Some Preliminary 3DMark06 results

Status
Not open for further replies.
H

HeavyH20

[H]ard|Gawd
Joined
Oct 28, 2004
Messages
1,180
These are some QA runs, but they give you a nice idea of how things will run. For more cards and results, type in numbers from 32 through 54 in the URL. Some will be invalid, but look for 6800 Ultra, x800XT PE and other results!

You get a score even with an SM 2.0 card, you just run two tests. And, better have a monitor that supports 1280 x x1024.


FX-60, x1800XT - full run:

http://service.futuremark.com/compare?3dm06=34

Here is a 7800 GTX 512 MB card run:

http://service.futuremark.com/compare?3dm06=50

And a 7800 GT run:

http://service.futuremark.com/compare?3dm06=52
 
Guess what though, in 3DMark06 the CPU result is now figured into the Overall score. So yep, different CPU's and speeds will effect the overall score.
 
Yup, and the tests feature heavy multi-threading.

EDIT: a more complete list

tyellbee said:
Radeon X1800 XT + AMD Athlon(tm) 64 X2 2614 MHz
3DMark Score 4256 3DMarks
SM 2.0 Score 1630 Marks
SM 3.0 Score 1703 Marks
CPU Score 1938 Marks

Radeon X1800 XL + AMD Athlon(tm) 64 X2 2615 MHz
3DMark Score 3452 3DMarks
SM 2.0 Score 1296 Marks
SM 3.0 Score 1333 Marks
CPU Score 1938 Marks

Radeon X1300 Pro + AMD Athlon(tm) 64 X2 2610 MHz
3DMark Score 1322 3DMarks
SM 2.0 Score 452 Marks
SM 3.0 Score 485 Marks
CPU Score 1939 Marks

Radeon X850 XT PE + AMD Athlon(tm) 64 X2 2609 MHz
3DMark Score 2368 3DMarks
SM 2.0 Score 1158 Marks
SM 3.0 Score N/A
CPU Score 1945 Marks

Radeon X800 XT PE + AMD Athlon(tm) 64 X2 2609 MHz
3DMark Score 2276 3DMarks
SM 2.0 Score 1110 Marks
SM 3.0 Score N/A
CPU Score 1934 Marks

Radeon X800 XT + AMD Athlon(tm) 64 X2 2610 MHz
3DMark Score 2156 3DMarks
SM 2.0 Score 1047 Marks
SM 3.0 Score N/A
CPU Score 1941 Marks

Radeon X800 XL + AMD Athlon(tm) 64 X2 2614 MHz
3DMark Score 1870 3DMarks
SM 2.0 Score 900 Marks
SM 3.0 Score N/A
CPU Score 1939 Marks

Radeon X800 GTO 256MB + AMD Athlon(tm) 64 X2 2610 MHz
3DMark Score 1514 3DMarks
SM 2.0 Score 720 Marks
SM 3.0 Score N/A
CPU Score 1941 Marks

RADEON X800 GTO 128MB + AMD Athlon(tm) XP/MP/4 2209 MHz
3DMark Score 844 3DMarks
SM 2.0 Score 409 Marks
SM 3.0 Score N/A
CPU Score 795 Marks

7800 GTX 512MB + AMD Athlon(tm) 64 X2 2621 MHz
3DMark Score 5552 3DMarks
SM 2.0 Score 2384 Marks
SM 3.0 Score 2343 Marks
CPU Score 1655 Marks

7800 GT + AMD Athlon(tm) 64 2016 MHz
3DMark Score 3236 3DMarks
SM 2.0 Score 1452 Marks
SM 3.0 Score 1457 Marks
CPU Score 796 Marks

GeForce 6800 Ultra + AMD Athlon(tm) 64 2409 MHz
3DMark Score 2442 3DMarks
SM 2.0 Score 1044 Marks
SM 3.0 Score 928 Marks
CPU Score 930 Marks

6800 GT + Intel Pentium 4 3005 MHz
3DMark Score 2178 3DMarks
SM 2.0 Score 960 Marks
SM 3.0 Score 893 Marks
CPU Score 651 Marks

6800 GS + AMD Athlon(tm) 64 1800 MHz
3DMark Score 1933 3DMarks
SM 2.0 Score 807 Marks
SM 3.0 Score 766 Marks
CPU Score 705 Marks

6600 GT 128MB + Intel Pentium 4 2668 MHz
3DMark Score 1116 3DMarks
SM 2.0 Score 548 Marks
SM 3.0 Score 292 Marks
CPU Score 690 Marks
Click to expand...
 
