Newbie to Dual Video Cards..... Question

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

I have a Dell laptop and have never experienced Dual cards. I am
getting ready to get a new desktop system. I have been reading about
Dual NVIDIA GeForce™ 6800 256 megabyte cards.

Ok here is my question. If you have dual 128 Megabyte cards is
performance the same or worse or better than having just 1 256 meg
card.

Ok another question as a follow up... If you have Dual 256 meg cards
how much performance gain do you have over 1 256 meg card.

Thank you for your time.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Oh Yes, Sorry I was meaning SLI.

What do people think of Geforce SLI 2 256 meg cards?

There are several models og 256 megabyte GEForce cards, with different
price points. Are the more expensive versions really worth that extra
money compared to added features and/or performance?

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Conservative.Nate@gmail.com wrote:
> I have a Dell laptop and have never experienced Dual cards. I am
> getting ready to get a new desktop system. I have been reading about
> Dual NVIDIA GeForceT 6800 256 megabyte cards.
>
> Ok here is my question. If you have dual 128 Megabyte cards is
> performance the same or worse or better than having just 1 256 meg
> card.

I'm assuming you mean SLI (two video GPUs working together) when you say
"dual", and not dual head (two video outputs) which is what "dual" will have
most people think of in a video card context.

The memory doesn't add up, so two 128 MB cards in SLI will only give you 128
MB memory.

Performancewise, an SLI solution will be faster than a single GPU if the
software is written to take advantage of it, or the drivers know about that
piece of software and can help out. In other cases, it won't be faster than
a single GPU, and in some cases it will be slightly slower due to more
overhead in the drivers.
And with two 128 MB cards instead of a single 256 MB one, you'll have less
usable memory.

> Ok another question as a follow up... If you have Dual 256 meg cards
> how much performance gain do you have over 1 256 meg card.

In the range of -2% to 85%, according to the tests I've seen, and depending
on what you run.

Regards,
--
*Art

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Conservative.Nate@gmail.com wrote:
> I have a Dell laptop and have never experienced Dual cards. I am
> getting ready to get a new desktop system. I have been reading about
> Dual NVIDIA GeForce™ 6800 256 megabyte cards.
>
> Ok here is my question. If you have dual 128 Megabyte cards is
> performance the same or worse or better than having just 1 256 meg
> card.
>
> Ok another question as a follow up... If you have Dual 256 meg cards
> how much performance gain do you have over 1 256 meg card.
>
> Thank you for your time.
>

SLI is a waste of time, money and effort. Skip it and either snag a
mobo with pcie (a good one is hard to find, not to mention it's barely
supported) or stick with AGP, which doesn't seem to be going away
anytime soon.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Wijja wrote:

> Conservative.Nate@gmail.com wrote:
>> I have a Dell laptop and have never experienced Dual cards. I am
>> getting ready to get a new desktop system. I have been reading about
>> Dual NVIDIA GeForce? 6800 256 megabyte cards.
>>
>> Ok here is my question. If you have dual 128 Megabyte cards is
>> performance the same or worse or better than having just 1 256 meg
>> card.
>>
>> Ok another question as a follow up... If you have Dual 256 meg cards
>> how much performance gain do you have over 1 256 meg card.
>>
>> Thank you for your time.
>>
>
> SLI is a waste of time, money and effort. Skip it and either snag a
> mobo with pcie (a good one is hard to find, not to mention it's barely
> supported) or stick with AGP, which doesn't seem to be going away
> anytime soon.

Huh? (a) there are many decent PCI Express motherboards. Any with a 900
series Intel chipset has it, as do those with the latest nvidia and via
chipsets for AMD. (b) while SLI may be a "waste of time, money, and
effort" it is available only on PCI Express motherboards, and many of the
"better" ones provide it. (c) PCI Express doesn't need any special kind of
"support". (d) Intel has decreed that AGP Shall Die and they appear well
on the way to getting their wish in that regard.

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

In article <d3mepa02a7i@news4.newsguy.com>, J. Clarke says...

