Value of integrated graphics on mobo?

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Psi-Tau Paladin wrote:
> Piotr Makley <pmakley@mail.com> wrote in news:94C781C1ACFA131E75@
> 130.133.1.4:
>
>
>>Are PCI graphics cards more or less expensive than the same card
>>for AGP?
>
>
> They are going up in price since not many people make them anymore.

It depends on the card. Most PCI cards *are* more expensive
than the AGP version of the same card, but for the Radeon 9200
the prices are about the same.

I suspect the reason is that the PCI version of the Radeon 9200
is selling like hotcakes. Hence better economies of scale come
into play for the manufacturer and their is more competition
among the vendors. This card is selling *very* well to people
who have integrated video and no AGP slot.

It is also a good card for people who already have a dual monitor
AGP card and want a cheap upgrade to a quad-display system. The PCI
Radeon 9200 plays nicely together with all of the AGP Radeon
8xxx and 9xxx cards.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

"Psi-Tau Paladin" <keairemoonMAPSON@hotmail.com> wrote:

>> Are PCI graphics cards more or less expensive than the same
>> card for AGP?
>
> They are going up in price since not many people make them
> anymore.
>

Although OCI graphics cards may be going up in price --- is that
from a base price which is lower or highter that the price of AGP
graphics cards?

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Piotr Makley <pmakley@mail.com> wrote in news:94C797CE8FC2031E75@
130.133.1.4:

> Although OCI graphics cards may be going up in price --- is that
> from a base price which is lower or highter that the price of AGP
> graphics cards?
They were still cheaper about 2 or so years ago and they were equal about
last year due to supply. At this point in time I would say that most
cards in the pci version will cost more than the equivilant AGP version
due to limited supply. Of course there will always be places trying to
get rid of stock.

i.e.
http://computing.kelkoo.co.uk/b/a/cp_111601_brand_pny.html

PNY Verto GeForce FX 5200 PCI (128 MB) = 69 pounds
PNY Verto GeForce FX 5200 Ultra AGP (128 MB) = 60 pounds

Pine XFX MX 400 PCI (64 MB) = 36 pounds
Pine XFX MX 400 AGP (64 MB) = 27 pounds

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Rob Stow wrote:
>> The amount of data transfered to the graphic card in 3d games is
>> completely independant of the resolution, since typically games do
>> not scale geometry details (which would make the amount of data
>> transfered different), thus this statement is completely false.
>
>
> Do a simple test: put a Radeon 9200 in a PCI slot. Run some gaming
> benchmarks. Repeat the benchmarks with an AGP version of that card
> in that same machine: same GPU running at the same clock, same type
> and amount of RAM at the same speed. Note that the benchmarks are
> very nearly identical at low resolutions, but the AGP card edges
> ahead at higher resolutions.
This is simply not true at all. If you did test this, you likely didn't
use identical cards - for instance you might have got the 64bit memory
interface version of the 9200 (afaik only hercules produces a 128bit pci
version) which is only about half as fast (and at lower resolutions
could still keep up if your game is cpu limited).
There were lots of comparisons between AGP and PCI when AGP was new, and
resolution just doesn't matter.
Resolution COULD make a difference, but only if you're running out of
local graphic memory because of the higher resolution - but this is
unlikely to happen, since the z/frame/back buffer don't use that much
memory (if you're not using FSAA which you can't really use with those
low-end cards).

>
> You missed in my previous post that my original AGP vs PCI comparison
> was for a Radeon 9200 - same GPU, same amount of RAM on the card.
> How then do you explain that the PCI version of the card keeps up
> with the AGP version until higher resolutions are reached ?
Since it's just not true, I don't need an explanation ;-)

Roland

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Roland Scheidegger wrote:
> There were lots of comparisons between AGP and PCI when AGP was new, and
> resolution just doesn't matter.
> Resolution COULD make a difference, but only if you're running out of
> local graphic memory because of the higher resolution - but this is
> unlikely to happen, since the z/frame/back buffer don't use that much
> memory (if you're not using FSAA which you can't really use with those
> low-end cards).