Brent_Justice said:
Guess what though, in 3DMark06 the CPU result is now figured into the Overall score. So yep, different CPU's and speeds will effect the overall score.
Click to expand...


Not sure how much, yet. An FX-60 with an x850 XTPE compares to a AMD 64 2.4 GHz with a 6800 Ultra.
 
Also note that the GT is not the overclocked 470/1100 version but at 430/1000! That gives me hope for my 500/1250 gt :) .
 
Just when we were approaching 20K again...they go and do this.

Nice scores btw...can't wait to try out the free version.
 
Note the updated list in my post - interesting results.

An X1300Pro with an AMD X2 CPU outscores a GeForce 6600GT with a Pentium 4.
 
dderidex said:
Note the updated list in my post - interesting results.

An X1300Pro with an AMD X2 CPU outscores a GeForce 6600GT with a Pentium 4.
Click to expand...

The FX-60 on that test would help out the x1300Pro. The 6600 GT is hampered by a single core P4. Not sure how much the CPU figures into the final score, but it now affects the overall result. I also would like to see how much SM 3.0 support helps scoring.
 
now you can't even compare the same cards in 3dmark, you also have to make sure the cpu is the same

factoring in the cpu and all now it really shows you how useless that score means, i mean, what does that number tell me about FEAR performance? absolutely nothing, sorry
 
Brent_Justice said:
now you can't even compare the same cards in 3dmark, you also have to make sure the cpu is the same

factoring in the cpu and all now it really shows you how useless that score means, i mean, what does that number tell me about FEAR performance? absolutely nothing, sorry
Click to expand...

Yeah, but who cares about FEAR performance? I don't, I never play the game.

In fact, I don't play ANY of the games you include in reviews (nor does almost the entire readership of SimHQ, in fact), so...those results are just as useless as 3dMark0# when you get down to it.

Unless you set out to benchmark *every single game* available, game benchmarks are exactly as 'useful' or 'useless' as 3dMark. They show you how a specific card performs in a specific situation that may or may not be completely meaningless depending on the reader. How well a card performs in FEAR, or Quake 4, or even FS2004 doesn't give ANY indication how it will perform in 'Silent Hunter III' or 'Lock On: Modern Air Combat' or 'Pacific Fighters'....so how are reviews using them any more useful to a simmer than 3dMark is?

In such context, all ANY of these tests or 3dMark scores show is relative position of one graphics card to another. If card A gets twice the score in 3dMark06 as card B...it's probably a better card. That doesn't show how it will perform in 'Dangerous Waters', though, does it? No more than knowing that card A gets 45.6 fps in FEAR at 1600x1200 with 4xAA and 16xAF and card B only get 30.0fps in FEAR at 1280x1024 with 4xAA and 16xAF. Both pieces of information are equally 'useless' in this case for providing SPECIFIC data, but equally 'useful' in providing RELATIVE data, so why not just use the quicker (and better looking) test to get said information?

I'm sure this argument will be re-treaded a million times in the coming days, and nobody will ever change their position on it either way, so...I dunno, seem a waste of bandwidth to have again.
 
The CPU can be taken out as a factor by using the same zero base system for all video card reviews. So, as of Jan 1, I would say it is an FX-60 on a particular mobo with 2 GB of DDR. No biggie, really. For reviews on card potentials, it keeps it apples to caramel apples, anyway.

The CPU is a key component when playing most games at higher resolutions. So, it should be a factor. If you want to measure actual game potential perfromance, the CPU is has to be considered. Why else would we want an FX-60 for gaming? If the bench simply measured video card performancelike 05, then it really does not reflect real world gaming experience. They have moved in the right direction.

