nVidia release new all singing all dancing dets.

[alexsok]

LOL, that is what I thought when I first saw the thread.. Speacial care for Nvidia driver release I guess.

Hehe

I think both John and Dave agree that once a signifcant driver release occurs, it's worth making a thread about it!

 

[maskrider]

Hmmmm interesting...

You could try the NOLF 2 demo..49 megs

ftp://3dgamers.in-span.net/pub/3dgamers1/games/noonelivesfrvr2/nolf2ohioteaser.exe

But NOLF 2 demo is not a benchmark, I have also got noticeable increase in speed with the feedeemark 2001SE on my ELSA 925VIVO (Ti4600), nature is now average at 74.8fps.

Overall is now 9805 feedeemarks with my P4 1.9 (not Northwood) with 512M PC800 RDRAM.

I have forgotten the old numbers.

The followings are for those people who want to know the detail numbers

Car chase low - 127fps
Car chase high - 39fps
Dragothic low - 190.9fps
Dragothic high - 101.5fps
Lobby low - 123fps
Lobby high - 54.6fps
Nature - 74.8fps
Fill-rate single - 1070.4MTexels/s
Fill-rate multi - 2325.3MTexels/s
High poly 1 light - 51.8MTriangles/s
High poly 8 light - 12.5MTriangles/s
EMBM - 133.2fps
DOT3 - 152.6fps
VS - 99.7fps
PS - 122.8fps
APS - 96.1fps
Point Sprites - 30.2MSpries/s

 

[Doomtrooper]

No offense guys with the Madonion scores...

1)It doesn't reflect real world gaming
2)Everyone optimizes (this is proabably #1 on optimizations)

I'm not trying to rain on anyones parade here, but I would be more concerned with faster Anisotropic and say Serious Sam 2 gained 15 fps vs a few more 3Dmarks

At least the control panel looks alot better..

 

[Galilee]

Er, does this mean drivers threads are now okay in this forum?

LOL, that is what I thought when I first saw the thread.. Speacial care for Nvidia driver release I guess.

Didnt people talk about the new Catalyst when they came?
This driverupdate is not another beta, it's something "new". Like the fact that aniso is now in the control-panel. Finally putting and end to that argument.

 

[king_iron_fist]

Mine don't seem to work properly - When I run control panel I get a rundll error. Nature test seems up by 20fps which seems to account for the score increases everyone is on about. apart from that and a load of artefacts and tearing I don't see any real difference.

OT: We seem to have chosen a very poor day to release our detonator article over at www.boogletech.com

 

[Joe DeFuria]

Er, does this mean drivers threads are now okay in this forum?

Heh, I'm generally as against "driver threads" as the next guy, but this is a special case. This is a "major" revision (similar to Catalyst release), and the vendor is officially making performance increase claims.

So, while I do detest the useless "These Drivers Rock" comments, (complete with comparing average FPS scores from previous drivers to peak FPS scores with these...) I am interested to see in what way (if any) the vendor's claims can be validated, and if there is any impact on image quality (positive or negative) with these drivers.

So far, performance benefits seem to be limited to 3D Mark nature score increase, but I'm waiting for some more varied tests from several sources.

 

[Reverend]

Anyway, I'll be running some quick comparisons between the 30.82 vs 40.41 in the following games (once I finish downloading the dang drivers in about half an hour's time!) :

GLQuake - Tenebrae mod
Tiger Woods 2002
F1 2002
UT2003 ( )

...mostly (save for GLQuake) using FRAPS in my own innovative ways

Let's hope this thread doesn't get locked

 

[Joe DeFuria]

Looking forward to the comparison Rev. Please include image quality impressions and screenies if you can.

 

[Neeyik]

Doomtrooper - can you just stop the 3DMark/anyone who posts 3DMark results digs, jibs and comments? There hardly goes a thread without you passing some remark about how useless it is...

Quite frankly, the individual "game" tests in 3DMark are just as valid as any other 3D benchmark. Does a timedemo in Q3A tell you anything about how many frames per second you'll get while playing online? The only thing benchmarks are good for is comparisons - run a stack of them, change a variable, run them again and see what has changed. And for that, all you need is an application that you can run over and over again...doesn't matter whether it's 3DMark, Serious Sam or Jet Set Willy.

 

[alexsok]

Mine don't seem to work properly - When I run control panel I get a rundll error. Nature test seems up by 20fps which seems to account for the score increases everyone is on about. apart from that and a load of artefacts and tearing I don't see any real difference.

Same thing happened to me.

Completely remove the old drivers and install these ones.

 

[Reverend]

No offense guys with the Madonion scores...

1)It doesn't reflect real world gaming
2)Everyone optimizes (this is proabably #1 on optimizations)

I'm not trying to rain on anyones parade here, but I would be more concerned with faster Anisotropic and say Serious Sam 2 gained 15 fps vs a few more 3Dmarks

At least the control panel looks alot better..

I think the folks here would be interested in 3DMark's individual scores versus "3DMarks". That's how I have always viewed 3DMark's usefulness anyway.