> (b) while SLI may be a "waste of time, money, and
> effort" it is available only on PCI Express motherboards,

WRONG. SLI first appeared in the late 90's on PCI boards with the 3Dfx
cards.

--
Conor

"Of all the things I've lost, I miss my mind the most." O.Osbourne.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Conor Turton wrote:

> In article <d3mepa02a7i@news4.newsguy.com>, J. Clarke says...
>
>> (b) while SLI may be a "waste of time, money, and
>> effort" it is available only on PCI Express motherboards,
>
> WRONG. SLI first appeared in the late 90's on PCI boards with the 3Dfx
> cards.

You are talking about a different product with the same name. The only
resemblance between the current technology called "SLI" and the one that
3dfx was shipping is that both allow two video boards to feed one monitor.

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Wijja wrote:

> SLI is a waste of time, money and effort. Skip it and either snag a
> mobo with pcie (a good one is hard to find, not to mention it's barely
> supported

BS. PCIe already is very well supported, and not only by Windows...

> ) or stick with AGP, which doesn't seem to be going away
> anytime soon.

AGP is dead. All new chipsets and mobos are PCIe. Not to forget that most
current and certainly all future GPU generations are PCIe-only (only made
AGP compatible by brigde chips which also limit performance)...

Buying an AGP system today is riding on a dead horse...

Benjamin

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Benjamin Gawert wrote:

> Wijja wrote:
>
>> SLI is a waste of time, money and effort. Skip it and either snag a
>> mobo with pcie (a good one is hard to find, not to mention it's barely
>> supported
>
> BS. PCIe already is very well supported, and not only by Windows...
>
>> ) or stick with AGP, which doesn't seem to be going away
>> anytime soon.
>
> AGP is dead. All new chipsets and mobos are PCIe. Not to forget that most
> current and certainly all future GPU generations are PCIe-only (only made
> AGP compatible by brigde chips which also limit performance)...

Do you have any hard examples where the performance of any current
production board is "limited" by the AGP bridge chip? I'm not talking
about some kind of theoretical limit based on unusable bandwidth but a
real-world difference greater than the few percent that often occurs
between different manufacturers using the same chipset?

> Buying an AGP system today is riding on a dead horse...

Unfortunately this is true, but that is due to the abuse of power by Intel
and not to any innate superiority of PCI Express.

>
> Benjamin

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

In article <d3msqi12n3s@news4.newsguy.com>, J. Clarke says...

> > WRONG. SLI first appeared in the late 90's on PCI boards with the 3Dfx
> > cards.
>
> You are talking about a different product with the same name. The only
> resemblance between the current technology called "SLI" and the one that
> 3dfx was shipping is that both allow two video boards to feed one monitor.
>
Oh dear.....you've been into PCs like what, 5 years tops?

They worked the same. One card did so many display lines, the other did
the rest. Both worked on exactly the same principle. Oh and dual
monitor support doesn't work on the current PCIe cards in SLI mode.


--
Conor

"Of all the things I've lost, I miss my mind the most." O.Osbourne.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Conor Turton wrote:

> In article <d3msqi12n3s@news4.newsguy.com>, J. Clarke says...
>
>> > WRONG. SLI first appeared in the late 90's on PCI boards with the 3Dfx
>> > cards.
>>
>> You are talking about a different product with the same name. The only
>> resemblance between the current technology called "SLI" and the one that
>> 3dfx was shipping is that both allow two video boards to feed one
>> monitor.
>>
> Oh dear.....you've been into PCs like what, 5 years tops?

So let's see, how many IBM 360 core dumps do _you_ have in your junk pile,
grandpa?

> They worked the same. One card did so many display lines, the other did
> the rest.

Nope. 3dfx "SLI" stood for "Scan Line Interleaving". One board did half
the scan lines, the other did the rest. That is _not_ a mode that is
supported by nvidia "SLI", which either splits each frame into two parts or
has the boards alternate, with one board rendering one frame and the other
rendering the next.

> Both worked on exactly the same principle.

