x16 instead of
the current x16/x1 in non-SLI mode ). They also employ some form of
passive cooling called "Stack Cool 2" -- no pictures of the
motherboards available yet.......... )

For more details on the X16 chipsets, see:-

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2493

John Lewis
- Technology early-birds are flying guinea-pigs.

 

[First of One]

These motherboards use the new nForce4 X16 chipsets, giving x16 PCIe
channels to BOTH video cards ( or any other card occupying the extra
"video" slot if not operating in dual-SLI mode --- x16/x16 instead of
the current x16/x1 in non-SLI mode ).

Blah blah, and they possess absolutely zero performance advantage over the
Asus A8N.

They also employ some form of
passive cooling called "Stack Cool 2"

Asus engineers have been looking at too many Pamela Anderson pictures...

 

[Dragoncarer]

Blah blah, and they possess absolutely zero performance advantage over the
Asus A8N.

Isn't it 8x/8x anyway?

 

[John Lewis]

Blah blah, and they possess absolutely zero performance advantage over the
Asus A8N.

I like to maximally future-proof when building a new very
high-performance system - for which I am presently beginning to list
the components. The x16 (instead of x1) extra "video" slot for those
of us not necessarily contemplating SLI is very interesting indeed -
think of the AGEIA physics chip on a x16 PCIe board, for
example............

Asus engineers have been looking at too many Pamela Anderson pictures...

Anything to keep them happy churning out new ideas in passive
cooling.

John Lewis

 

[First of One]

I like to maximally future-proof when building a new very
high-performance system - for which I am presently beginning to list
the components.

Other areas of your system will go obselete long before the PCIe x8
interface to each SLI card is maxed out. M2 CPU socket. > 4GB RAM. BTX form
factor...

The x16 (instead of x1) extra "video" slot for those
of us not necessarily contemplating SLI is very interesting indeed -
think of the AGEIA physics chip on a x16 PCIe board, for
example.

The things that must happen for AGEIA to be useful:

- Working silicon
- Board implementation (though BFG is on this)
- Good drivers
- Extension in DX or certain Open API to support it
- Game support
- A tangible improvement in speed/realism once the above requirements are
met

The last one may actually be a major stumbling block, since the AGEIA card
will have to communicate with the vid card. PCIe x16 is still *much* slower
than local video RAM access.

Anything to keep them happy churning out new ideas in passive
cooling.

http://www.hardocp.com/article.html?art=ODAx
Abit's passive heatpipe cooler is actually very effective at carrying heat
away from the southbridge and voltage regulators. However, you STILL need a
fan in the case to whisk the heat out.

 

[John Lewis]

Other areas of your system will go obselete long before the PCIe x8
interface to each SLI card is maxed out. M2 CPU socket. > 4GB RAM. BTX form
factor...

Enjoy BTX especially when your video card overheats in spillover
exhaust air from the CPU (particularly the Intel variety of CPU).
Remember that BTX was dreamed up by those power-saving wizards ---
Intel, with nary a thought of properly cooling anything else except
their inefficient CPUs, RAM and Intel's embedded-video chip-sets.
No thought of efficiently-cooling third-party peripheral cards - not
at all in Intel's financial interest. Not many takers yet for
BTX....... I wonder why ?

The things that must happen for AGEIA to be useful:

- Working silicon
- Board implementation (though BFG is on this)
- Good drivers
- Extension in DX or certain Open API to support it
- Game support
- A tangible improvement in speed/realism once the above requirements are
met

The last one may actually be a major stumbling block, since the AGEIA card
will have to communicate with the vid card.

Really ? And what does a physics chip have to communicate DIRECTLY to
the video card ?

PCIe x16 is still *much* slower
than local video RAM access.


http://www.hardocp.com/article.html?art=ODAx
Abit's passive heatpipe cooler is actually very effective at carrying heat
away from the southbridge and voltage regulators.

Yuk, and it sacrifices the parallel and serial ports for zero
benefit.. The current Asus A8N-SLI Premium passive solution is much
better than this hack. And poor old Abit is having a teeny weeny
financial problem at the moment anyway. Seems as if they have
concentrated too much on gimmicks like "Fatal1ty", OTES, uGuru etc,
and not listened to their customers. I have been a faithful Abit
customer for years from BH6 to IC7-G with nary a break. My next
motherboard purchases will NOT be Abit.

However, you STILL need a
fan in the case to whisk the heat out

Sure, of course. Preferably 2.

John Lewis

 

[First of One]

Enjoy BTX especially when your video card overheats in spillover
exhaust air from the CPU (particularly the Intel variety of CPU).
Remember that BTX was dreamed up by those power-saving wizards ---
Intel, with nary a thought of properly cooling anything else except
their inefficient CPUs, RAM and Intel's embedded-video chip-sets.
No thought of efficiently-cooling third-party peripheral cards - not
at all in Intel's financial interest. Not many takers yet for
BTX....... I wonder why ?

High-end BTX Gateway system. Keep in mind 90% of high-end video cards go
into these OEM boxes.
http://www.hardocp.com/article.html?art=ODA1

BTX was indeed dreamed by Intel with little common sense, along with PCIe,
DDR2, LGA CPU packaging... The ONLY reason the AMD64 platform adopted *one*
of these technologies is for SLI-support.

Really ? And what does a physics chip have to communicate DIRECTLY to
the video card ?

