3dMark 2001 is the best game I ever played!

[VBboy]

Actually, no.

Why is everyone so obsessed with the stupid 3dMark benchmark results?! It rarely corresponds to the real-life performance (e.g. games), just like with the 8500 vs 4600 (you know what I'm talking about). Everyone is bragging about their mama's big fat 3dMark scores, while in real life that may mean nothing. Especially on NewEgg.com, where customers post things like "Wow, this card gave me an X improvement over my Y 3dMark score.". Besides being good eye-candy, 3dMark is useful for maybe burn-in testing, that's all

Agree or disagree?

 

[ElDonAntonio]

agree

 

[VTrider]

I also agree...kinda but let me explain.


I agree that many users put way too much significance (sp?) on thier 3DMark 2001SE scores, esp. in regards to real-life performance. I've been running the 3DMark benchmark prolly since day one and I still do to this day. The thing is that it's still a benchmark, relative to your system. I get a new video card, and/or new CPU and one of the first things I do is run 3DMark 2001SE just to see the difference from the old setup.

I find their database of scores not only useful but an accurate resource when I want to compare my system (cpu/vid) performance to my own. It still puts things in perspective. I guess what i'm saying is that although you feel it rarely corresponds to real-life performance (to which I agree to a point), a high/low 3DMark 2001SE score usually does correspond appropriately to to high/low performance.


-VTrider

 

[Rand]

Go to the MadOnion forums if you want a laugh. The commonly given advice on boosting ones 3DMark includes dropping to point-sampling, setting mipmapping to the lowest settings possible, alter LOD bias to the fastest settings available, overclock everything to the absolute limits regardless of whether you get graphical issues or stability problems... as long as it makes it through 3DMark once!

They'll do anything and everything it takes even if it completely kills visual quality to get a higher score. I've even seen people handing out recommendations to diusable rendering of all textures just to get a higher score!

If they go through all this just to get a high score, what is the point of benchmarking?
Quite obviously their not going to play real world games with such settings.

Personally, I count it as lucky none of them have yet figured out that you can disable rendering in 3D and allow 3DMark to run through it's testing suite without rendering anything. I'm sure they'd probably do that to get a higher score also.


3DMark is decent as a very vague indicator of gaming performance, but it definitely does not corellate perfectly with real world application usage, and both nVidia and ATi have heavily optimized their drivers specifically for 3DMark in a manner quite impossible in typical games.

3DMark's worth as a benchmark has been diminishing ever since release, and at the present time it's of only minimal benefit.
It's vaguely useful for determining if their are any severe performance issues, and perhaps mildly useful as a stress test.
As a benchmark it'll really only give you a hint of real world application performance at this point in time.

Probably the only very useful aspect left in 3DMark is MadOnion's catalogue of average performance recieved with each processor/graphics card configuration. By comparing your setup to other the catalogue of average scores you can get a very rough idea of how much performance gain you would see by upgrading.
Though even that is of dubious benefit as the individual requirements can vary drastically depending upon the type of game in question.

 

[MiExStacY]

but it is good to test if you videocard is stable running it in loop overnight with prime95,but for benchmark i do laugh at those spending there whole life to break there records,especially those that upgrade from a gf3 ti200 to a gf3 ti500 to get 1-2k more points

 

[Wingznut]

3dMark is good as one way to compare tweaks to your system. But to use it as to imply that one pc is better than another because it gets a higher score... Well, that's just plain silly.

Notice that I said "one way" above. There is no complete benchmark to test system performance.

 

[Acanthus]

3Dmark 2001 game demos are actually based on real games dont forget, these games are now outdated and 3Dmark 2003 will be DX9 and have a whole new sleu of graphics engines, i know car chase was based on the Mercedez Benz Truck Racing Engine, Dragothic on the Theif II engine, Lobby on the Max Payne engine, and Nature on a proprietary engine from (insert companys name here) its a more developed version of the engine used in Dark Age of Camelot by Mythic entertainment. It says the companies names in the credits at the end of the demo.

I dont throw a ton of wieght on 3dmark scores because it doesnt cover all games, and nowadays 2k1 is showing its age, 2k3 on the other hand i will throw some merit behind until the engines its based on fade.

 

[VBboy]

Don't get me wrong, I have no problem with 3dMark as a single benchmark. But to bet everything on it, especially to drop your videocard's quality settings just to get a higher 3dMark score, is really silly. Even MaximumPC noticed that some PC manufacturers started shipping computers with MipMap rendering set to "fast", which decreases visual quality a lot -- just to get a higher 3dMark score. Why not just run it at 8 bit at 320x200, and get the highest score?..