So, and FX-60 system with a single 7800 GTX may game better than a AMD 3000 with 7800 GTX in SLI. 3Dmark05 would not reflect this, but real world game tests would. And, now, 3DMark06 will test this factor. I think this is a good development.
 
Radeon X1300 Pro + AMD Athlon(tm) 64 X2 2610 MHz
3DMark Score 1322 3DMarks
SM 2.0 Score 452 Marks
SM 3.0 Score 485 Marks
CPU Score 1939 Marks

Radeon X800 XL + AMD Athlon(tm) 64 X2 2614 MHz
3DMark Score 1870 3DMarks
SM 2.0 Score 900 Marks
SM 3.0 Score N/A
CPU Score 1939 Marks

548 point difference between an x800xl, and a x1300pro.

WHAT?
 
TruthInRuin said:
548 point difference between an x800xl, and a x1300pro.

WHAT?
Click to expand...
And yet another realization of what a worthless benchmark 3DMark is. :D
 
pxc said:
And yet another realization of what a worthless benchmark 3DMark is. :D
Click to expand...
No, it's really pointing out the shortfall of the X800XL (which still wins) for newer titles.

Notice the X800XL has double the SM2.0 score, but cannot run the SM3.0 tests at all? That's where the disparity is coming from.
 
I have a 7800GTX 512 @ 620/1.85 regular use. I'm insterested in seeing what I'll pull. Looks like around 6000 or higher going from that one benchmark.
 
SM 3.0 will be a required component to play the upcoming HDR based games as designed, so cards without SM 3.0 are penalized. None of the x800, x850 based cards will do well on this test. Again, a step in the right direction to mirror true gameplay potential.
 
Anyone notice all results were at 1280/1024? I wonder if this is the new default resolution.
 
Brent_Justice said:
new colors, textures and shadows to already existing tests, and added 1 new graphics test, i'm not impressed
Click to expand...
I dunno, those are pretty HUGE changes to the shadow system used. Quite an improvement!
 
Brent_Justice said:
and that means what?
Click to expand...
The same thing "42.5 fps in FEAR" means to someone who does not have that game, nor has any intention of buying it.

Why do you ask?
 
The only thing I wonder, is that Windows Vista is supposed to be out at the end of the year, and its supposed to include DirectX 10 right? So why come out with a new benchmark its going to be outdated within the year?

They had a good spacing, 2001, 2003, 2005...then 2006 :confused:
 
dderidex said:
The same thing "42.5 fps in FEAR" means to someone who does not have that game, nor has any intention of buying it.

Why do you ask?
Click to expand...

A x800xl would woop an x1300pro in FEAR right? Yet it scores pretty close to it in 3dmark06 simply because it has SM3. Theres your proof that FEAR would be a better judge of a cards performance, even if you don't have it.
 
With all do respect, gaming benchmark's, such as you state "42.5 fps in FEAR", provide readers with a better understanding of how a card performs - in one instance.

When I'm looking to buy a video card, it's for gaming, period the end. I don't spend hundreds of dollars, to get some numeric score in a benchmark program. I simply want a card that'll play todays games at high settings.

I'm not saying 3DMark# is useless, as I douse it for a couple of reasons, but when 3DMark# is dependent upon your processor . . . that's getting carried away. They have PCMark# to do this . . . why implement it on the 3DMark benchmark? I understand that your processor could bottleneck your GPU, etc . . . but let the GPU tests show that. I don't need 3DMark saying my processor hits above 1,000+ points when I have PCMark for that purpose.

My feeling is 3DMark06 is now going to be solely used to see if a system is performing properly, rather than if their graphic card is "good enough". It'll become a System benchmark, rather than a graphics card benchmark. I'm assuming users will fall back on '05 for that matter (graphics bench), along with AM3. They no longer can use 3DMark06 to compare cards against cards - Unless they focus in solely on SM2.0 or SM3.0 score - but that's just tedious.