Here's a recent test, and you can actually see that the difference gets
SMALLER between pci and agp with higher resolution (which actually is
expected, since the same amount of geometry data is transfered, but the
card has to work harder thus lower framerates, and the difference is
quite small to begin with at 800x600 and non-existant at higher
resolutions (compare the club3d lp and sapphire pci 64MB card, those
have same clocks, both 64bit memory interface, though the pci card has
only 64MB while the agp card has 128MB).
http://www.ati-news.de/HTML/Berichte/Sapphire/R9200-PCI/Sapphire-R9200-PCI-Seite5.shtml
(btw forget the aquamark results, this test definitely penalizes cards
with less ram, the 3dmark01 results though show that there is indeed a
difference between pci and agp cards - but again, the difference does
not grow with higher resolution).

Roland

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Roland Scheidegger <rscheidegger@gmx.ch> wrote:

> Roland Scheidegger wrote:
>> There were lots of comparisons between AGP and PCI when AGP
>> was new, and resolution just doesn't matter.
>> Resolution COULD make a difference, but only if you're
>> running out of local graphic memory because of the higher
>> resolution - but this is unlikely to happen, since the
>> z/frame/back buffer don't use that much memory (if you're not
>> using FSAA which you can't really use with those low-end
>> cards).
>
> Here's a recent test, and you can actually see that the
> difference gets SMALLER between pci and agp with higher
> resolution (which actually is expected, since the same amount
> of geometry data is transfered, but the card has to work
> harder thus lower framerates, and the difference is quite
> small to begin with at 800x600 and non-existant at higher
> resolutions (compare the club3d lp and sapphire pci 64MB card,
> those have same clocks, both 64bit memory interface, though
> the pci card has only 64MB while the agp card has 128MB).
> http://www.ati-news.de/HTML/Berichte/Sapphire/R9200-PCI/Sapphir
> e-R9200-PCI-Seite5.shtml (btw forget the aquamark results,
> this test definitely penalizes cards with less ram, the
> 3dmark01 results though show that there is indeed a difference
> between pci and agp cards - but again, the difference does not
> grow with higher resolution).
>


Roland, I am getting confused following this thread.

Are you saying that there is very little difference between the
same PCI and AGP graphics cards?

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Piotr Makley wrote:
> Roland, I am getting confused following this thread.
>
> Are you saying that there is very little difference between the
> same PCI and AGP graphics cards?

Yes, if the cards are otherwise identical.
(And keep in mind only low-end cards are available for pci, with
high-end cards which are much, much faster today you'd see more difference.)

Roland

 

[Michael]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

But hasn't ATI continued to build/release up to date Radeon cpu based video
cards? They're pretty high tech? OTOH, the PCI Radeon cards are much more
expensive than similar AGP based video cards?

"Roland Scheidegger" <rscheidegger@gmx.ch> wrote in message
news:c57f31$2otu17$1@ID-84205.news.uni-berlin.de...
> Rob Stow wrote:
> > A standard 32 bit/33 MHz PCI bus compares well with AGP 1x.
> > It has more than enough bandwidth for 2D work and is good
> > enough for 3D games if you are playing at a low res like
> > 640x480 or 800x600.
> The amount of data transfered to the graphic card in 3d games is
> completely independant of the resolution, since typically games do not
> scale geometry details (which would make the amount of data transfered
> different), thus this statement is completely false.
> The reason PCI graphics cards are too slow at higher resolutions has
> nothing to do with the pci bus itself, but simply because only
> lowest-end graphic chips are available compared to AGP card (high-end
> graphic chips would definitely be limited by the pci bus in newer games,
> but you could easily crank up resolution as much as you'd wanted without
> loosing performance if such pci cards would exist).
>
> Roland

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Michael wrote:

> But hasn't ATI continued to build/release up to date Radeon cpu based video
> cards? They're pretty high tech? OTOH, the PCI Radeon cards are much more
> expensive than similar AGP based video cards?

No, the Radeon 9200's and 9200SE's are very similarly
priced for the AGP and PCI versions - and at some
vendors they have *exactly* the same price.

