3dMark not a worthy benchmark anymore?

[TravisBickle]

I notice that on the big websites 3dmark is not used as a benchmark much on last year's cards I was looking at. isn't it respected any more, and why?

 

[MrGrim]

I can't see how it ever was. All you have to do to score well was get a GeForce2 GTS.

 

[MustangSVT]

the scores do not reflect real life gaming or applications performance due to how it overstresses one feature over other that is not used as much in most programs. Such as fillrate.

it's good source to see if your components are working up to what its supposed to do since there are similar machines with similar settings benchmarked in their database. However comparing different video cards to each other was not clear cut with it.

also it helps to check your system's stability(spell?) since it'll stress every component.

 

[IBMer]

3dMark2000 totally reflects a real world situation:

1. The 3dMark2000 tree:

3dmark2000

2. The 1024x768x32 benchmarks for MDK2:

MDK2

 

[Taz4158]

It's a useless benchmark. The only validity it has is to benchmark your OWN system.

 

[Deeko]

anymore?? It never was. I would rate it as the worst benchmark ever created.

 

[IBMer]

Again people take a look at what I wrote. The benchmark clearly mimics the same thing you see in MDK2 and Probally Evolva or really any T&L game that was devoloped for HW T&L. What part of this do you not understand?

 

[MustangSVT]

Geforce DDR faster performer then Radeon DDR? i dont think so. dont make a fool out of urself and stop while u have a chance.

 

[Deeko]

IBMer, MDK2 is heavily in favor of nvidia cards. 3DMark is NOT a good benchmark, what part of that do you not understand? Do you honestly think that a GeForce SDR is faster than a Voodoo5?

 

[lsd]

<< Geforce DDR faster performer then Radeon DDR? i dont think so. dont make a fool out of urself and stop while u have a chance. >>


Maybe you're the fool..

 

[Bartman]

OUCH!

 

[JayPatel]

its basically good for eyecandy. as a performance benchmark i think it sucks...

 

[Deeko]

lsd, the DDR only won in the 640x480 test. The Radeon won in both 1024x768 and 1600x1200. Probably the ones in between, too. So I wouldn't call him a fool, for he was right.

 

[MustangSVT]

i guess i dont need to add more to that. and all the facts about 3dmark and benchmarks have been said already so i dont have to add on to that either.

i cant change ppl's minds, i can only share mine.

 

[IBMer]

IBMer, MDK2 is heavily in favor of nvidia cards. 3DMark is NOT a good benchmark, what part of that do you not understand? Do you honestly think that a GeForce SDR is faster than a Voodoo5?

When almost all benchmarks show heavily in favor of one brand, why does this mean something. I have the Radeon, V5 and GTS and I know how each performs.

BTW the GeForce SDR has more stable geometry in games like Giants and Sacrifice where they have LOD engines for T&amp;L.

 

[Deeko]

Most benchmarks do NOT show what 3DMark and MDK2 show. They DO NOT. And the GF SDR isn't better than the V5. 3DMark supporters can go cry about it, 3DMark simply is not good.

 

[ahfung]

Deeko, have u realized that new games like Giants/Alice/Sacrifice DOES reflect a trend?

As I said beforem, 3D Mark 2000 is a benchmark for upcoming(now is current already) games. So anyone use the default score to judge how well their video cards plays legacy games is a fool.

And please listen:

3D Mark 2000 DOES exposes the weakness of display cards with low memory bandwidth. What you need is to benchmark it at 1024x768x32 or above. Then you will see how weak MX and SDR are.

Say it again, 3D Mark 2000 DOES NOT suck.

 

[lsd]

<< lsd, the DDR only won in the 640x480 test. The Radeon won in both 1024x768 and 1600x1200. Probably the ones in between, too >>


The Radeon was basically TIED with a card which is one generation older than it. And the DDR did not only win in the 640x480 tests.

 

[ahfung]

GF DDR only has a memory bandwidth of 4.8GB/s, verus 5.87GB/s of Radeon DDR. Under 1024x768x32, a memory bandwidth limited situation, the DDR still ties with a newer generation card which has 1GB/s extra memory bandwidth and bandwidth saving techniqes like Hyper Z. What else can you complain about a card released more than 1 year ago?

 

[Jim]

My card is better than your card!!!!!!!!!!!!!!!!!

 

[DaveB3D]

3Dmark is only for people who understand what the results mean. If you don't understand the results they are pointless. Unfortunately, most don't understand them...

 

[Taz4158]

IBMer sure trolls a lot of boards.....

 

[IBMer]

It isn't trolling if you put proof and logic behind your posts, unlike somepeole that just insult others.....

 

[TravisBickle]

3dMark is a forward-looking benchmark, right?
so why don't we use 3dMark 1999 for more real world performance?
(sorry, that was my more facile side talking)

 

[DaveB3D]

Deeko,

Actually MDK2 uses immediate mode rendering, something that many cards simply aren't optimized for. Typically optimizations go more the Q3 route of DrawElements. Like with the V5 that was long the case, but as the drivers progressed they got much better. The drivers we had internally (which aren't released yet, but hopefully will be soon by x3dfx) are much better than before. So good in fact that you can enable the T&amp;L checkbox and still have great performance.


Taz,

What don't you STFU. IBMer has shown a great deal of intelligence in the posts I've seen while I've been posting here. You've not yet shown anything to prove you to do the same, so until you do you have no business talking.