FutureMark 3DMark06 Benchmark Overview [Now with Download Link]

[MBrown]

Originally posted by: DAPUNISHER

Originally posted by: gac009
1951 with the rig in the sig.

SM 2.0 990
CPU 1095

I know its not a top of the line pooter, and it never was, but I thought some people would be interested in seeing how a bottom middle range rig would do.

Do you have hyperthreading turned off?

 

[Munky]

Originally posted by: Steelski

Originally posted by: Steelski

Originally posted by: Matt2

Originally posted by: fierydemise

Originally posted by: Matt2
Nice to know that I'll be scoring higher than X1800XT CF

actually the X1800XT crossfire scores a 5815 and 7800GT SLI scores 5732.

I'm pretty sure I can squeeze out 100 points or so I doubt they even OCd those cards so I'll be breaking 6k for sure. And yes, I know the X1800XT CF can be overclocked too. Dont sink my ship!

So what's the deal with ATI's scores? They seem to be a bit... low...

You have to remember that the big big BIG thing for ATI is AA and AF in these kinds of tests. I would like to see a GTX beating them in a AA+AF enviroment

Also the Nvidia drivers are newer than the 5.13's from ATI , come to think of it. there should be a new driver out very soon.

I just saw the Firingsquad one with the AA on, I am eating my words a little. But i am still sure a driver revision will help matters a lot.

Oh, drivers are always getting tweaked for 3dmark. I expect to see new Ati drivers next week with optimizations for 3dmark06. Maybe that programmable mem controller will do the trick?

 

[jEnus]

At work, we just got out 1900xt master card, already have the 1900 slave. Maybe I'll run this and post the scores.

 

[Steelski]

Originally posted by: jEnus
At work, we just got out 1900xt master card, already have the 1900 slave. Maybe I'll run this and post the scores.

Maybee, What do you mean Maybee......
Do it....Do it NOW

 

[Elfear]

Originally posted by: Steelski

Originally posted by: jEnus
At work, we just got out 1900xt master card, already have the 1900 slave. Maybe I'll run this and post the scores.

Maybee, What do you mean Maybee......
Do it....Do it NOW

Yes, yes! (he says in his Igor voice).

 

[xtx4u]

Originally posted by: Rage187

Originally posted by: Frostwake
I guess Ati is getting low scores because it was tested without AA... and we all know nvidia > ati without AA so.. lets wait for some AA results and check it after

yes, lets keep changing the default test until ATi wins and then it's a valid test.

so u don't play games with AA and AF???

 

[Ackmed]

Originally posted by: munky

Oh, drivers are always getting tweaked for 3dmark. I expect to see new Ati drivers next week with optimizations for 3dmark06. Maybe that programmable mem controller will do the trick?

6.1's just came out. With a small boost from what Ive read.

 

[Munky]

Originally posted by: Ackmed

Originally posted by: munky

Oh, drivers are always getting tweaked for 3dmark. I expect to see new Ati drivers next week with optimizations for 3dmark06. Maybe that programmable mem controller will do the trick?

6.1's just came out. With a small boost from what Ive read.

I guess that shows how often I check for updates...

 

[Steelski]

6.1's out...This means war.
Hilarious.

 

[xtx4u]

Originally posted by: Ackmed

Originally posted by: keysplayr2003

Originally posted by: Ackmed

Originally posted by: Rage187

Originally posted by: Frostwake
I guess Ati is getting low scores because it was tested without AA... and we all know nvidia > ati without AA so.. lets wait for some AA results and check it after

yes, lets keep changing the default test until ATi wins and then it's a valid test.

When synthetic tests do not have the same results as real games, its a problem with the benchmark. Also, who buys a high end card, such as a X1800 or GTX and doesnt use AA/AF? Honestly. Most reviews of high end cards, dont show low resolutions, or tests without AA/AF for a reason.

I dont have a problem with synthetic benches in themselves, but as I said, when results dont match real world gaming, I dont see how anyone can count them as valid. Just as Sisoft Sandra shows a huge (about 1000 point or about 20%) increase from 2T to 1T, games show nothing, but perhaps a 1-2fps gain. While I do like the graphics in 3dmarks, I just dont see how anyone can take the results over real games.

If the ATI card did better, you would not be making a peep. You're unhappy with the results, so automatically there must be something wrong with the testing methods, without a doubt. ATI may have a better AA implementation above 4x, but everyone runs 3dmarkXX with default setting for a good baseline to compare to. Don't worry, with a few optimizations on ATI's part, I'm sure they'll be ahead.

Um, ATi isnt ahead with 2x or 4xAA, with 8xAF. The 256MB GTX is faster in every test, except 2 I think, and thats at 1600x1200 and less than a FPS difference. Which goes against the vast majority of video cards reviews, when they use games. Making it an pretty unvalid "test". It doesnt reflect real world performance.

I would be making "a peep" if the 512MB GTX scored less than a X1800XT too. Because that too would not match real world performance with games. Dont assume you know everything, or what I would/wouldnt do, because you dont.

As for your request for a download link, this is what I used; http://www.bjorn3d.com/download/index.php?dlid=2 Took a while to get it to actually download, but when it did, I was at 750k all the way.

