PDA

View Full Version : Further investigation of Tech-Reports allegations with NOLF


Pages : [1] 2 3

gstanford
06-07-03, 06:10 AM
First off, a link to the Tech-Report article that set all this off for the pedantic amongst us (you know who you are...)
http://tech-report.com/etc/2003q2/3dmurk03/index.x?pg=1

Note the zipped savegame file needs to be renamed to .ace (it's really an ace archive, but this forum won't upload ace files).

This is a continuation of the 3DMark2003 saga brought up by tech-report and peoples concerns that nVidia is cheating in games.

Those of you with good memories will remember reverend (anthony tan) of beyond3d used to have his own little website where he would review and comment upon things.

he used NOLF1 (the first game in the No-one Lives Forever) series to show the effects of anisotropic filtering in on of his articles.

I have reproduced his test as closely as is possible here, with the following twist.

The game is tested normally, then tested again as "3DMark03.exe".

I ran 3 tests:

1) no anisostropy applied whatsoever
2) 8x anisotropy as NOLF.exe
3) 8x anisotropy as 3DMark03.exe

Unfortunately I'm only going to post the images as jpegs (95% quality in irfanview) since the .tiff and .png versions are massive.
If anyone wants to host the tiffs (2.30mb apiece) tell me after all attachments are loaded (I'm only 56k - this could take a while)...

NOLF is a DirectX game, just as 3dMark2003 is a DirectX benchmark.

attachment for this post is the savegame used to set the screenshots up (slot2.zip) - (remember - actually an ace archive).

gstanford
06-07-03, 06:11 AM
This post contains game setting screen 1

gstanford
06-07-03, 06:13 AM
This post contains game settings screen 2

by way of explanation environment mapping and model chrome affect the lighting in parts of the image - we are only interested in the anisotropic differences so they are disabled.

gstanford
06-07-03, 06:14 AM
This post contains game settings screen 3

gstanford
06-07-03, 06:15 AM
This post contains screen 1 of the detonator settings

gstanford
06-07-03, 06:16 AM
This post contains image 2 of the detonator settings

gstanford
06-07-03, 06:27 AM
Moderators/admin!!!

Why won't your damned bulletin board work as advertised???

I am trying to upload a 370K jpeg image and the forum tells me I have exceed the file size on 102400 bytes...

gstanford
06-07-03, 06:37 AM
This is the test image from NOLF.exe with no anisotropic filtering applied (for reference purposes) Lower quality 40% due to forum upload issues.

gstanford
06-07-03, 06:40 AM
This is the test image from NOLF.exe with 8x anisotropic filtering applied (for reference purposes) Lower quality 40% due to forum upload issues.

gstanford
06-07-03, 06:42 AM
This is the test image from 3DMark03.exe with 8x anisotropic filtering applied (for reference purposes) Lower quality 60% due to forum upload issues.

uttar could you delete your first post. Thanks for the reply. A brain fart on my part.

gstanford
06-07-03, 06:44 AM
Okay as everybody should now (hopefully - the jpeg compression may have messed things up, but if someone will host files I can supply tiffs) be able to see there is absolutely no difference between NOLF.exe and 3DMark03.exe.

Discuss.

Uttar
06-07-03, 06:48 AM
Hmm, it certainly seems to me there are no IQ differences. Is there any performance difference, however? Because if there isn't, nVidia might be testing for more than just the filename. Or it might be application-specific.


Uttar

gstanford
06-07-03, 06:55 AM
no performance difference that I can pick up (and NOLF is one of my favorite games - I have a very good feel for how it performs).

extreme_dB
06-07-03, 07:09 AM
I don't understand why it was even necessary to post all those pics. Anyway, is there a link to the original article or discussion?

On a side note, I played NOLF for a bit and it was one of my worst experiences in FPS gameplay. In one level at the beginning where you have to snipe bad guys from a window, they just pop out of places continually like it's Duck Hunt or something hehe. Was NOLF trying to be a parody of bad gameplay as well?

The textures on some levels gave me a headache.

Hyp-X
06-07-03, 07:13 AM
Originally posted by Uttar
nVidia might be testing for more than just the filename. Or it might be application-specific.


It could test if the app uses the DX9 API.
NOLF1 is either a DX7 or DX8 game.

DaveBaumann
06-07-03, 07:17 AM
The 3DMark AF changes occur when its selected by the application, not through the driver control panel.

gstanford
06-07-03, 07:19 AM
I could post a derogatory remark about extreme_dB at this point, but, frankly at the moment he does not need my help in looking like a clown.

NOLF is a directX 7 game.

I could repeat the tests with morrowind (if the rename would work - Ultima 9 would not). That is directX 8. Fairly obviously I have no game directX 9 graphics wise to test with.

extreme_dB
06-07-03, 07:19 AM
Wait a minute, I was under the impression that someone had claimed to have found an anomaly in NOLF using the renaming trick, and gstandford was posting pics to refute it.

Or is this just a test of a game? In which case this thread is pointless IMO.

extreme_dB
06-07-03, 07:25 AM
I don't understand what's going on here or why you would want to make a derogatory remark about me? :confused:

Any app detection for AF optimization would only be done for a benchmark (synthetic/game) I would think.

I'm obviously missing something.

gstanford
06-07-03, 07:33 AM
run back to rage3d, troll.

You are in a nVidia specific forum on a nVidia specific website at the moment. If you have nothing constructive to add to my thread, kindly delete your posts in it, leave and don't return.

digitalwanderer
06-07-03, 07:44 AM
Originally posted by gstanford
run back to rage3d, troll.

You are in a nVidia specific forum on a nVidia specific website at the moment. If you have nothing constructive to add to my thread, kindly delete your posts in it, leave and don't return.
Uhm...you're the guy who made a thread loaded with picts and mostly your own posts when you could have just posted on post saying, "I tried running NOLF & running it after renaming the .exe and didn't see/notice any differences."

No offense, but this IS a bit of overkill to make a point that no one is arguing with you...why don't you go back to [t]ardOCP where you apparently belong. ;)

digitalwanderer
06-07-03, 07:46 AM
Originally posted by extreme_dB
I don't understand what's going on here or why you would want to make a derogatory remark about me? :confused:

Any app detection for AF optimization would only be done for a benchmark (synthetic/game) I would think.

I'm obviously missing something.
If it helps, either you ain't the one missing it in this thread or I'm missing it too. ;)

There was some discussion yesterday about this....the drivers can identify the program by more than just the name of the .exe. :)

Sorry you wasted all your time, but the pictures are pretty. ;)

ChrisRay
06-07-03, 07:47 AM
I'm trying to see what I am looking at here is all

gstanford
06-07-03, 07:48 AM
So says another dyed in the wool fanatic. The advice I gave Extreme_dB goes equally for you, digitalwanderer.

Some of the fanATIcs getting around here would do well do read the sticky in this forum "No ATI discussions in here!"

Now, unless you have something of value to contribute to the thread one way or the other, please leave.

EDIT: I have never been a hardocp member, and only visited their forums once ages ago to smack down the allegations of another fanATIc on another forum, so thats how much you know...l

aapo
06-07-03, 07:57 AM
Originally posted by DaveBaumann
The 3DMark AF changes occur when its selected by the application, not through the driver control panel.

This is a very good point. Gstanford, you should test a game that can internally set AF on/off. Otherwise your test scenario idea was worth the try, IMHO.