A couple of games like MaxPayne and Serious Sam is what I call the best way to benchmark a videocard. Besides, 3dMark is not tweaked or optimized for max. performance at all!! For all we know, their code is extremely inefficient, but who cares -- it's not a game, it's just a program. The only programs that have the right to be slow and inefficient are benchmarks

My biggest question is this: who would PAY for 3dMark 2001 SE?.. I mean, what value do you get out of it?! You like fancy graphics, get a 3d fish screen saver

 

[blackhawk]

Its simply a way of checking for improvements with card/settings changes on your machine. You can find out what an overclock will give you, what your limiting hardware is and what the new video card offers in speed improvements on your machine.

Its not a game but a benchtest for a result for comparison purposes. If someone wants to brag about it, more revenue for MadOnion!

 

[LanEvoVI]

Originally posted by: VBboy
Actually, no.

Why is everyone so obsessed with the stupid 3dMark benchmark results?! It rarely corresponds to the real-life performance (e.g. games), just like with the 8500 vs 4600 (you know what I'm talking about). Everyone is bragging about their mama's big fat 3dMark scores, while in real life that may mean nothing. Especially on NewEgg.com, where customers post things like "Wow, this card gave me an X improvement over my Y 3dMark score.". Besides being good eye-candy, 3dMark is useful for maybe burn-in testing, that's all

Agree or disagree?

Agreed!

 

[Insidious]

Agreed (sort of)

It's a nice tool to compare tweaks on a given machine, but just plain idiotic to try and compare one machine to another.

But of course, only an idiot would destroy the video quality of a system to prove they have the best video process.

 

[Goi]

Then apparently there're quite a number of idiots in AT

Seriously, I don't even bother to run 3DMark2001SE much...maybe once or twice, but to me its just a senseless benchmark that doesn't neccessarily reflect real world performance. I benchmark the games I play, and that's Q3A. Sure, its old, and it doesn't scale that well, but its what I play, so its what I'm concerned about. I could care less if it increased my 3DMark2001SE score by 3X if it didn't increase my Q3A performance.

 

[DeRusto]

Agreed, but...

I only use it when to see what sort of performance my setup SHOULD be getting..and if I have a problem, I may run it again to see how the score has changed.. take, for instance, when I noticed that Serious Sam was running very poorly on my machine and that HalfLife was running at about half the frames of my friends comp.. (me Radeon 8500/him GF2MX400)..I ran 3dMark and saw that my score dropped from 8000 to 5000..that was a problem

I have it fixed now..but 3dMark did sort of help me figure it out...

 

[Dug]

Why is everyone so obsessed with the stupid 3dMark benchmark results?! It rarely corresponds to the real-life performance (e.g. games

Because #1- its EASY, just hit the benchmark button. (I don't even know how to benchmark most games, too many variables)
#2- It looks better than other benchmarks. (Well except now there's UT2003)
#3- There's one number at the end.
#4- Bigger number means better performance
#5- When you hear 20000+ 3dmarks, it peaks your interest.
#6- Nobody else has come out with a good looking, easy benchmark
#7- People need to show off some kind of number- like a 0-60 time for cars.
#8- There's no such thing as real-life performance because games aren't real life.

 

[Goi]

Interesting points, but several of them are flawed IMHO.

#1- its EASY, just hit the benchmark button. (I don't even know how to benchmark most games, too many variables)

That's definitely true, but there are also many other options available when it comes to tweaking, and that's a good thing too. Anyway, the ease is a bonus for beginners, but I'm guessing most of us here are able to perform a simple benchmark in the games we play.

#2- It looks better than other benchmarks. (Well except now there's UT2003)

That's true to some degree too, but it depends on your taste. However, benchmarks aren't meant to be good looking, they're meant to just produce performance figures. The stuff you see on screen is simply a side effect of the measurement process, and having it look good is simply a bonus.

#3- There's one number at the end.

Which is good for beginners, but not so good for more experienced people. What does that number mean? How is it obtained? Also, actually there's a breakdown of individual scores in 3DMark2001SE which you forgot to mention.

#4- Bigger number means better performance

Not neccessarily. It simply means better performance for that particular benchmark. Even within the benchmark there are many different "game scenarios", some video cards perform better at one than another.

#5- When you hear 20000+ 3dmarks, it peaks your interest.

Well I don't know about you or the rest of the people here, but it doesn't peak my interest.

