• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

3dmark vs the real world

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Big_Baller

Member
Joined
Oct 1, 2002
Location
Seattle, WA
I saw this article and liked it so I thought I would let you guys read the article. Great review its done by ace hardware. I have hardly heard of them but the atricle seemed solid enough.

In the end, 3dmark = junk on a stick imho.
 
I guess I just take it for granted that 3DM03 is a GPU bench (and a good one at that)

But I can see your point as it's advertised as a gaming bench.

Since it determines a score based on the GPU gaming benches only, and leaves out cpu testing in overall score, I can see how someone out to demonize 3Dm03 would have an easy time of it.

It has it's limited purpose, just as many other benches do.
 
I find it funny that 3dmark didn't even give a half-off representation of your gaming proformance. Instead it gave about the exact opposite results. lol Bench-marketing. Darn straight.
 
I use 2k1SE, PCMark and a handfull of games that ARN'T popular with the online review sites to test such things. Really takeing 3dmark03 to basicly a GPU only bench wasn't a wise move. Gameing rigs are the sum of the parts in the end.
 
i ran my system below and only got a few frames per second better than the 350 mghz pII computer....just shows how this so called "gamers benchmark" is not valuable whatsoever in judging real life gaming. I dont even see the purpose of running this benchmark anymore...
 
Falcon-K said:
i ran my system below and only got a few frames per second better than the 350 mghz pII computer....just shows how this so called "gamers benchmark" is not valuable whatsoever in judging real life gaming. I dont even see the purpose of running this benchmark anymore...

How about a GPU bench?
 
Big_Baller868 said:
I saw this article and liked it so I thought I would let you guys read the article. Great review its done by ace hardware. I have hardly heard of them but the atricle seemed solid enough.

In the end, 3dmark = junk on a stick imho.


ACE normally has very good reviews and technical articles but sadly this was not one of their best . It totally misses the point . 3dm 2003 is made specifically for GPU stress testing in techniques agreed on by the big players as being representative of future software .

It is therefore no small wonder that CPU speed made no real difference . We already know that . It is very important to specifically test the GPU and not only have tests like 3dm 2001SE which was too easily affected by mem timings and CPU changes . For example an overclocked GF4 MX 460 in a fast rig will give very good 3dmarks in 2001SE and make it look better than lets say a GF3 or Matrox Parhelia on a midrange system or slightly lower system . But that is rubbish in the real world . No 3dMark score can make that MX run Doom3 properly or any directX 8 class program for that matter . But those scores would fool many people because 2001SE allowed the system to make a vcard look better than it really was , and it allowed cards to get scores in tests when they should get zero .

What 2003 has done is limit external system factors and at the same time be harsh on cards for not supporting standard features .This is very important, when directX9 games come out there will either be no workaround for such features in games or the workaround to allow lower cards will be slower +/- uglier . You cannot permit a vcard to parade with high scores that are attributable to the cpu or mem timing or fsb . All together these changes from 2001 SE have produced a much bettter vcard benchmark .

Some people compare it to todays games and say that it is not representative :rolleyes: . DUH ......... that is because it is meant to be a future or forward looking benchmark just like all of the others before it . 2000 and 2001SE didn't represent real world performance at the time either . But look at them now , there are very few cases where a card is consitently better than another in 3dmark 2000 or 2001SE but is slower in real world D3d tests ! 2003 has to be given time . It is actually amazing how many supposedly experienced reviewers strongly attack 2003 but ignore the same and worse faults/limitations in 2000 , 2001SE and any other benchmark for that matter . Sadly, many have been infected by a certain IHV who has an axe to grind and benchmarks to defraud . ( I'm not lumping ACE's in this crowd and hopt to never have to do it )

Finally as a side note , I have no problem with overall game syatem benchmarks like 3dmark 2001SE as these have their place but we all need to stop getting confused about the scope and aim and limitations of each benchmark out there .
 
Falcon-K said:
i ran my system below and only got a few frames per second better than the 350 mghz pII computer....just shows how this so called "gamers benchmark" is not valuable whatsoever in judging real life gaming. I dont even see the purpose of running this benchmark anymore...

They should never have called it the Gamers' Benchmark but the 3dbenchmark or something like that .

But with that said , it is aimed to the future not the past or the present , which is what any current game is ! This is FutureMark not Today'sGamesMark , for that , go and buy some of today's games . So don't expect it to correlate with 'real world' situations , it won't ; at least not for 1.5 to 2 years , by which time it will be CPU limited , and even midrange cards will be 10,00 plus points , just as is the case with everybody's favourite 2001SE ! And by that time just as with 2001SE it will then be USELESS as a predictor of future performance since the world would have caught up and surpassed it .
 
AKULA said:
I use 2k1SE, PCMark and a handfull of games that ARN'T popular with the online review sites to test such things. Really takeing 3dmark03 to basicly a GPU only bench wasn't a wise move. Gameing rigs are the sum of the parts in the end.

I see your point, but the accuracy of 3DM01 is highly dubious. For example, I can affect the score several thousand points just by changing my FSB and memory timings......Things that wouldn't show the same kind of increase in a game.

One think I like about 3DM03 is it's ability to instantly pinpoint an issue to either the GPU or somewhere else in your system. While a low score in 3DM01 could mean ANYTHING, a low score in 3dM03, and you KNOW it's in the GPU, or GPU settings:)
 
Well that several thousane points is merely a percentage. And it is only a benchmark that can give you a closer look at where you might lie vs all the other computers.

As for Futuremark.... they did a stupid thing, relying so mych on the video card. Then again, it is called 3dMark 2003. Too bad for them they claim that it is an overall gaming performance analysier, or something like that.

All the previous 3dMark programs were more closely an overall system bench, even to 99MAX. But those were the days.
 
Yes the silly old days ......... when a video card test was actually a memory bandwith , memory timing and CPU benchmark .

As I said there is a set place for GPU tests and system tests , maybe 3dmark2003 should have had a system test version or setting , I would have liked that , but I can in no way say that the days of 98,99,2000 and 2001SE are good old days .
 
Back