Only if an airplane and an airship work on exactly the same principle.

You might want to google "SLI nvidia 3dfx" and do a little research before
you parade your ignorance again.

> Oh and dual
> monitor support doesn't work on the current PCIe cards in SLI mode.

I don't believe that I ever claimed that it did, so what is your point?

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Conor Turton <conor@conorturton.com> wrote:
> In article <d3msqi12n3s@news4.newsguy.com>, J. Clarke says...
>
>> You are talking about a different product with the same name. The
>> only resemblance between the current technology called "SLI" and the
>> one that 3dfx was shipping is that both allow two video boards to
>> feed one monitor.
>>
> Oh dear.....you've been into PCs like what, 5 years tops?
>
> They worked the same. One card did so many display lines, the other
> did the rest. Both worked on exactly the same principle.

Not only that, but nVidia bought 3dfx who held the patent on it, which is
why they can come out with SLI. They've changed what the SLI acronym stands
for, and modernized it for a different generation of cards, but there's
little question that this is one of the reasults of nVidia taking over all
3dfx assets.

Now the question is, will we see a monster with 4 GPUs, like the Voodoo5
6000 had? Or a 2x2 solution?

Regards,
--
*Art

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Arthur Hagen wrote:

> Conor Turton <conor@conorturton.com> wrote:
>> In article <d3msqi12n3s@news4.newsguy.com>, J. Clarke says...
>>
>>> You are talking about a different product with the same name. The
>>> only resemblance between the current technology called "SLI" and the
>>> one that 3dfx was shipping is that both allow two video boards to
>>> feed one monitor.
>>>
>> Oh dear.....you've been into PCs like what, 5 years tops?
>>
>> They worked the same. One card did so many display lines, the other
>> did the rest. Both worked on exactly the same principle.
>
> Not only that, but nVidia bought 3dfx who held the patent on it, which is
> why they can come out with SLI. They've changed what the SLI acronym
> stands for, and modernized it for a different generation of cards, but
> there's little question that this is one of the reasults of nVidia taking
> over all 3dfx assets.

If in fact it worked anything like the 3dfx product there might be some
justice to that, but the current "SLI" doesn't work by rendering alternate
scanlines, it works by rendering half the screen in a block or by rendering
an entire screen and then letting the second board render the next.

> Now the question is, will we see a monster with 4 GPUs, like the Voodoo5
> 6000 had? Or a 2x2 solution?
>
> Regards,

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

In article <d3oahl01p7d@news1.newsguy.com>, J. Clarke says...

> So let's see, how many IBM 360 core dumps do _you_ have in your junk pile,
> grandpa?
>
Found Google I see.

> > They worked the same. One card did so many display lines, the other did
> > the rest.
>
> Nope. 3dfx "SLI" stood for "Scan Line Interleaving". One board did half
> the scan lines, the other did the rest. That is _not_ a mode that is
> supported by nvidia "SLI", which either splits each frame into two parts or
> has the boards alternate, with one board rendering one frame and the other
> rendering the next.
>
Exactly the same principle as the 3Dfx ones with the exception that one
did alternating scan lines, the other halves of the frame.


>

--
Conor

"Of all the things I've lost, I miss my mind the most." O.Osbourne.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Conor wrote:

> In article <d3oahl01p7d@news1.newsguy.com>, J. Clarke says...
>
>> So let's see, how many IBM 360 core dumps do _you_ have in your junk
>> pile, grandpa?
>>
> Found Google I see.

Huh? What does Google have to do with anything?

>> > They worked the same. One card did so many display lines, the other did
>> > the rest.
>>
>> Nope. 3dfx "SLI" stood for "Scan Line Interleaving". One board did half
>> the scan lines, the other did the rest. That is _not_ a mode that is
>> supported by nvidia "SLI", which either splits each frame into two parts
>> or has the boards alternate, with one board rendering one frame and the
>> other rendering the next.
>>
> Exactly the same principle as the 3Dfx ones with the exception that one
> did alternating scan lines, the other halves of the frame.