Like I said, I'm not trying to be rude, but rather trying to explain why a gaming benchmark IS important. Remember, some, if not most, people buy video cards to play games, and some want cards that will be able to handle the game at max settings and a high res. When a benchmark gives us the FPS for the particular game, it helps us a whole lot more than "4,000 3DMarks". We understand that a game my not be "playable" when the card is only delivering 36fps on certain settings.

On the contrary, entusiasts who build systems for benching and stability, well then programs such as 3dMark are more useful than a gaming benchmark. It allows them to decide if their graphic card, processor, or RAM, meets todays "Top-of-the-line" list. It all boils down to what the user wants.

I hope I made my point clear rather than being an arrogant ass.

Cheers.

Note: I know 3DMark06 isn't only dependent on the processor - it was just an example I used.
 
Shadow27 said:
A x800xl would woop an x1300pro in FEAR right? Yet it scores pretty close to it in 3dmark06 simply because it has SM3. Theres your proof that FEAR would be a better judge of a cards performance, even if you don't have it.
Click to expand...

The bench tries to predict upcoming peformance in games, not all current games. If you want to compare FEAR results, look at the SM 2.0 support results alone. The x800XL is twice as fast (assuming the same CPU). Right on the money.

Now, if there is a new game coming out with SM 3.0 and HDR, how much faster will the x800XL be than the x1300 Pro? Well, the x800XL will not play the game in native mode. The score handicaps the cards with no SM 3.0 support. Simple, but effective concept on the scoring algorithm.

Prior critisms of benchmarks claimed they did a poor job reflecting true gaming performance. Have to agree with that. If it is a simple video benchmark, then all you are measuring is how fast a video card can push triangles, etc. If you make it CPU dependent, then it becomes more of a gaming benchmark. The CPU can cannot be ignored when building a gaming system. If anything, 3DMark06 is more relevant to gaming than previous benchmark programs.

So, an FX-60 equipped system with a single 7800 GTX 512 MB card may be a better gaming choice than a AMD 64 3200+ with 7800 GTX cards in SLI. In 3DMark05, this would not be reflected in the score, but in real world gaming, the FX-60 system might be faster. We will see how 06 pans out.
 
jnick said:
With all do respect, gaming benchmark's, such as you state "42.5 fps in FEAR", provide readers with a better understanding of how a card performs - in one instance.
Click to expand...

Exactly - and that's why using FEAR fails in this case!

Sitting here right now, if I wanted to know the BEST video card to buy for 'TESIV: Oblivion'...what can tell me that? What "game test" will give me a score that maps out EXACTLY to the performance in that game?

That's right - none will! So why does it matter if you use FEAR to get a relative idea of how cards stack up vs one another or 3dMark06 instead?

Indeed for 'Oblivion', that *does* heavily use SM3.0 and is HEAVILY multithreaded, 3dMark06 will probably better represent how systems will stack up vs one another than FEAR will, which is a single-threaded SM2.0 game.

Again, if all you care about is playing FEAR, and you bought your $x thousand PC *JUST* to play FEAR...then by all means, just look at FEAR benchmarks. Or, if you bought your PC to play FEAR, Quake 4, Doom 3, Sims 2, etc....then sure, benchmarks in those games will help you.

What if you bought your PC to play 'Pacific Fighters', 'Dangerous Waters', 'Silent Hunter III', 'TESIV: Oblivion'? Then you could look at [H]'s "in-game experience" reviews all you want and they will TELL YOU NOTHING MORE THAN 3dMARK WILL!!

Indeed, they may well be misleading. As Shadow27 pointed out (although he didn't mean to), the X800XL whoops the X1300Pro in FEAR. Great. The X800XL must be a MUCH better buy for 'Oblivion' then, right? Except, you won't get any of the SM3.0 effects then, but, oh well, right? And hey, the X800XL outperforms the X1300Pro in FEAR, so it must ALSO be the better card to get for AoE3, right? Only, it looks like a totally different game when run in SM2.0 mode instead of SM3.0.


vs
 
dderidex said:
And hey, the X800XL outperforms the X1300Pro in FEAR, so it must ALSO be the better card to get for AoE3, right? Only, it looks like a totally different game when run in SM2.0 mode instead of SM3.0.


vs
Click to expand...

