GeForce 3 slower than GeForce2 MX!


I’m sure I’m hitting a software path on the GF3 when I’m running through a hardware path on the GF2 MX. The GF3 perfomance is pants - the GF2 MX performance is acceptable.

The difference occurs when I render some things with glPolygonOffset enabled, texturing enabled and alpha test enabled. Texture formats are GL_RGBA ( of course) - not sure what other information I can give.

Need more details.

What drivers you running on each card? What PC setup do you have? What are you exactly doing in your scene? What extensions are you using etc etc…

Is that a high poly scene?
I’m asking because of the invariance issues between vertex shaders/fixed function. GF3 might implement poly offset with vertex shaders, but I’m going out on a limb here, admittedly. AFAIK these issues should already be resolved in later drivers but that may come at a performance cost. Again AFAIK there should be a glHint(something) introduced for that.

GF 2MX has
GF 3 Ti200 has (???)

This is when I do right click on desktop and look at driver details. The GF2 is running on a Pentium III 733mhz - the GFIII is running on a Celeron 900mhz.

Isn’t a celeron based on a P2 or something… The Lower clocked P3 is probably the much more faster cpu, and prolly has SSE, unlike the celeron, which the video card drivers are prolly using.

Other than that I dont know.


The newer Celerons have SSE as well. Looking at some D3D benchmarks (3DMark2001) of comparable (to what was stated above) systems indicate the GF3 system as scoring nearly twice as high.

[This message has been edited by DFrey (edited 04-10-2002).]

Yea - I’m sure the Celeron is a ****E cpu, but I figure the GeForce 3 would compensate for that!

I’m sure I’m hitting some kind of software path with my particular code section. I’m not sure why this would be the case - anyway, we can live with it for now - pending further investigation of course (deadline tommorrow - no time to start on version 2 already

Most cellys from 800 MHz and up actually use a 100 MHz bus and support SSE, so they’re very similar to many 800 Pentium-IIIs (though slower than the CuMine “800EB” versions with 133 bus). It’s quite likely not your problem. More likely chipset driver issues, memory speed grades, or something like that.