I see. If you believe that that is "the same principle" then you also must
believe that airships and airplanes work on "the same principle" because
both of them fly.

However since you seem to (a) not know much about engineering and (b) seem
to enjoy feeble attempts at being abusive, there's not much point in
continuing this until you grow up.

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

J. Clarke wrote:
> Conor wrote:
>
>
>>In article <d3oahl01p7d@news1.newsguy.com>, J. Clarke says...
>>
>>
>>>So let's see, how many IBM 360 core dumps do _you_ have in your junk
>>>pile, grandpa?
>>>
>>
>>Found Google I see.
>
>
> Huh? What does Google have to do with anything?
>
>
>>>>They worked the same. One card did so many display lines, the other did
>>>>the rest.
>>>
>>>Nope. 3dfx "SLI" stood for "Scan Line Interleaving". One board did half
>>>the scan lines, the other did the rest. That is _not_ a mode that is
>>>supported by nvidia "SLI", which either splits each frame into two parts
>>>or has the boards alternate, with one board rendering one frame and the
>>>other rendering the next.
>>>
>>
>>Exactly the same principle as the 3Dfx ones with the exception that one
>>did alternating scan lines, the other halves of the frame.
>
>
> I see. If you believe that that is "the same principle" then you also must
> believe that airships and airplanes work on "the same principle" because
> both of them fly.
>

Since no one else picked up on this, I guess I will....

Both fly using lift........



One produces lift via drag, the other via lighter than air gasses.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Arthur Hagen wrote:

> Conor Turton <conor@conorturton.com> wrote:
>
>>In article <d3msqi12n3s@news4.newsguy.com>, J. Clarke says...
>>
>>
>>>You are talking about a different product with the same name. The
>>>only resemblance between the current technology called "SLI" and the
>>>one that 3dfx was shipping is that both allow two video boards to
>>>feed one monitor.
>>>
>>
>>Oh dear.....you've been into PCs like what, 5 years tops?
>>
>>They worked the same. One card did so many display lines, the other
>>did the rest. Both worked on exactly the same principle.
>
>
> Not only that, but nVidia bought 3dfx who held the patent on it, which is
> why they can come out with SLI. They've changed what the SLI acronym stands
> for, and modernized it for a different generation of cards, but there's
> little question that this is one of the reasults of nVidia taking over all
> 3dfx assets.
>
> Now the question is, will we see a monster with 4 GPUs, like the Voodoo5
> 6000 had? Or a 2x2 solution?
>
> Regards,

An interesting side note is the video card for the Mac G5 that runs the
huge LCD panel (31" I believe) is a dual, dual card. That is, there are
4 DVIs on the back. Each powers half of the panel, therefore 2 DVIs are
required to run one panel (dual port) and each card will run two panels
(dual dual, as it were). This seems to be more in line with the "old" SLI.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

Benjamin Gawert wrote:

> Wijja wrote:
>
>
>>SLI is a waste of time, money and effort. Skip it and either snag a
>>mobo with pcie (a good one is hard to find, not to mention it's barely
>>supported
>
>
> BS. PCIe already is very well supported, and not only by Windows...
>
>
>>) or stick with AGP, which doesn't seem to be going away
>>anytime soon.
>
>
> AGP is dead. All new chipsets and mobos are PCIe. Not to forget that most
> current and certainly all future GPU generations are PCIe-only (only made
> AGP compatible by brigde chips which also limit performance)...
>
> Buying an AGP system today is riding on a dead horse...
>
> Benjamin
>
>

I'd say more like a dying horse.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

