The truth isn't very useful for measuring NVIDIA cards-BJORN

[see colon]

Mind you, I fail to see wha the popularity of a game has to do with its use a benchmarking tool. Does a game being unpopular mean its shaders are less PS 2.0? Does popularity directly corrolate with the quality of the shaders or number?

ok, one more time. tomb raider: angel of darkness is a fantastic benchmark for ps2.0, but if noone is playing the game, it's more of a synthetic benchmark than anything else. compare this game to quake3 (at the time of quake3's release). since quake was a top selling game, and a top licensed engine, everyone who was purchasing a video card for gaming would look at quake3 benchmarks because it demonstrated how it and to some extent other games based on it's technology would run. ut2003 had a similar life as a benchmark. i have a hard time believing anyone based their hardware purchase around how fast they could run TR:AOD, but some may have based a purchase on the fact that TR:AOD is shader intensive and card x performs bettin it it than card y. the same information could be gathered from any number of the synthetic benchmarks that are out now (3dmark, shadermark, aquamark, x2 rolling demo).
c:

ps- i'm in no way saying that we should only bench "popular" games. more information is always better.

 

[jb]

since quake was a top selling game, and a top licensed engine, everyone who was purchasing a video card for gaming would look at quake3 benchmarks because it demonstrated how it and to some extent other games based on it's technology would run.

To me I can hear were you are coming from. However the above point we have found it is usally not the case. Meaning perfromance in Q3 != Profromence in RtCW != JK2 != Alice != SOF, ect but thats a minor point..so I will shudda up now

 

[see colon]

Meaning perfromance in Q3 != Profromence in RtCW != JK2 != Alice != SOF, ect

that's why i said to some extent. all of those games add new features to the engine, and some of those (i believe) actualy run on the team arena engine. but still, the performance delta is somewhat predictable in quake3/ta engine games.
c:

 

[Quitch]

Synthetic or game, what's the difference? Most benchmarks within games use canned, un-game like situations. A PS 2.0 shader is a PS 2.0 shader, and whether or not it's a game, or a good game, or a much played game, is neither here-nor-there.

 

[jjayb]

So basically anything that shows Nvidia in a bad light should not be used. Whether it's a game, a game benchmark, or a synthetic benchmark. First we had the "can't use 3dmark, it's not a game" argument. Now we are getting "can't use tombraider, even though it is a game it's not very popular."

Funny nobody argued the same about aquamark. Exactly the opposite in fact. It was praised as a benchmark because it was based off of an actual game. It was one of the tools the anti-3dmark people used: "Aquamark is great because it's based off of an actual game"

How many people actually play aquanox? I don't personally know one single person who bought aquanox, yet we don't hear any complaints about using aquamark as a bench because it's not a very popular game.
Again, it's not okay to use any bench that doesn't show Nvidia in a good light. Whether it's a game, a synthetic or make believe, if it shows Nvidia in a bad light certain people will say it's not a valid benchmark for whatever reason.

 

[see colon]

First we had the "can't use 3dmark, it's not a game" argument. Now we are getting "can't use tombraider, it's not very popular'.

this isn't what i'm saying at all. i'm saying that is a game is never played and only benchmarked (and the engine is never licensed), what makes it any differant than a synthetic benchmark like 3dmark's "game" tests.

i could argue a book's worth of letters on why you should not benchmark with TR:AOD, but in fact i like the game as a menchmark. it's got a nice, full feature set and stresses the video system enough to produce some interesting results. and while i can look at those results and infer that cardX has trouble with multiple complex (as in ps2.0) shaders, i won't be running to the store to purchase cardY just for it's ability to play TR:AOD at a faster clip. all things being equal i look at TR:AOD's results just like i do 3dmarks.
c:

ps- i can't say this enough about 3dmark, either. nVidia has always "cheated" at 3dmark. they even got caught doing it in 3dm2001. how does the fact that they now admit to cheating in 3dmark2003 make it any less valid. i see reviews now that won't run '03 because of the "controversy" but still use 2001 like it is somehow "cheat" free.