just ran a little test too see how much faster my G3 was over my V5500, well first off i ran my test on the 2001se test, i ran it with just the test the v5500 could run with the same settings including havin em both run in 3d software. well the cards are both overclocked, the v5500 at 195 and the G3 was at 230/460, well anyways here's what happen, and yes the v5500 did out perfporm the G3 in this test, now where in the hell is that G2 of mine  ::)






System Configurations

My System
Compared System


Operating System Microsoft Windows 2000 Microsoft Windows 2000

DirectX Version 8.1 8.1



Mobo Manufacturer

Mobo Model VT8366A-8233 VT8366A-8233

AGP Rates
(Current/Available) N/A / 1x 2x N/A / 1x 2x 4x



CPU AMD Duron(tm) Processor 2071 MHz AMD Duron(tm) Processor 2071 MHz

FSB 166 MHz 166 MHz

Memory 512 MB 512 MB



Graphics Chipset 3dfx Voodoo5 5500 NVIDIA GeForce3

Driver Name Amigamerlin 2.5 SE for Voodoo 4/5 NVIDIA GeForce3

Driver Version 5.00.2195.0232 6.14.10.4403

Video Memory 32 MB 64 MB



Program Version 3DMark2001 SE 3DMark2001 SE

Resolution 800x600 16bit 800x600 16bit

Texture Format Compressed Compressed

FSAA Disabled Disabled

Z-Buffer Depth 16bit 16bit

Frame Buffer Double Double

Rendering Pipeline D3D Software T&L D3D Software T&L



Detailed Test Results
3DMark Score
(3D marks) 3975

3871




Game 1 - Car Chase - Low Detail
(FPS) 55.8

50.7


Game 1 - Car Chase - High Detail
(FPS) 23.6

23.0


Game 2 - Dragothic - Low Detail
(FPS) 57.5

59.0


Game 2 - Dragothic - High Detail
(FPS) 19.3

26.5


Game 3 - Lobby - Low Detail
(FPS) 99.2

98.4


Game 3 - Lobby - High Detail
(FPS) 49.6

40.0


Game 4 - Nature
(FPS) No hardware support

N/A




Fill Rate (Single-Texturing)
(MTexels/s) 576.7

860.3


Fill Rate (Multi-Texturing)
(MTexels/s) 647.3

1690.1




High Polygon Count (1 light)
(MTriangels/s) 6.9

6.7


High Polygon Count (8 lights)
(MTriangels/s) 2.5

3.9




Environment Bump Mapping
(FPS) No hardware support

N/A


DOT3 Bump Mapping
(FPS) No hardware support

N/A




Vertex Shader
(FPS) 25.4

30.6


Pixel Shader
(FPS) No hardware support

N/A


Advanced Pixel Shader
(FPS) No hardware support

N/A




Point Sprite Speed
(MSprites/s) 1.6

5.5

and yes the v5500 did out perfporm the G3 in this test

???

Well, the only time I tested a GeForce3 Ti200 (no OC, tested only for selling it), it scored 8279pts in 3D Mark 2001SE Build 330, 1024x768 32bit (AthlonXP 2100+) .

I don't think a Voodoo5 could ever touch a GF3 in 3D Mark, so something must have gone terribly wrong with your test (?)
--------------------
Look at your CPU limited fps, this seems to be the error.
The Low Detail fps are exceptionally low. Try benching with a more powerful CPU.

No not really FalconFly, to make it a fair test i ran both cards on just the tests that the v5500 could run plus i had the g3 cad run with 3d software t&l instead of hardware t&l just like the v5500, so since the g3 didnt get 4 extra test to score on and had to run the same 3d software t&l setting it scored lower than the v5500. most ppl just test straight up the way the card comes, i wanted to no if under the same conditions if the g3 could really beat my v5500 and all being the same it could not. now with all the g3 settings at default it ran at almost 10,000 on that test, so the card is faster in 2001 standards but in a equal test it was slower, just goes to show that t&l saved nvidia's ass because the g2 is even slower :P

hope that clears it up ;)

Well, I'd say you've proven that the Detonator Driver you used eats a few more CPU cycles when compared unter non-T&L conditions.

I wounder though why anyone would disable T&L if present and useable...

And as said, in order to use the Fillrate of a GeForce3, you better up the CPU power alot, even more so under Software T&L conditions... ;)

PS.
I wouldn't call that a 'fair' comparison.
This is like disabling Hyper-Z on an ATI example, because a tested competitor would not possess a bandwidth-saving feature ::)
Either that, or you'd have to change the Topic (Cards tested under non-standard conditions using disabled features)

A test under 1600x1200 32bit would be more useful to compare the Cards apart from CPU limitations.

Ah, i see what you mean falconfly, i just meant tested at same settings hwen possible, sry my fault :-[,  and its is a 2500 barton that is o/c to 2800 barton, not sure why it says duron

First of all, testing a video card in 3DMark 2001 with Software TnL proves the real power of the V5 against the G3. This test is of true value, since in games TnL doesn't count as much as fillrate do. If a card is enough fast, it can outperform even a G4, although there are special cases under which this might happen.

But!!! It is unfair to do as most of the tester (from magazines/home) do, to test a video card using Tnl and under no. Simply test the cards with same settings to have a true score.

A test run with a ProSavageDDR, under similar circumstances proved the unfair chance of the Hardware Tnl, against normal Tnl. On Mafia, the test was done by starting a new game at 800x600, 32 bit. On normal Tnl, the video card would perform slow, but when running with Hardware TnL, the framerate was near 60fps!!!, and in the game menu IT MAKES A DIFFERENCE!!!

So, if V5 would lie that it has and uses TnL... a G3 total killer in action... And if there are any glitches, no problem they can be avoided by the dricer even if the V5 has no Hardware TnL...

Make a good card lie that has a TnL, and you get a very fast one!!!