#6- Nobody else has come out with a good looking, easy benchmark

True, but again the ease and good lookingness is more for beginners.

#7- People need to show off some kind of number- like a 0-60 time for cars.

Yup, and I'm not opposing that a 3DMark2001SE score is ONE of the numbers to show, but many people are using it as the ONLY number worth showing, which is what I'm put off by.

#8- There's no such thing as real-life performance because games aren't real life.

OK, you're taking it literally now, real-life/real-world isn't meant to be taken literally, it simply means performance in the games you play, apps you run. Now, if all you play and run all day is 3DMark2001SE, then that's the benchmark you gotta use since its a direct reflection of the performance you're gonna get, but I pray that most people have more life than to do that all day long.

 

[Dug]

Hey- I was just answering the guys question.

 

[VBboy]

Thanks for your feedback, guys!

You know, just gathering some information, not trying to make anyone fight or anything. I do agree that 3dMark is easy to benchmark, but I don't agree that benchmarks should be good-looking However, it's nice to have a pretty benchmark, as it helps you wait for it to end (as opposed to seeing something like Prime 95 do its thing).

As far as 3dMark being based on actual game engines... MaxPayne and other games that it's based on went through a lot of optimization after being released, but I'm pretty sure 3dMark kept their engines unchanged, so it doesn't even correspond to those very games...

I am going to make my own VbBoyMark Pro 2004. Watch out, Mad Pepper!

 

[Maggotry]

I think 3DMark is great because my 2 year old son loves to watch it! He also likes watching the other popular benchmarks: Commanche 4, Code Creatures, DroneZmark, and VulpineGL. Other than just watching the pretty pictures, I don't give a damn about 3DMark scores.

 

[Acanthus]

ill admit if i saw 20000+ 3dmarks it would get my attention, i play the latest games (warcraft 3, UT2003, soon to be AC2, Dark Age of Camelot Shrouded Isles XP, Dronez, Sacrifice, more) and it DOES somewhat tell you your overall performance in DX8/8.1, (it reflects in no way the ability to render Doom III or any DX9 affiliate) Basically higher 3dmark scores = higher overall performance (this performance will vary from game to game depending on texturing, lighting, Aniso, FSAA, poly count, Graphics engine, and more.

I agree with the 0-60 idea, people use it as a standardized measurement of 3dspeed because its easy to use, and being the average of 4 games instead of a single graphics engine, i feel its more accurate. Obviously if your directly benching the games you are going to play thats even better. I watch my average FPS in UT (its always something like 59.98 stupid refresh rate).

 

[dexvx]

No single benchmark is ever perfect. Thats why benchmarking companies try and do multiple benchmarks (SPEC, 3dMark, Winstone, SysMark, etc) to try and tabulate scores. But keep in mind that whenever a benchmark gets remotely close to popular, people try and cheat the benchmark so that they get a higher score and try and bloat their system to be as good as a system that is better.

Thats why a lot of benchmarks need to be updated on a yearly basis. I look forward to the next 3dMark because 3dMark has always been at the forefront of consumer level 3d visual technology.

 

[woolmilk]

what 3dmark is missing is a comparison of rendered images.

 

[Acanthus]

the "pro" version has IQ tests for comparison, also 3Dmark is one of the only benchmarks that is hard to "fool" i believe 2003 is rumored to have a way to detect mipmap settings. so setting to fastest or best quality will be able to be seen in the future.

 

[ElFenix]

i wonder if ut2k3 benchmark uses some preset stuff or if it uses the settings defined by the user in the game?

 

[VBboy]

Hehe, if any of you can remember, there was no way to benchmark Doom and Doom II. Those were the times. Life was simple. Either you computer could run Doom II, or it sucked!

 

[merlocka]

Reasons 3dmark is dumb:

1) threads titled "post your 3dmarks!" - these should instead be titled "computer dorks unite and join in this group video masturbation!"
2) people tweaking their systems at the expense of video quality just to be "the highest score with T-bird and geforce3 ti200", at the same time they are askewing the database.
3) video card manufacturers - their driver teams spend hours and hours optimizing their drivers to score high in 3dmark when they could be fixing bugs instead.

Reasons I like 3dmark:

1) I can compare changes in my system - overclocking, new drivers, new video card, etc... and have a semi reliable set of datapoints to reference.
2) I can look up systems on the ORB to determine how systems scale and how different systems perform on different tests.
3) When I assemble systems for people, I can throw that on there for a nifty 3d-demo. Amazing that regardless of the score they get they always think it must be the fastest score out there