For example, about five weeks ago I bought four
128 MB ATI Radeon 9200 PCI cards for $139.95 each
at FutureShop (Canadian electronics store chain).
The AGP version of the card at that store had exactly
the same price.



>
> "Roland Scheidegger" <rscheidegger@gmx.ch> wrote in message
> news:c57f31$2otu17$1@ID-84205.news.uni-berlin.de...
>
>>Rob Stow wrote:
>>
>>>A standard 32 bit/33 MHz PCI bus compares well with AGP 1x.
>>>It has more than enough bandwidth for 2D work and is good
>>>enough for 3D games if you are playing at a low res like
>>>640x480 or 800x600.
>>
>>The amount of data transfered to the graphic card in 3d games is
>>completely independant of the resolution, since typically games do not
>>scale geometry details (which would make the amount of data transfered
>>different), thus this statement is completely false.
>>The reason PCI graphics cards are too slow at higher resolutions has
>>nothing to do with the pci bus itself, but simply because only
>>lowest-end graphic chips are available compared to AGP card (high-end
>>graphic chips would definitely be limited by the pci bus in newer games,
>>but you could easily crank up resolution as much as you'd wanted without
>>loosing performance if such pci cards would exist).
>>
>>Roland
>
>
>

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Rob Stow wrote:
> Michael wrote:
>
>> But hasn't ATI continued to build/release up to date Radeon cpu based
>> video
>> cards? They're pretty high tech? OTOH, the PCI Radeon cards are much
>> more
>> expensive than similar AGP based video cards?
>
>
> No, the Radeon 9200's and 9200SE's are very similarly
> priced for the AGP and PCI versions - and at some
> vendors they have *exactly* the same price.
>
> For example, about five weeks ago I bought four
> 128 MB ATI Radeon 9200 PCI cards for $139.95 each
> at FutureShop (Canadian electronics store chain).
> The AGP version of the card at that store had exactly
> the same price.

Are you sure though the PCI cards have a 128bit memory interface? AFAIK
PCI cards always have the name "9200" and not "9200SE" even if in fact
sometimes they only have the 64bit interface (you can see it easily, if
they have only 4 (tsop) memory chips in total, then they have a 64bit
memory interface).
And of course these 9200 cards are not exactly "up to date", they might
not be that old but the technology is rather outdated.

Roland

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Roland Scheidegger wrote:

> Rob Stow wrote:
>
>> Michael wrote:
>>
>>> But hasn't ATI continued to build/release up to date Radeon cpu based
>>> video
>>> cards? They're pretty high tech? OTOH, the PCI Radeon cards are much
>>> more
>>> expensive than similar AGP based video cards?
>>
>>
>>
>> No, the Radeon 9200's and 9200SE's are very similarly
>> priced for the AGP and PCI versions - and at some
>> vendors they have *exactly* the same price.
>>
>> For example, about five weeks ago I bought four
>> 128 MB ATI Radeon 9200 PCI cards for $139.95 each
>> at FutureShop (Canadian electronics store chain).
>> The AGP version of the card at that store had exactly
>> the same price.
>
>
> Are you sure though the PCI cards have a 128bit memory interface? AFAIK
> PCI cards always have the name "9200" and not "9200SE" even if in fact
> sometimes they only have the 64bit interface (you can see it easily, if
> they have only 4 (tsop) memory chips in total, then they have a 64bit
> memory interface).
> And of course these 9200 cards are not exactly "up to date", they might
> not be that old but the technology is rather outdated.

Umm you can make a 128bit interface by just moving it twice rather than
once. The PCI bus doesn't support 128bit wide datatransfers anyway so
you just write twice. Another reason why AGP rules

Though i would still like to see a dual AGP system one day.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Yoyoma_2 <Yoyoma_2@[at-]Hotmail.com> wrote:

>> Are you sure though the PCI cards have a 128bit memory
>> interface? AFAIK PCI cards always have the name "9200" and
>> not "9200SE" even if in fact sometimes they only have the
>> 64bit interface (you can see it easily, if they have only 4
>> (tsop) memory chips in total, then they have a 64bit memory
>> interface). And of course these 9200 cards are not exactly
>> "up to date", they might not be that old but the technology
>> is rather outdated.
>
> Umm you can make a 128bit interface by just moving it twice
> rather than once. The PCI bus doesn't support 128bit wide
> datatransfers anyway so you just write twice. Another reason
> why AGP rules
>
> Though i would still like to see a dual AGP system one day.


