[MrBungle]

Why 3DMark06 Shouldn't Be Taken Too Seriously

by Mr. Bungle

1) quad cores get a huge points advantage in 3DMark, but not in any game out today.

2) the stock benchmarks runs with no anti aliasing and minimal texture filtering. the true test of a gaming system isn't how many frames per second you can get at minimum settings; but rather how high you can turn up the settings while maintaining playable frames.

3) the sea monster is an idiot. i mean come on, all it had to do was jump up from directly beneath the airship and knock it out of the sky. then it could have picked the people out of the water one at a time and enjoyed a tasty meal. instead, it goes around trying to show off, flying over the airship on its first pass. the sea monster got what it deserved.

feel free to add to my list if you want. there may be some things i missed.

 

[iampoor]

Depending on yor Gpu's core and memory speed certain combinations might get higher scores even though in the real world/other games they will do better

 

[Criswell]

Quote:

Originally Posted by iampoor Depending on yor Gpu's core and memory speed certain combinations might get higher scores even though in the real world/other games they will do better

What?

I agree Bungle, it should all be ran on two different settings, the default setting, and Full Setting where everything is enabled. But then again scores should only be compared if they are running the same resolution.

 

[Mhill2029]

If you get higher scores in 3DMark06, you'll have higher frame rates in games. I trust the results on it myself....

 

[iampoor]

What I meant was..

Supose your core is set to 500mhz and your memory is at 2.0ghz

Now suppose it only needs that memory at 1.6ghz

Now there is 400mhz of unused bandwith.

ANother game could use all 2.0ghz of abndwith but only need a 400mhz core speed.

It was just an idea I had....

 

[t4ct1c47]

Interesting post MrBungle.

A Rep +1 me thinks

Quote:

Originally Posted by iampoor What I meant was.. Supose your core is set to 500mhz and your memory is at 2.0ghz Now suppose it only needs that memory at 1.6ghz Now there is 400mhz of unused bandwith. ANother game could use all 2.0ghz of abndwith but only need a 400mhz core speed. It was just an idea I had....

That is so wrong its untrue. For starters you do realise that no game in existence has access to the actual hardware of the graphics card or has any say on how well it performs? The game application uses the drivers, themselves attached to the very kernel of the operateing system, to tell the hardware what to do. As a rule of thumb, the higher the resources (hardware specs and capabilities) the better performance you get in a given application.

The only time when this is untrue is in cases where people have core, shader and memory clocks on a graphics card, rather than just core and memory. Overclocking via software dynamically raises the shader clock as the core clock is raised. However, some shader to memory frequencies will be detrimental to performance if the shader clock itself is unable to operate at the correct syncronisation.

Bottom Line

Game applications do not control the bandwidth of communication between the GPU and RAM on the graphics card. This role is performed by the onboard MCH. Higher resources (clocks) means more performance in rendering the images you see on your screen.

 

[iampoor]

Quote:

Originally Posted by t4ct1c47 That is so wrong its untrue. For starters you do realise that no game in existence has access to the actual hardware of the graphics card or has any say on how well it performs? The game application uses the drivers, themselves attached to the very kernel of the operateing system, to tell the hardware what to do. As a rule of thumb, the higher the resources (hardware specs and capabilities) the better performance you get in a given application. The only time when this is untrue is in cases where people have core, shader and memory clocks on a graphics card, rather than just core and memory. Overclocking via software dynamically raises the shader clock as the core clock is raised. However, some shader to memory frequencies will be detrimental to performance if the shader clock itself is unable to operate at the correct syncronisation.

hmmm I sorta see but alright.

Just like I said it was just an idea I had.

I have no proof of anything

Also I know its not common.

But I was suggesting a possible situation

 

[jonny1989]

well 06 is a year or two old now and where benching card that came out this year on it,I think that you will see better socres with more tests and test options on 07,And by the way when does 07 come out?

 

[MrBungle]

Quote:

Originally Posted by Mhill2029 If you get higher scores in 3DMark06, you'll have higher frame rates in games. I trust the results on it myself....

the idea came to me while i was setting up my new CrossFire tonight. i realized that i now have 1 Gigabyte of VRAM. this gives me no advantage whatsoever on 3DMark, but i can now run games on any size monitor i please.

then i put FEAR in to give it a run. 1280x1024 rez, most settings maxed except only 4xAA and 4xAF with soft shadows turned off. i was averaging about 100 fps in the game. then i switched to my friend's 22" monitor, set the game at 1680x1050, 6xSSAA, 16xHQAF, soft shadows on and all other settings maxed. the result? the exact same framerate i got with lower rez and lower settings. i can't wait to try Stalker and all my demanding games now.

my 3DMark06 score was about 60% higher on my first CF run, but the score was only higher because i got more frames per second. it doesn't take into account the fact that i can now power a 60" high def TV. it doesn't know that i have 96 shaders or that i can play with SSAA and HQAF at crazy high levels without affecting framerates by much at all.

anyway, i broke 10k my first time out with no OC

. not bad for a couple of year-old video cards and a budget processor.