The physics output has to go somewhere, right? All the geometry transform is
still done by the video card. In fact, the more the game depends on
programmable shaders, the worse the I/O impact. All the talk about geometry
instancing and hardware flow control means the video card has to manage its
communication with the physics card.

Yuk, and it sacrifices the parallel and serial ports for zero
benefit.. The current Asus A8N-SLI Premium passive solution is much
better than this hack.

The current A8N SLI passive solution leads to nowhere. It connects the
voltage regulator heat sink with the chipset heat sink, but relies on "bleed
air" from the CPU fan to cool everything.

And poor old Abit is having a teeny weeny
financial problem at the moment anyway. Seems as if they have
concentrated too much on gimmicks like "Fatal1ty", OTES, uGuru etc,
and not listened to their customers. I have been a faithful Abit
customer for years from BH6 to IC7-G with nary a break. My next
motherboard purchases will NOT be Abit.

Abit, DFI, these brands have no OEM sales, so their marketshares are almost
negligible compared to the giants like Asus, ECS, and MSI. All it takes is
for a few enthusiasts to fart the wrong way and Abit's financials are the
gutter. Making a desirable product isn't enough. Chaintech was smart to exit
the M/B business.

And Fatal1ty isn't all a gimmick. The Fatal1ty X800XL is the first retail
card with factory-backed overvoltage adjustments, the key to successful
overclocks.

Sure, of course. Preferably 2.

Which partially negates the benefit of a passive (and expensive) heat-pipe
solution. Sure, the fans are bigger, but noiseless they aint.

 

[John Lewis]

The physics output has to go somewhere, right?

Yes, back to the primary CPU, of course -- just as it would if the
physics computation was executed on the second core of a dual-core GPU
( or a SPE, in the case of PS3 ), except that the dedicated AGEIA
hardware would execute faster.

All the geometry transform is
still done by the video card.

Sure, after it has been told where to place the 'objects' by the
CPU/physics algorithms.

In fact, the more the game depends on
programmable shaders, the worse the I/O impact. All the talk about geometry
instancing and hardware flow control means the video card has to manage its
communication with the physics card.


The current A8N SLI passive solution leads to nowhere. It connects the
voltage regulator heat sink with the chipset heat sink, but relies on "bleed
air" from the CPU fan to cool everything.

Actually from the case fan ( which is far more useful and effective
than Abit's noisy, junk OTES )

Abit, DFI, these brands have no OEM sales, so their marketshares are almost
negligible compared to the giants like Asus, ECS, and MSI. All it takes is
for a few enthusiasts to fart the wrong way and Abit's financials are the
gutter. Making a desirable product isn't enough. Chaintech was smart to exit
the M/B business.

And Fatal1ty isn't all a gimmick. The Fatal1ty X800XL is the first retail
card with factory-backed overvoltage adjustments, the key to successful
overclocks.

Huh ?? <<Factory-backed>> overvoltage adjustments ?? What's the
difference between that and the normal overvoltage adjustments on
enthusiast boards ?
More Abit weasel-words like OTES, uGURU etc... ?

Which partially negates the benefit of a passive (and expensive) heat-pipe
solution. Sure, the fans are bigger, but noiseless they aint.

Ah, but chip-set fans are notoriously unreliable, which is the primary
reason for the move to passive MB cooling - mostly a function of heat
and consequent lack of bearing-lubrication. And
a 120mm 1500RPM case fan is a lot less noisy than a 25mm 6000RPM
whiner.


John Lewis

 

[First of One]

Sure, after it has been told where to place the 'objects' by the
CPU/physics algorithms.

Today's games are not CPU-limited at high resolutions, so designing hardware
to offload work from the CPU won't provide any gain. And we are assuming in
future games the CPU actually does most of the physics algorithms, all the
while nVidia is numerically integrating wave equations on the GPU to
generate water effects. The more advanced shader programming becomes, the
more we have to rely on the video card to handle all the flow control and
load-balancing.

An AGEIA card would be an ass-backwards solution. If anything the chip
should be integrated onto the video card, to remove the I/O bottleneck. Even
then it still needs API-, driver-, and game-support. After all that, it
still needs to show a tangible performance/quality benefit.

Actually from the case fan ( which is far more useful and effective
than Abit's noisy, junk OTES )

You don't need a hype-power heatpipe solution to get benefit from case fans.
The southbridge is near the front fan. The voltage regulators are near the
rear fan. Transferring heat to the back of the PCB like Asus' new design
does squat because there's no airflow there.

Huh ?? <<Factory-backed>> overvoltage adjustments ?? What's the
difference between that and the normal overvoltage adjustments on
enthusiast boards ?

Read it carefully, "Fatal1ty X800XL". When was the last time you saw easy
overvoltage adjustments on a video card? Overclocking and cranking up the
voltage on video cards is where it counts, since most games are fillrate- or
shader-limited at the resolutions we want to play at. Even a watercooling
rig on the video card doesn't do much without the ability to crank up the
voltage.

Ah, but chip-set fans are notoriously unreliable, which is the primary
reason for the move to passive MB cooling - mostly a function of heat
and consequent lack of bearing-lubrication.

See above. The incremental performance of a heat pipe isn't much compared to
just plain heat sinks when the sinks are all close to case fans anyway.
Asus' solution is just as gimmicky as Abit's.