Yeah and we all know the x1300Pro has huge power backing its sm3 capability , right ?

Nothing better than playing AoE 3 at 6/9.3 min/max FPS without AA/AF @ 1280x960. (source: http://www.xbitlabs.com/articles/video/display/games-2005_5.html )

Support for eyecandy means exactly nothing if the card isn't fast enough to actually put out a reasonable amount of FPS with eyecandy enabled.
 
Danylo said:
Support for eyecandy means exactly nothing if the card isn't fast enough to actually put out a reasonable amount of FPS with eyecandy enabled.
Click to expand...
There is a known bug in AoE3 with the drivers used in that test. Either Cat 6.1 or 6.2 will be bringing a fix to the table that will improve X1K-series cards numbers in that game.

Besides, who in their right mind buys a <$100 card and expects to run at 1280x960?

Nonono, try 800x600 or 1024x768 at the most. And it'll still look good with all the effects the game has to offer!
 
Who in their right mind buys a <$100 card and expects to run SM3 ? Besides 8x6 with or without sm3 is looking like crap.

And is there even any reliable source for this "mysterious" driver bug?
 
Danylo said:
Who in their right mind buys a <$100 card and expects to run SM3 ? Besides 8x6 with or without sm3 is looking like crap.
Click to expand...
Ummm...resolution matters a LOT more than shader version.

A < $100 card is going to be playing AoE3 at 800x600 or 1024x768...PERIOD. Doesn't matter what settings you have in-game. So...might as well enjoy SM3 if you are going to play at that resolution.

Danylo said:
And is there even any reliable source for this "mysterious" driver bug?
Click to expand...
Sure - do a search on Rage3d or Beyond3d. Don't recall if it was Sireric or Catalyst Maker who commented on it, but it's confirmed that 6.1 or 6.2 will have a fix for it.
----------
ANYWAY, back on topic: site with approximately one bajillion pics of 3dMark06

Those shots are *something*. Really gets me - every time a new 3dMark comes out, you inevitably look at the pics and think "wow...now THAT is what next-gen games should look like - I can't understand HOW we could have thought the last 3dMark looked 'next-gen'!"

Every

Friggin

Time
 
I don't think 3DMark represents what any current game or any future game will look or perform like.
 
Sorry for going OT one last time but :


dderidex said:
Ummm...resolution matters a LOT more than shader version.
Click to expand...
Did I say something different ? Your point was that the x1300 is superior for SM3 games like AoE3, because of more eyecandy - Whats the point of having to drop to 8x6 without AA/AF which looks like crap (as stated above) just to be able to turn on some additional SM3 effects ? Where did the eyecandy go ;) ? SM3 effects are nice if used with a fast card that can actually make use of them without turning down _every_ other option to raise image quality , but they're certainly no selling point for some lowcost video card like the x1300.

A < $100 card is going to be playing AoE3 at 800x600 or 1024x768...PERIOD. Doesn't matter what settings you have in-game.
Click to expand...
A card like the x800gt for instance is capable of playing at higher res though. http://www.newegg.com/Product/Product.asp?Item=N82E16814102573 - 99$ after MIR.
Sure - do a search on Rage3d or Beyond3d. Don't recall if it was Sireric or Catalyst Maker who commented on it, but it's confirmed that 6.1 or 6.2 will have a fix for it.
Click to expand...
YOU have to prove your point, not the one you're arguing with.
 
Brent_Justice said:
I don't think 3DMark represents what any current game or any future game will look or perform like.
Click to expand...

Quoted for truth.

I'm curious though if they're going to call it "The gamers benchmark" again. As it looks now - because of the way SMP & SM3 etc. is factored in the end result, the actual SCORE will have just about nothing to do with the performance of already released games and/or any games released in the near future.
 
Status
Not open for further replies.
Back
Top