View Full Version : Detonator 29.42 vs. 30.82 driver performance

08-09-02, 02:30 AM
Please view the attachment. I compiled it for those looking for more than a subjective opinion on how the new dets perform.

08-09-02, 04:16 AM
Nice work there :) Very consistent with my experience as well. Too bad Aniso. + FSAA performance hasn't improved one bit :( But then again live streaming now does work, which is nice :p

08-09-02, 09:23 AM
nice work! I'll tell the front newspage about it! ;)

08-09-02, 11:20 AM
Ha ha all that work for the most unexciting news ever. Even with the drivers going through a major revision number change the performance is almost identical for DX.

08-09-02, 01:52 PM
Will we get more new drivers from nvidia that does speed up the GF4 ti 4400 speed pretty much or a lot, in no FSAA or AA and with FSAA and AA on maximum ?


08-09-02, 02:17 PM
I would like to see that too! I had really high hopes last night when I started benchmarking but it would appear that performance wise nothing has changed.
It was nice of NVIDIA to provide some hooks in the driver to allow the tweakers of the world to play though.
Thanks Matthyahuw for getting me on the front page, I also saw a link on [H]. One evenings bordom turns into 15 minutes fame :)

08-09-02, 02:19 PM
lunix do you think nvidia will release a lot faster drivers soon ? Like right when the ATi 9700 comes out ?

08-09-02, 02:25 PM
intim83, I have no friggin idea. What with the architecture being around so long I wouldnt hold my breath for any more large improvements. Not that they wouldnt be welcomed. I have to remind myself that a year ago if someone would have told me that I would be getting over 10k 3dmarks I would have laughed in their face.

08-09-02, 02:27 PM
Well, I think what most of us really want is better FSAA + Anisotropic performance, isn't it? I've got a kick-a** 3D performance whitour either of them (my 3DMark2001se score at default is 14.500+), but what I'd really like is better performance at 1280z1024x32 4xFSAA and 4x Ani. Here I get around 3100 3dmarks for comparisons sake. Right now performance is actually sufficient at 1280x1024x32 with 4xFSAA for most games (I do sims mostly, like FS2002, GP4) but 4x Ani looks SO much better, but it does push even the almighty Ti4600, overclocked too, a little over the edge. That wouldn't happen I suppose, if the fillrate wasn't cut in half when Ani. filtering is enabled. I will continue to hope, but it may look like an upgrade instead :(

08-09-02, 02:27 PM
lunix true and right now people will laugh about 20000 to 30000 3dmark2001SE scores. But we will have those kind of scores in 3dmark2001SE at least 1 to 3 years from now.

08-10-02, 07:43 AM
On the news page.

Keep in mind that the changes made to the driver set did not hurt performance. If you recalll back to the TNT days, a new driver set was released that really hurt the TNT's performance. This lead to a lot of driver mixing to get the right balance of speed and stability. No such concern nowadays...

In the weeks before the new product launches, there are several driver revs designed to get the new products working with out taking down the legacy chips. Expect more lateral gains.

08-10-02, 08:06 AM
A big aniso perfomance increase can be achieved using the latest release of Riva Tuner! :D

08-10-02, 10:47 AM
I wouldn't call the performance using RT, BIG.
Marginal at the most, I know how to use RT so it's not me.

I would be more interested in hearing users thoughts on 2D.
And I would like the Vibrancy setting to stop reverting to default when you bring up a DOS box, or related.

2D in a lot of apps that use a window with graphics in, is still not right.
If any of you have used an emulator that uses a graphical front-end in a window, you might have seen this.
The AtariST emu SainT uses this technique, and on NVidia cards it screws up the display.
There are couple of apps that use it, but their names slip my mind.

I also get a lot of graphical corruption on the desktop if I copy or move files from one place to another.
An example is if you drag a winzip archive, with the right mouse button, to a different location and click unzip to here. It always leaves a corrupt image of that menu and needs a few refreshes(F5) to get rid of it.
I'm a PC engineer, and I have tried all the driver sets from 23.11 to 30.82. And using Norton ghost to test out fresh installs of WinME, Win2k, and WinXP, they all do the same. Yet I've seen no mention of it here.
Is it just me ?


08-10-02, 12:45 PM
You probably already tried this, but just haven't mentioned it but it would be almost criminal for someone not to bring up that the graphical corruption could be caused by your 300mhz OC on your proc. As silly as it sounds, knowing that if you're savy enough to GET a 300mhz OC, then you're obviously savy enough to back off the OC to see if the corruption goes away. Mostly I'm just thinking about it out loud to give you the chance to correct me if my thinking is astray... :)

08-10-02, 03:05 PM
moshpit ,
Yeah, thanks. It's been like this ever since my Geforce 3 and the 23.11 drivers.
It's definately the drivers as I have tested the 2D problem on a few mates PCs.
Perhaps it's just a combination of certain hardware and OS that makes this just unlucky ?
But the thing is, take out the Ti4600 and slap in an 8500 and it goes away. Even a Voodoo5500 I have doesn't produce the problem.

I don't suppose anybody would be willing to try it out for me here ?

Download the new Geforce tweaker called aTuner

www.3dcenter.org - I think.

And run in, preferably in WinME, but 2k and XP is ok.
Now do you see the whole window ?
I can only see a part of the window, and a lot of the check boxes are cut off.

08-10-02, 07:55 PM
i just downloaded aTuner, i see all the checkboxes and the whole window seems to be there. i use winME and a gf3 w/ 29.42 drivers

08-11-02, 03:49 AM
Thanks for doing that.

Hmmm, I'm now puzzled.
3 fresh OS's, all doing the same thing with aTuner.
It only happens on fixed windows, i.e. Windows you can't resize.