THG looks at 3dmark2003

[human2k]

linky

 

[yllus]

You have to hand it to Tom's people for at least one thing: They certainly know how to stir the pot.

 

[CraigRT]

hate tomshardware.. i wish someone would fire the guy

 

[Pastore]

I feel like toms hardware visits like all the forums possible, and just mimics what all the little whiny bastards have been saying about the new stuff. Toms writers are just one big n00b.

 

[human2k]

Originally posted by: Beast1284
I feel like toms hardware visits like all the forums possible, and just mimics what all the little whiny bastards have been saying about the new stuff. Toms writers are just one big n00b.

 

[Wolfsraider]

wow thanks for the read

nvidia may well be on track to rid them selves of 3dmark.
after reading that,i am left to wonder at 3dmarks reasoning for just reskinning the images? to stress test?
why would you use code so poorly written?wouldn't it benifit (futuremark) to optimize the code to run on any machine in a true game enviroment?

am i wrong in thinking that simple code would give a better test of the machine?since no two video cards are the same (even in reading / running the code),rather than reskinning,why didn't they use a conversion of one skin to another?or would i be wrong in thinking this?

the one rule i have read over and over in coding is keeping it clean.why then would this not fit that rule?

doing mods for morrowind(yes i know this may not fit this scenario) this is the exact thing they warn about "not doing" right?

please explain whether or not i am on the mark here as i am a newbie thats very confused.

thanks

mike

 

[Snoop]

Originally posted by: Wolfsraider
wow thanks for the read

nvidia may well be on track to rid them selves of 3dmark.
after reading that,i am left to wonder at 3dmarks reasoning for just reskinning the images? to stress test?
why would you use code so poorly written?wouldn't it benifit (futuremark) to optimize the code to run on any machine in a true game enviroment?

am i wrong in thinking that simple code would give a better test of the machine?since no two video cards are the same (even in reading / running the code),rather than reskinning,why didn't they use a conversion of one skin to another?or would i be wrong in thinking this?

the one rule i have read over and over in coding is keeping it clean.why then would this not fit that rule?

doing mods for morrowind(yes i know this may not fit this scenario) this is the exact thing they warn about "not doing" right?

please explain whether or not i am on the mark here as i am a newbie thats very confused.

thanks

mike

Good questions, but dont expect much feedback as the Fanboi's are going to attack.
IMO, 3dmark is dung. Its worthless except to see the pretty rendered scenes (which btw, arent that great), and to tell your e-buddies how cool you are because your machine scored xxxx in 3dmark. It means Jack-sheit in the real world. Further, I noticed on bapco's sight, they and mad onion are collaborating on benchmarks now, which lowers their credibility even further in my book.

 

[merlocka]

There have been alot of us on these forums with the same thoughts regarding 3Dmark, and not because it "favors" any video card.

The reality is that ATI and nVidia have spent a ton of resources optimizing code for this useless benchmark. This time could have gone into optimizing their drivers for actual games.

I can hardly wait until Doom3 arrives for 2 reasons... I'm sure it will be a great game, and it will provide the next real benchmark for video cards (and optimizing for doom3 will only benifit consumers).

 

[DX2Player]

Whatever....Nvidia was part of the team till December, to say that they didnt have a chance to influence it is just retarted. Its just a buisness tactic and well thats ok. In a year from now Nvidia might be on top again and well see if the fans still think its a BS test, and if not.........whatever.

 

[Spicedaddy]

Pixel Shader 2.0 41.1 fps 12.0 fps 14.2 fps

wtf? nVidia should stop whining about PS 1.4 and start optimizing PS 2.0 on the FX...


BTW, would falling back to PS 1.3 instead of 1.1 in the tests that use 1.4 (for cards that don't support 1.4) improve performance?

