Why is 3dmark such the rage?

[KfistoRok]

How come when it comes to video cards all it seems to be about is 3dmarks? People are so obsessed with it that Nvidia and ATI play dirty to get more points. I think its getting old. Anyone else?

[billstuck]

It's common way to compare relative performance. Yea it's the games that count, but everyone likes to know how they are doing compared to everyone else.

[Falcon-K]

yea thats why i dont really care about 3dmarks anymore

[El<(')>Maxi]

What's wrong with a little competition? Hey, were all here on OverClocker's Forum's right?

For many of these guy's it's just a hobby. Even the one's at the top, and an expensive one at that. You can spend hundred's of hour's modding hardware. And considering the amount of money and time people put into their system's to make them faster, there really is no surprise in it's popularity.

Am I faster than the next guy? Did I put together a great performing machine? Besides, pushing the envelope is a good way to innovate, and is partly responsible for the development of the hardware that we all use.

So if you wanna game, then game. If you wanna bench too, then do it. What's the problem?

As far as this goes,

Originally posted by KfistoRok
People are so obsessed with it that Nvidia and ATI play dirty to get more points.

It's all about the money.


Maxi~

[SpaceyWilly]

yeah I agree 100% with you.

3dmark benchmarks seem to be becoming more and more useless as time goes on, especially with the news that ATI and NVidia are 'cheating' on them.

I think it would be good if
a)someone else made a decent benchmark program
b)Futuremark straightened up it's act.

I will say that for testing the results of, say, overclocking or something where you're looking at a before/after with the same comp 3dmark is still an excellent program. However, I think that for testing out your brand new video card, there's nothing like a good session of Halo or Maybe a little bit of C+C Generals, just to see what performance you've REALLY gained.

[Aeon Flux]

I used to use 3Dmark as a reference to how well I was doing on my modding. Then one time a while back I ran a little test and found out that the more times you run 3Dmark your score goes down. I know its common knoweldge now but it was a suprise to me years ago. Also I don't like using synthetic benchmarks after seeing that my athlon 1.1 tbird plowed my friends tbred. Its all in the configuration. Of course I enjoy modding and configuring. It's a great hobby. But I try and restrict myself to what I have now rather then buying mroe. For instance Im going to be water cooling ym computer soon. I made my own water block. My old geforce 2 I actually ripped the old heat sink off and put on a pentium heatsink and a 500 rpm fan and sandwhiched a fan on the other side of the pcb to cool it off. I then took the stock GPU heatsink and hacked it apart to put a heat sink the ram. As for your question I also don't understand the popularity with 3dmark . Its fun to watch but its inaccurate and in reailty doesnt provide bragging rights in any way shape or form.

[john240sx]

the best way to eliminate 'optimizations' would be to use a time demo from a predetermined game and compare results with one another. after a while (if this blows up), we would simply switch games for the time demo. by doing this it would be virtually impossible to 'cheat' on these benchmarks as we could switch games at any givin time.

the problem with this is everyone would need to have the same game to be able to compare results. this is why synthetic benchmarks are so popular.

i say we throw everyone for a loop and use 3DMark2000... i seriously doubt there's any optimizations in the current drivers for that benchmark :p

[Aeon Flux]

Originally posted by john240sx
the best way to eliminate 'optimizations' would be to use a time demo from a predetermined game and compare results with one another. after a while (if this blows up), we would simply switch games for the time demo. by doing this it would be virtually impossible to 'cheat' on these benchmarks as we could switch games at any givin time.

the problem with this is everyone would need to have the same game to be able to compare results. this is why synthetic benchmarks are so popular.

i say we throw everyone for a loop and use 3DMark2000... i seriously doubt there's any optimizations in the current drivers for that benchmark :p
I hate to say it but any enhancements on the driver level for 3dmark 2001 will also work for 3dmark2000

[JigPu]

lol :) The key to benchmarking is having an extensive suite to measure performance with. While 3DMark in itself contains a suite of tests, lately there have been cheats which degrade it's value. However, running several other benchmarks help elimiate false FPS gains from cheating.

When I benchmark my card I use 3DMark03, GLExcess, and Aquamark3. If I'm really in the mood (or doubuting the gains I got from a driver install), I'll run some of my others such as FogCity, Particle Fury, Particle Blend, and D3DBench. That, and playing UT2K3 allways helps me see if a new deriver helped :D

JigPu

[Aeon Flux]

Originally posted by Aeon Flux

I hate to say it but any enhancements on the driver level for 3dmark 2001 will also work for 3dmark2000
Sorry, the rabbit hit the enter button and I didnt get to finish. Anyway any enhancements on the driver level that helps 2dmark2001 will also help 3dmark2000. The enhancements that optimize scores in 3dmark2001 are specific settings for direct X. Certain things are turned off and quality is reduced to have a faster frame rate and improve the score. These enhancements that "tweak" the direct x options to improve scores will also due the same for 3dmark2000. The enhancements are basic enhancements that, I hate to say it, will also help some games and other programs that use the same functions of direct X taht were optimized.

[john240sx]

well my idea did a hindenburg (sp?) :(

yeah, i guess we need a fresh benchmark. i wish GLExcess did DX benching too but that would kind of go against it's name, it's a nice benchmark to watch though (only a few times, after that it gets boring)

[shiyan]

I'm surprised no one mentioned it yet, but there is also an element of my **** is bigger than your **** for some of the people who do it.

that is not the most important reason though. for people who run it all the time, they do it because they enjoy overclocking, vmodding, and tweaking.

but it is addictive also. you're always after a higher score, you try to get to top 1000, top 100, fist page of the ORB... takes time and money.

I guess it is also a proof of your skills at overclocking (and how much money you spend on computers). I'd really like to know how much money's worth of computer stuff guys who are always on the 1st page of ORB have used....

the graphics card makers focus on it because a lot of review sites use it.

[R4z0r4mu5 Pr|m3]

what would be a good idea would be for a benchmark to first ask how much money did you put into your system? then we can have a score that delivers who got the best bang out of the buck score :p

perhaps this will force down the prices, especially for mid-range cards and other parts

[john240sx]

that's easy. just run any benchmark that gives a score, divide that score buy how much you've spent and you got your PPD (points per dollar).

[Falcon-K]

Originally posted by john240sx
that's easy. just run any benchmark that gives a score, divide that score buy how much you've spent and you got your PPD (points per dollar).

yea its not like we are talking about serious calc here...

[SpaceyWilly]

hey that's a nifty idea

I like it

But if you just ask people how much they've spent, they're sure to lie.

So you just need to make a program that hacks into peoples' bank accounts and finds out how much money they've spent on their computer.

Although Im sure NVidia would find a way to cheat that, too...