 

[Special_K]

07 should come out by the end of this year.

My 2 cents to this is kind of like how games should put specs for systems on the back of their boxes

Min, recommended, and 60 fps no lag.

Anyone else agree?

 

[iampoor]

Quote:

Originally Posted by Special_K 07 should come out by the end of this year. My 2 cents to this is kind of like how games should put specs for systems on the back of their boxes Min, recommended, and 60 fps no lag. Anyone else agree?

I think thats a good idea

However

1. it would cause prolems suppose someone as stuff loading down there system so don't get that amount

2. it depends on where you are in the level (eg details...)

 

[Fusion Racing]

Quote:

Originally Posted by iampoor hmmm I sorta see but alright. Just like I said it was just an idea I had. I have no proof of anything Also I know its not common. But I was suggesting a possible situation

good idea though... even if it wasn't true - it was good

i find the problem with 3dmark is that i can run 500/1180 on that, but in any game that will just artifact like hell and crash - it means it doesn't really reflect the real in game performance

the benchmark should run at 1600x1200 or 1680x1050 for widescreen, with full AA and AF on, then it would be closer to game performance

 

[MrBungle]

Quote:

Originally Posted by Special_K Min, recommended, and 60 fps no lag. Anyone else agree?

ha, wouldn't that be nice? the way i always look at the boxes - i ignore Minimum Specs altogether and treat Recommended Specs as the real minimum. that's the way it seems to me.

 

[Fusion Racing]

Quote:

Originally Posted by MrBungle ha, wouldn't that be nice? the way i always look at the boxes - i ignore Minimum Specs altogether and treat Recommended Specs as the real minimum. that's the way it seems to me.

yeh that's exactly what i do... minimum its pointless to look at - because there is always the chance it will run, but run so crap you cant really play it.

 

[Puckbandit35]

i totally agree on the sea monster part...

and how come there is really no huge splash when he goes back into the water?????????

PHAIL

 

[trueg50]

Also, Nvidia and ATI optimize their drivers for 3DMark, just look at the 8600's, and also the 7900's when they came out, they did amazing in 3DMark's. However the 8600's especially just don't deliver when it comes to what matters; frames per second.

 

[MrBungle]

the dragon totally phailed. that's why we don't have any dragons anymore.

 

[Kopi]

Quote:

Originally Posted by MrBungle 3) the sea monster is an idiot. i mean come on, all it had to do was jump up from directly beneath the airship and knock it out of the sky. then it could have picked the people out of the water one at a time and enjoyed a tasty meal. instead, it goes around trying to show off, flying over the airship on its first pass. the sea monster got what it deserved.

ahahah that made me laugh preeetttyyyy good.

 

[PhillyOverclocker]

I get low scores on 3DMark '06 even though I can max out CSS and BF2142. I never lag or stutter unless it's a server on the other side of the globe, and that's unavoidable with cable. I can't wait for FIOS!

 

[Unknownm]

Quote:

Originally Posted by MrBungle Why 3DMark06 Shouldn't Be Taken Too Seriously by Mr. Bungle 1) quad cores get a huge points advantage in 3DMark, but not in any game out today. 2) the stock benchmarks runs with no anti aliasing and minimal texture filtering. the true test of a gaming system isn't how many frames per second you can get at minimum settings; but rather how high you can turn up the settings while maintaining playable frames. 3) the sea monster is an idiot. i mean come on, all it had to do was jump up from directly beneath the airship and knock it out of the sky. then it could have picked the people out of the water one at a time and enjoyed a tasty meal. instead, it goes around trying to show off, flying over the airship on its first pass. the sea monster got what it deserved. feel free to add to my list if you want. there may be some things i missed.

1) I disagree. Quad cores gets higher points because 3dmark06 knows there a quad core and takes use of them. Some games don't / can't use quad core because they where programmed before quads came out

2) I'm 50%/50%. Futuremark only did 4x AF because it's even for all video cards that are DX9 (Geforce FX, 6, 7, 8 and Radeon X1300 +)

3) I disagree. That's not a point of "Why 3DMark06 Shouldn't Be Taken Too Seriously". If your mad about that benchmark email futuremark