 

[IKeelU]

I don't really know much about how directx works, but I thought the purpose of such an API was so that developers wouldn't have to write card-specific code. I mean, if you design a program around DX9.0 for example, you're not writing for the FX or the 9700 directly. And if your graphics card/drivers aren't optimized for, say, UT2003, wouldn't that mean they aren't optimized for DX8.1 in general?
That being said, if the FX did poorly using the 42.63 drivers, wouldn't it imply that those drivers simply weren't optimized for DX9.0 (or whatever) features? Maybe someone can clear this up for me...

 

[Wolfsraider]

i care not whether ati or nvidia like or dislike the benchmark.

optimizations legit or otherwise are going to happen.

i am not conspiracy theorizing here.bapco etc.. is not my concern.

benchmarking has always been a nice way to test the performance of my machine.

but i see now that benchmarking is getting way out of hand by users and promoters

benchmarking is a way to test components we all use,but when the honesty of the benchmarks are compromised ...who wins?

i own ati but i agree with nvidia on this benchmark as well.

the hell with the marketing folks

what happened to the old

"if you build it and build it right they will come" sales that is.

something is sorely missing.

mike

 

[merlocka]

Originally posted by: IKeelU
I don't really know much about how directx works, but I thought the purpose of such an API was so that developers wouldn't have to write card-specific code. I mean, if you design a program around DX9.0 for example, you're not writing for the FX or the 9700 directly. And if your graphics card/drivers aren't optimized for, say, UT2003, wouldn't that mean they aren't optimized for DX8.1 in general?
That being said, if the FX did poorly using the 42.63 drivers, wouldn't it imply that those drivers simply weren't optimized for DX9.0 (or whatever) features? Maybe someone can clear this up for me...

I think the point nVidia made is that they proffered 2 driver sets, the only difference is that the latter set improved 3dMark03 scores by almost 50%.

No other obvious benifit to consumers. That work could have gone to optimizing drivers for an actual game.

 

[Wolfsraider]

Originally posted by: merlocka

I think the point nVidia made is that they proffered 2 driver sets, the only difference is that the latter set improved 3dMark03 scores by almost 50%.

No other obvious benefit to consumers. That work could have gone to optimizing drivers for an actual game.

but why stop there?

couldn't they just code to ignore 8 out of the 11 reskins by using an algorithm?

isnt this what is possible in this benchmark?

hardware that passes the benchmark are more important to these companies than games are,since they dont sell games,but do sell video cards.therefore they have a vested intrest in benchmarking utilities.
edit/;
then why would nvidia back out of this benchmark-i mean they proved it can be optimized with just a driver update,50% optimized no less: end edit


mike

 

[merlocka]

Originally posted by: Wolfsraider

Originally posted by: merlocka

I think the point nVidia made is that they proffered 2 driver sets, the only difference is that the latter set improved 3dMark03 scores by almost 50%.

No other obvious benefit to consumers. That work could have gone to optimizing drivers for an actual game.

but why stop there?

couldn't they just code to ignore 8 out of the 11 reskins by using an algorithm?

isnt this what is possible in this benchmark?

hardware that passes the benchmark are more important to these companies than games are,since they dont sell games,but do sell video cards.therefore they have a vested intrest in benchmarking utilities.
edit/;
then why would nvidia back out of this benchmark-i mean they proved it can be optimized with just a driver update,50% optimized no less: end edit


mike

Can you repeat that in english?

 

[Wolfsraider]

but why stop there?

couldn't they (video card companies)just code to ignore 8 out of the 11 reskins by using an algorithm?

isnt this what is possible in this benchmark?



hardware that passes the benchmark (get the highest scores)are more important to these companies than games are.
since they dont sell games.these video card companies sell video cards.

therefore they have a vested intrest in benchmarking utilities.(the average joe looks at or is shown benchmarks in making video card purchases)

edit/;
then why would nvidia back out of this benchmark-i mean they proved it can be optimized with just a driver update,50% optimized no less: end edit

does this make more sense

<----lives in a cave and struggles getting ideas out lol

mike

 

[Noid]

I still *snicker* when I hear of Tom's Hardware...
Not cuz I have any opinion about them,,, but cuz..

