View Full Version : See the differance AGPx1 vs AGPx2

29th September 2001, 12:21


Thats few benches that i have done on my matrox. As you can see AGPx2 wins hands down.

The card was overclocked 150/200

And all the tests were done in bilinear filtering.

Oh no why the pictures arnt showing up :o

29th September 2001, 12:31
Seeing that we can't view your results, it wouldn't matter. The only place where AGP1x vs AGP2x (or 4x for that matter) really how a difference is in a few benchmarks. Try your tests on real world games and apps and your claims fall though the floor.

29th September 2001, 12:37
I did my tests in Quake 3 i think thats real enough.

If anyone can put those pics up i can send you them by e-mail

29th September 2001, 12:43
You can send them to me and I'll put them up

29th September 2001, 13:40
Kastuvas's pics...


29th September 2001, 13:46
Tho it appears you have a typo in Test 10's resolution.

What OS and drivers are you using?

29th September 2001, 13:58
You are also using a 16meg G400 which is far more dependent on AGP transfers than a 32meg card. >95% have 32 meg cards.

29th September 2001, 14:06
I am using Windows 98 SE.
The drivers i was using were 6.21 with Turbo GL without Turbo GL i cant enable AGPx2 well i can but the games freeze after 10 seconds.

Also Turbo GL makes funny textures in Quake 3 above 1024x768x32 and evrything max but its fine at 1024x768x16 and everything max

29th September 2001, 14:14
On a 32 Mb G400 you get a diference less than 1-2 fps on Quake3, even in the Quaver demo. Been there done, that.

The *only* test where you could see a noticeable difference was on Dagooth Moor Zoological Gardens that were like 5 fps apart (maybe due to the big geometry throughtput into the AGP bus) - but you really donīt play that too often, do you :)

29th September 2001, 14:21
Also TurboGL isn't intended for use with PD6 up.

14th October 2001, 17:47
I recently compared a G400 MAX 2X AGP card forced to run 1X (otherwise nothing in 3D will run), versus a 4X AGP card at 4X AGP.

I got about the same frame rates in the QuakeIII 1.27 demo (42-44 fps thereabouts). I run 1024x768x16, with textures set to 75%, and bilinear filtering, on an 800 Thunderbird. I could not tell the difference in any of the other 3D games either. The 4X AGP didn't improve the graphics corruption in parts of Alice and the Return to Castle Wolfenstein demo I experience.

I forget exactly, but there was negligible difference in 3DMark2000/2001 also.

So if you guys have a 1X AGP (forced) card, I wouldn't fret. And I imagine a 1X AGP card that overclocks better than a 4X AGP card will perform the better of the two.

15th October 2001, 12:22
The last time I checked there was quite some improvement between AGP1x and AGP4x in 3DMark2000. Even the change from AGP2x to AGP4x, while not making that much of a difference, still gave consistently higher results.

And this was on a VIA chipset that some say can't do AGP4x at all :rolleyes:


EDIT: P.S., I don't believe it makes much of a change in actual games, though.

15th October 2001, 15:06
"EDIT: P.S., I don't believe it makes much of a change in actual games, though."

Yep, case in point:

Geforce256 DDR: 5500 3DMark2000 score
QuakeIII 1.27 demo: 41 fps

Voodoo5 PCI: 4400 3DMark2000 score
QuakeIII 1.27 demo: 46 fps

These settings are 1024x768x16, with all settings to HIGH.

Also in Startopia, the Voodoo5 is smooth on the menu screen while the Geforce is jerky.