Do you mean to say that there are almost no dual monitor AGP
graphics cards?

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Piotr Makley wrote:
> Yoyoma_2 <Yoyoma_2@[at-]Hotmail.com> wrote:
>>
>>Though i would still like to see a dual AGP system one day.
>
> Do you mean to say that there are almost no dual monitor AGP
> graphics cards?

There are many, many, dual monitor AGP cards out there -
it has almost become the standard for AGP cards.

Hence I took her (his?) remark as intending to convey a
desire for a machine with more than one AGP slot, to
which the answer is, "Forget it - it ain't going to happen."
Especially now with PCI-Express just around the corner.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Piotr Makley wrote:

> Yoyoma_2 <Yoyoma_2@[at-]Hotmail.com> wrote:
>
>
>>>Are you sure though the PCI cards have a 128bit memory
>>>interface? AFAIK PCI cards always have the name "9200" and
>>>not "9200SE" even if in fact sometimes they only have the
>>>64bit interface (you can see it easily, if they have only 4
>>>(tsop) memory chips in total, then they have a 64bit memory
>>>interface). And of course these 9200 cards are not exactly
>>>"up to date", they might not be that old but the technology
>>>is rather outdated.
>>
>>Umm you can make a 128bit interface by just moving it twice
>>rather than once. The PCI bus doesn't support 128bit wide
>>datatransfers anyway so you just write twice. Another reason
>>why AGP rules
>>
>>Though i would still like to see a dual AGP system one day.
>
>
>
> Do you mean to say that there are almost no dual monitor AGP
> graphics cards?

No i mean there are almost no boards with dual AGP slots where you can
add a 2 quad monitor AGP cards if you want

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Rob Stow wrote:

> Piotr Makley wrote:
>
>> Yoyoma_2 <Yoyoma_2@[at-]Hotmail.com> wrote:
>>
>>>
>>> Though i would still like to see a dual AGP system one day.
>>
>>
>> Do you mean to say that there are almost no dual monitor AGP graphics
>> cards?
>
>
> There are many, many, dual monitor AGP cards out there -
> it has almost become the standard for AGP cards.
>
> Hence I took her (his?) remark as intending to convey a
> desire for a machine with more than one AGP slot, to
> which the answer is, "Forget it - it ain't going to happen."
> Especially now with PCI-Express just around the corner.

Yeah i clarified that in another post. The original intent was because
PCI-66mhz is still pretty slow, especially when handling large data.
PCI-X is really good but your right its not adopted yet. But what would
prevent a dual AGP slot motherboard? It could be usefull when trying to
render high-performance simulations that are sectioned. Though i don't
know if the bus supports it.

From what i know, PCI-X still won't have the same kind of access that
the AGP card has, like having to go through a PCI bridge, direct access
to memory with textures etc...

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Yoyoma_2 wrote:

> Rob Stow wrote:
>
>> Piotr Makley wrote:
>>
>>> Yoyoma_2 <Yoyoma_2@[at-]Hotmail.com> wrote:
>>>
>>>>
>>>> Though i would still like to see a dual AGP system one day.
>>>
>>>
>>>
>>> Do you mean to say that there are almost no dual monitor AGP graphics
>>> cards?
>>
>>
>>
>> There are many, many, dual monitor AGP cards out there -
>> it has almost become the standard for AGP cards.
>>
>> Hence I took her (his?) remark as intending to convey a
>> desire for a machine with more than one AGP slot, to
>> which the answer is, "Forget it - it ain't going to happen."
>> Especially now with PCI-Express just around the corner.
>
>
> Yeah i clarified that in another post. The original intent was because
> PCI-66mhz is still pretty slow, especially when handling large data.
> PCI-X is really good but your right its not adopted yet. But what would
> prevent a dual AGP slot motherboard? It could be usefull when trying to
> render high-performance simulations that are sectioned. Though i don't
> know if the bus supports it.
>
> From what i know, PCI-X still won't have the same kind of access that
> the AGP card has, like having to go through a PCI bridge, direct access
> to memory with textures etc...

sorry forgot a "not" somewhere. like NOT having to go through a pci bridge.

