Matrox confirms Parhelia is AGP 4X

[High_Jumbllama]

*High_Jumbllama thinks about pulling out his quasitronic automodulating, multifrequency, pulsed, phase amplified, quantum sighted, phaser rifle with attached titanium alloy blade launcher next time somone posts from the Inquirer as if it is actual news.

[rugger]

Originally posted by High_Jumbllama
*High_Jumbllama thinks about pulling out his quasitronic automodulating, multifrequency, pulsed, phase amplified, quantum sighted, phaser rifle with attached titanium alloy blade launcher next time somone posts from the Inquirer as if it is actual news.

You would be better served by one of those heavy a** IBM Model-M keyboards

[Dr Mordrid]

With all the editing and graphics stuff I've used it was nigh on impossible to tell if the card was running 2x or 4x.

Dr. Mordrid

[Jon P. Inghram]

I don't know what you guys are talking about... when set my new motherboard to AGP 4x I noticed quite an improvement over 2x!


There ya go, see how easy that was? Watch, I can do it again!


The Parhelia looks like it might be good, until you check the block diagram. Unfortunatly there is no UberZ or other z-buffer enhancements shown, so while I'm sure it'll have decent 2D graphics it'll still not hold a candle to the GeForce 6 TI 2^34.


Heck, why stop while I'm on a roll!


Matrox has always been business oriented, but with this new card it seems they're going to attempt to reenter the gaming market. We at Idiotechnica belive that this is a mistake on Matrox's part, mainly because they've shown that while they make decent business cards, they simply cannot compete with the gaming performance and driver quality of nVidia.


Might as well try one more...


Some suspect that Matrox is going to try for the professional 3D rendering market with their new Parhelia-512 chip, but we suspect they'll miss the mark due to their poor history with OpenGL support. (Oops, that actually accurate... doh )

[The PIT]

Anyone care to benchmarks with a few of todays games at agp x1 x2 x4 and with some prof graphics programs too see if theres any noticable differance in rendering times or fps?

[Mikko]

How many of you people already have AGP-8x motherboard??

I don't, but maybe I'll just buy a new Parhelia-xxx after couple of years, if I'll have AGP-8x motherboard at that time.

Until then I will certainly be very happy with the Parhelia-512 AGP-4x version!

[Jon P. Inghram]

Not exactly what I would call a "professional" program, but here's the difference between AGP 1x and 4x on my system with 3DMark2K2 SE:

AGP 1x: 1507,http://service.madonion.com/compare?2k1=3541745

AGP 4x: 1543,http://service.madonion.com/compare?2k1=3524801

[edit]

Downloading SPECviewperf 7.0, that ought to be a more accurate test of "professional" usage. It's 288 megs though... it'll be a while at 95 kB/s

[/edit]

[Indiana]

I did something like this on my RadeonVIVO with 3DMark2000 some time ago - the difference between 1x and 2x is small and the one between 2x and 4x is minimal...


P.S.: Those MadOnion "benchmarks" are under the few programs where you see the biggest difference between AGP modes because of them using excessively high geometric and texture data (considering todays standards).

[TdB]

i don´t think any current matrox cards are agp limited, perhaps some of the faster gf4ti benches would be more relevant, as they are closer to a parhelia in that regard.

[Jammrock]

Is it just me, or does this whole argument seem rather rhetorical?

AGP is a high-performance interconnect between the core-logic chipset and the graphics controller for enhanced graphics performance for 3D applications. AGP relieves the graphics bottleneck by adding a dedicated high-speed interface directly between the chipset and the graphics controller as shown below.
...http://developer.intel.com/technolog...al/chapt_1.htm ...
This figure is a representation of how textures would be stored to system memory and then delivered directly to the frame buffer. This removes bandwidth-intensive 3D and video traffic from the constraints of the PCI bus and allows the ability to use more textures than can typically fit into local memory for high detailed images and texture maps. This also reduces the local memory requirements on the graphics card.

Translation: AGP 8x will only greatly improve video card performance for those cards that store their textures in SYSTEM MEMORY. I.E. Integrated graphics solutions and older cards with small amounts of onboard RAM. Graphics cards such as the Matrox Parhelia, GeForce 4 Ti4600, ATi Readon 8500, etc, which store the textures in 64-128 MB of special high speed RAM will not see great performance increases.

Jammrock

[Maggi]

hmmm ... considering I almost daily generate previews of +1GB in system RAM, me thinks I could benefit from AGP8x.

Is AGP support in WinXP as crippled as it is in Win2k or did they write a new GART ?

Cheers,
Maggi

[DuRaNgO]

Of course you'll see a difference between 1x and 4x, now see the difference between 2x and 4x/8x will be very small

[Jammrock]

Maggi, okay, in your case you would notice a difference if you have enough RAM that you are not running into virtual memory. For a gamer though, not likely to be a significant difference, unless you have an integrated solution.

JAmmrock

[Jazzz]

i remember seeing a site, a while back, comparing AGP 4x 2x and 1x for 3dmark 2001 and the Ti4600, the only part which saw a decrease was the Nature demo, the others were all roughly the same. No need for AGP 8x really 4x is just fine, when there are more mobos which support AGP 8x then matrox may support it them but there is no real need

[Maggi]

Originally posted by Jammrock
Maggi, okay, in your case you would notice a difference if you have enough RAM that you are not running into virtual memory. For a gamer though, not likely to be a significant difference, unless you have an integrated solution.

Jammrock

that's what I figured, but does anybody know if AGP support in XP is not restricted as it is in Win2k ?

Thanx,
Maggi