When I was at the AMD 'Reality Check' here in Chicago ... they got 'boo'ed

I kinda feel sorry they get so much flack.
_________

As for the benchmark ... I just wanted to see some eye candy .. and I was dissapointed...
As for its tests .. I like the way 2001SE displayed what was being tested in each case... if a card didnt support it .. it said so
The only thing I did like was the extraction of more system specs.

 

[ViRGE]

Nvidia does make some good points here about just how poorly written the 1.1 algorithms seem, compared to the 1.4 algoritms. I'd really like to hear an explaination from Futuremark on why they didn't at least take up the advice on not doing multiple skinning.

 

[DAPUNISHER]

As the saying goes "A fool and his money are soon parted" and many fools will upgrade their vid cards with the thought that they need the latest 300-400$ card to game in 6 months

It's the old "smoke and mirrors" gag :frown: I'm not a serious gamer so my 8500LE@300/315 (11106 3Dmark2001SE) will suffice for some time yet, and since I play more ghost recon than anything else and it plays great with everything on highest settings I'm content

 

[Wolfsraider]

time to realize this benchmark is useless if everyone can cheat this easily lol

read this warning big jpegs 1280x1024

 

[human2k]

Its just a benchmark, you didnt pay anything thing to use it, quit complaining, use something else....sheesh!

 

[kuk]

Originally posted by: Wolfsraider
time to realize this benchmark is useless if everyone can cheat this easily lol

read this warning big jpegs 1280x1024

Although I dislike 3dmark, none of their benchmarks are useless because of cheating.
A benchmarking application is used to detect problems in a system configuration and analyze differences between configurations of your own system (new CPU, video card, etc), while sometimes having a trusted parameter (ie. a hardware website). As long as this parameter doesn't cheat, there is no reason to invalidate benchmark programs because it allows cheating.

Cheating does not make your computer faster. Cheating does not make the other guys computer faster. You do not get money for having the l33test computer in a certain benchmark (well, not where benchmarking is not controlled by an 'audit'). Cheating is only a "problem" to those who have the urge to show the world how fast their computer is (or reafirm their manhood; sometimes both get mixed up).

Fixing these cheats is something nice, but in the end, for you and me, it really doesn't matter that much.

Kuk

 

[chizow]

Its not a benchmark, its a glorified screensaver that flashes a few numbers here and there. The ATI DX9 demos do this better anyways.

Its pretty obvious why the clowns at FutureMarketing are making synthetic benchmarks rather than ACTUAL games, they have no friggin clue wtf they're doing.

I'd hate to think what would happen if they actually made a game, it'd be hacked to all hell and you'd see minimum system specs like:
Required:
Radeon 9700pro or GeForce FX

Optional/Recommended:
CPU/PentiumMMX 133
RAM/64MB RAM
HDD/200MB Free Disk Space
Sound: Not required, it will only hurt your score.

LoL!

Chiz

 

[Wolfsraider]

Originally posted by: human2k
Its just a benchmark, you didnt pay anything thing to use it, quit complaining, use something else....sheesh!

ok maybe i am being vocal but i have that right

Although I dislike 3dmark, none of their benchmarks are useless because of cheating.

it isn't that big a deal.your right,but i find it wrong that this benchmark is so flawed.
"
if i am comparing my system to another system with lower clocks on their video and i have lower scores, ("i" being figurative as "i in this case being posters)then what worth is this benchmark?since 3dmark has been the "most widely used" benchmark, or at least on the top 3 list for most people,it also has the potential to be the most exploited.how can you determine if its cheaters or a hardware problem?

i was just being me sorry

i realize the error i made and back away from the keyboard...i just thought that this was an intresting add to this thread lol

good day