 

[Reverend]

But NOLF 2 demo is not a benchmark, <snip>

Monolith Productions is working on including some sort of benchmark in the retail game... may or may not get included in the retail game.

 

[NV25]

I gained 310 points with these drivers. The most significant increase was Nature going from 36 to 46 fps, which probably accounts for most of that.

My new score : 8112

Athlon XP 2200+
GeForce3
512mb PC2100 Ram

It might not mean much to me though because I have a 9700 on pre-order

Nice control panel, btw. And it's about time that d3d aniso was in there. I was sick of using Rivatuner!

 

[Reverend]

Looking forward to the comparison Rev. Please include image quality impressions and screenies if you can.

It's midnight over here. I'll have time to run very quick benchmarks and post my results but would not have time to investigate IQ differences (if there are any, good or bad) tonight. I'll be sure to provide any significant IQ differences tomorrow.

 

[maskrider]

Now the 8xAF numbers AF (no AF)

feedeemarks 7534 (9805)
Car chase low - 117.2fps (127fps)
Car chase high - 37.2fps (39fps)
Dragothic low - 119.4fps (190.9fps)
Dragothic high - 74.6fps (101.5fps)
Lobby low - 97.7fps (123fps)
Lobby high - 48.6fps (54.6fps)
Nature - 49.1fps (74.8fps)
Fill-rate single - 611.2MTexels/s (1070.4MTexels/s)
Fill-rate multi - 611.2MTexels/s (2325.3MTexels/s)
High poly 1 light - 49.1MTriangles/s (51.8MTriangles/s)
High poly 8 light - 12.3MTriangles/s (12.5MTriangles/s)
EMBM - 74.2fps (133.2fps)
DOT3 - 67.1fps (152.6fps)
VS - 80.3fps (99.7fps)
PS - 69.8fps (122.8fps)
APS - 44.5fps (96.1fps)
Point Sprites - 30.3MSpries/s (30.2MSpries/s)

 

[Galilee]

Anyway, I'll be running some quick comparisons between the 30.82 vs 40.41 in the following games (once I finish downloading the dang drivers in about half an hour's time!) :

GLQuake - Tenebrae mod
Tiger Woods 2002
F1 2002
UT2003 ( )

...mostly (save for GLQuake) using FRAPS in my own innovative ways

Let's hope this thread doesn't get locked

If NVIDIA is slow you can try this link:
http://www.stud.ntnu.no/~vidaralm/40.41_winxp.exe

 

[maskrider]

But NOLF 2 demo is not a benchmark, <snip>

Monolith Productions is working on including some sort of benchmark in the retail game... may or may not get included in the retail game.

Indeed, I am not willing to see any sort of benchmark utilities included in NOLF 2, I am merely replying to DT on the NOLF 2 demo part. I would like to see that NOLF 2 is being treated as solely a game than a benchmark tool.

I am very eagerly waiting for the arrival of NOLF 2 and for sure my pre-ordered R9700 PRO, ha ha !

 

[alexsok]

Sys:
P4 2ghz
512mb RAM
GF4TI4600

The highest score is bolded, ANISO perfomance to follow in the next post! (although I don't have 30.82 Aniso 8x results, sorry).

Score: 30.82: 10346 - 40.41: 11400

Game 1 LOW: 30.82: 165.6 - 40.41: 162.4
Game 1 HIGH: 30.82: 49.8 - 40.41: 49.3

Game 2 LOW: 30.82: 189.0 - 40.41: 215.8
Game 2 HIGH: 30.82: 109.1 - 40.41: 116.8

Game 3 LOW: 30.82: 142.9 - 40.41: 146.1
Game 3 HIGH: 30.82: 65.2 - 40.41: 65.2

Game 4 Nature: 30.82: 44.6 - 40.41: 76.6 (hovering around 100fps sometimes)

Fillrate (Single-Texturing): 30.82: 1067.0 MTexels/s - 40.41: 1063.7MTexels/s

Fillrate (Multi-Texturing): 30.82: 2320.3 MTexels/s - 40.41: 2326.6 Mtexels/s

High Polygon Count (1 Light): 30.82 50.7 MTriangles/s - 40.41: 55.0 MTriangles/s

High Polygon Count (8 Lights): 30.82: 12.6 MTriangles/s - 40.41: 12.6 MTriangles/s

EMBM: 30.82: 138.1 - 40.41: 152.6

DOT3: 30.82: 152.3 - 40.41: 163.3

VS: 30.82: 101.9 - 40.41: 98.8

PS: 30.82: 121.8 - 40.41: 123.5

ADV PS: 30.82: 88.8 - 40.41: 98.8

Point Sprites: 30.82: 30.5 MSprites/s - 40.41: 33.0 MSprites/s

 

[Snap]

So - Vertex Shader test drops in performance, pixel shader test shows a small rise but the nature test shows a massive rise...

 

[Oompa Loompa]

So the enhancement has nothing to do with fill rate or shading.

Geometry compression?