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Yoyoma_2 schrieb:
>
> Roland Scheidegger wrote:
>
> > Are you sure though the PCI cards have a 128bit memory interface? AFAIK
> > PCI cards always have the name "9200" and not "9200SE" even if in fact
> > sometimes they only have the 64bit interface (you can see it easily, if
> > they have only 4 (tsop) memory chips in total, then they have a 64bit
> > memory interface).
>
> Umm you can make a 128bit interface by just moving it twice rather than
> once. The PCI bus doesn't support 128bit wide datatransfers anyway so
> you just write twice. Another reason why AGP rules

Ahem. Roland was not referring to the system-to-graphics-card interface,
but the card's own memory interface. Apart from this, your first
statement is just plain wrong. Some graphics card basics may not hurt...

Stephan
--
Meine Andere Seite: http://stephan.win31.de/
PC#6: i440BX, 1xP3-500E, 512 MiB, 18+80 GB, R9k AGP 64 MiB, 110W
This is a SCSI-inside, Legacy-plus, TCPA-free computer
Mail to From: not read, see homepg. | Real gelesene Mailadr. s. Homep.

 

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

Stephan Grossklass wrote:

> Yoyoma_2 schrieb:
>
>>Roland Scheidegger wrote:
>>
>>
>>>Are you sure though the PCI cards have a 128bit memory interface? AFAIK
>>>PCI cards always have the name "9200" and not "9200SE" even if in fact
>>>sometimes they only have the 64bit interface (you can see it easily, if
>>>they have only 4 (tsop) memory chips in total, then they have a 64bit
>>>memory interface).
>>
>>Umm you can make a 128bit interface by just moving it twice rather than
>>once. The PCI bus doesn't support 128bit wide datatransfers anyway so
>>you just write twice. Another reason why AGP rules
>
>
> Ahem. Roland was not referring to the system-to-graphics-card interface,
> but the card's own memory interface. Apart from this, your first
> statement is just plain wrong. Some graphics card basics may not hurt...

Nah trust me its not for my lack of graphic card basics, its the fact
that i didn't really read it well. The joys of being bilingual i guess
haha

Keep up the flaming, there's not enough animocity in the world!

 

[Guest]

Archived from groups: alt.comp.hardware.overclocking.amd,alt.comp.periphs.mainboard.asus,alt.comp.periphs.mainboard.gigabyte,alt.comp.periphs.mainboard.msi-microstar (More info?)

Yoyoma_2 wrote:
> Yeah i clarified that in another post. The original intent was
> because PCI-66mhz is still pretty slow, especially when handling
> large data. PCI-X is really good but your right its not adopted yet.
> But what would prevent a dual AGP slot motherboard? It could be
> usefull when trying to render high-performance simulations that are
> sectioned. Though i don't know if the bus supports it.
Don't confuse PCI-X with PCI-Express. PCI-X is just an extension of PCI
and well adopted (in the server market). PCI-X might die together with
PCI and be replaced with PCI-Express though I guess it will take some
time (those markets don't adopt new standards fast).

> From what i know, PCI-X still won't have the same kind of access that
> the AGP card has, like having not to go through a PCI bridge, direct
> access to memory with textures etc...

I don't think PCI-Express will miss any features which might make it
slower than AGP (PCI-X possibly yes, but then again I don't think PCI-X
graphic cards exist). And AGP isn't really all that different from PCI
anyway (AGP1x was basically just PCI-66 (32bit), in fact on the good old
bx chipset the agp port can be used as PCI-66 instead - of course the
board manufacturer has to decide that).