J. Clarke wrote:

> Benjamin Gawert wrote:
>
>
>>Wijja wrote:
>>
>>
>>>SLI is a waste of time, money and effort. Skip it and either snag a
>>>mobo with pcie (a good one is hard to find, not to mention it's barely
>>>supported
>>
>>BS. PCIe already is very well supported, and not only by Windows...
>>
>>
>>>) or stick with AGP, which doesn't seem to be going away
>>>anytime soon.
>>
>>AGP is dead. All new chipsets and mobos are PCIe. Not to forget that most
>>current and certainly all future GPU generations are PCIe-only (only made
>>AGP compatible by brigde chips which also limit performance)...
>
>
> Do you have any hard examples where the performance of any current
> production board is "limited" by the AGP bridge chip? I'm not talking
> about some kind of theoretical limit based on unusable bandwidth but a
> real-world difference greater than the few percent that often occurs
> between different manufacturers using the same chipset?
>
>
>>Buying an AGP system today is riding on a dead horse...
>
>
> Unfortunately this is true, but that is due to the abuse of power by Intel
> and not to any innate superiority of PCI Express.
>
>

Hey, may I have some of whatever it is you are smoking? PCI Express is
not only replacing AGP, it is replacing PCI itself.

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

No one wrote:

> J. Clarke wrote:
>> Conor wrote:
>>
>>
>>>In article <d3oahl01p7d@news1.newsguy.com>, J. Clarke says...
>>>
>>>
>>>>So let's see, how many IBM 360 core dumps do _you_ have in your junk
>>>>pile, grandpa?
>>>>
>>>
>>>Found Google I see.
>>
>>
>> Huh? What does Google have to do with anything?
>>
>>
>>>>>They worked the same. One card did so many display lines, the other did
>>>>>the rest.
>>>>
>>>>Nope. 3dfx "SLI" stood for "Scan Line Interleaving". One board did
>>>>half
>>>>the scan lines, the other did the rest. That is _not_ a mode that is
>>>>supported by nvidia "SLI", which either splits each frame into two parts
>>>>or has the boards alternate, with one board rendering one frame and the
>>>>other rendering the next.
>>>>
>>>
>>>Exactly the same principle as the 3Dfx ones with the exception that one
>>>did alternating scan lines, the other halves of the frame.
>>
>>
>> I see. If you believe that that is "the same principle" then you also
>> must believe that airships and airplanes work on "the same principle"
>> because both of them fly.
>>
>
> Since no one else picked up on this, I guess I will....
>
> Both fly using lift........
>
>
>
> One produces lift via drag, the other via lighter than air gasses.

Nope. Not quite. In fact not really close. One flies by dynamic pressure,
the other by static. Very different principles.

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)

 

[Guest]

Archived from groups: alt.comp.periphs.videocards.nvidia (More info?)

No one wrote:

> J. Clarke wrote:
>
>> Benjamin Gawert wrote:
>>
>>
>>>Wijja wrote:
>>>
>>>
>>>>SLI is a waste of time, money and effort. Skip it and either snag a
>>>>mobo with pcie (a good one is hard to find, not to mention it's barely
>>>>supported
>>>
>>>BS. PCIe already is very well supported, and not only by Windows...
>>>
>>>
>>>>) or stick with AGP, which doesn't seem to be going away
>>>>anytime soon.
>>>
>>>AGP is dead. All new chipsets and mobos are PCIe. Not to forget that most
>>>current and certainly all future GPU generations are PCIe-only (only made
>>>AGP compatible by brigde chips which also limit performance)...
>>
>>
>> Do you have any hard examples where the performance of any current
>> production board is "limited" by the AGP bridge chip? I'm not talking
>> about some kind of theoretical limit based on unusable bandwidth but a
>> real-world difference greater than the few percent that often occurs
>> between different manufacturers using the same chipset?
>>
>>
>>>Buying an AGP system today is riding on a dead horse...
>>
>>
>> Unfortunately this is true, but that is due to the abuse of power by
>> Intel and not to any innate superiority of PCI Express.
>>
>>
>
> Hey, may I have some of whatever it is you are smoking? PCI Express is
> not only replacing AGP, it is replacing PCI itself.

And why is it replacing PCI? Because people are crying for it or because
Intel stuck it on all their chipset and gave the manufacturers no real
choice but to use it or abandon Intel?

--
--John
to email, dial "usenet" and validate
(was jclarke at eye bee em dot net)