Originally posted by: otispunkmeyer

Originally posted by: BroadbandGamer
Yeah, turn on 4x AA and I'll bet the X1800 XT wins.

sure

but then NV cant run the HDR. ATI can do HDR and AA, NV can do HDR or AA.

so if they put AA on.....does that mean HDR wont run? or the bench will skip those tests with HDR?

NV can do HDR+AA, just look at HL2 for proof of this. It just cant do the type of HDR+AA that Farcry uses, when ATi can.

oh boy... dont u mean Valve can make NV do HDR + AA? and Farcrys real hdr looks way better than Valve's wannabe HDR which took them to forever develope cause they can't get their lazy ass to work any faster, just as they delayed HL2's release.

 

[Gamingphreek]

Everyone seems to be blaming ATI's seemingly lower scores on lack of AA.

Sorry but i think it is because:

Also, while both NV and ATI hardware can do floating point blends, only NVIDIA hardware can perform hardware filtering on floating point render targets.

Which probably means ATI is using some other way to filter the FP Blending. I would venture to think that is why they get killed on the HDR/SM3 benches. As for the other ones, i suppose AA/AF may play a role in it, but it seems to me that every test is already so graphically intensive that doing that would kill performance for all cards, but im not sure as i dont own an X1xxx series card.

-Kevin

 

[thilanliyan]

1994 with the rig in the sig. I think the ram is hurting it a bit. Still, seems a bit low to me even without the SM3 tests.

 

[professor1942]

I get a whopping 1408

 

[Xed]

4290 with my rig in sig (only a single 7800gtx though, sli is disabled atm)

 

[Gamingphreek]

Originally posted by: xtx4u

Originally posted by: Ackmed

Originally posted by: keysplayr2003

Originally posted by: Ackmed

Originally posted by: Rage187

Originally posted by: Frostwake
I guess Ati is getting low scores because it was tested without AA... and we all know nvidia > ati without AA so.. lets wait for some AA results and check it after

yes, lets keep changing the default test until ATi wins and then it's a valid test.

When synthetic tests do not have the same results as real games, its a problem with the benchmark. Also, who buys a high end card, such as a X1800 or GTX and doesnt use AA/AF? Honestly. Most reviews of high end cards, dont show low resolutions, or tests without AA/AF for a reason.

I dont have a problem with synthetic benches in themselves, but as I said, when results dont match real world gaming, I dont see how anyone can count them as valid. Just as Sisoft Sandra shows a huge (about 1000 point or about 20%) increase from 2T to 1T, games show nothing, but perhaps a 1-2fps gain. While I do like the graphics in 3dmarks, I just dont see how anyone can take the results over real games.

If the ATI card did better, you would not be making a peep. You're unhappy with the results, so automatically there must be something wrong with the testing methods, without a doubt. ATI may have a better AA implementation above 4x, but everyone runs 3dmarkXX with default setting for a good baseline to compare to. Don't worry, with a few optimizations on ATI's part, I'm sure they'll be ahead.

Um, ATi isnt ahead with 2x or 4xAA, with 8xAF. The 256MB GTX is faster in every test, except 2 I think, and thats at 1600x1200 and less than a FPS difference. Which goes against the vast majority of video cards reviews, when they use games. Making it an pretty unvalid "test". It doesnt reflect real world performance.

I would be making "a peep" if the 512MB GTX scored less than a X1800XT too. Because that too would not match real world performance with games. Dont assume you know everything, or what I would/wouldnt do, because you dont.

As for your request for a download link, this is what I used; http://www.bjorn3d.com/download/index.php?dlid=2 Took a while to get it to actually download, but when it did, I was at 750k all the way.

Originally posted by: otispunkmeyer

Originally posted by: BroadbandGamer
Yeah, turn on 4x AA and I'll bet the X1800 XT wins.

sure

but then NV cant run the HDR. ATI can do HDR and AA, NV can do HDR or AA.

so if they put AA on.....does that mean HDR wont run? or the bench will skip those tests with HDR?

NV can do HDR+AA, just look at HL2 for proof of this. It just cant do the type of HDR+AA that Farcry uses, when ATi can.

oh boy... dont u mean Valve can make NV do HDR + AA? and Farcrys real hdr looks way better than Valve's wannabe HDR which took them to forever develope cause they can't get their lazy ass to work any faster, just as they delayed HL2's release.

Any card can do the HDR + AA that Valve used as they are rendering the HDR IIRC through the pixel shaders.

Far Cry uses OpenEXR HDR which requires a 32bit FP Blending to render HDR. It, arguably, looks better, but comes at a much larger performance hit. Ati can do this not because of superior hardware, but because they decided to implement the necessary core logic to allow the AA of FP blends.

If you notice, almost none of the cards save for CF systems can run this anyways.

Neither company has a real advantage here but i do applaud ATI for implementing the features to allow the user to decide.

-Kevin

 

[DidlySquat]

I submit a motion to ban people who don't have at least ATI X-series or Nvidia 6-series cards or better.

 

[Matt2]

Tminue -5 minutes until first '06 run

 

[beggerking]

Why does it have to be the same as TRAA? TRAA and AAA are both methods of applying AA to transparency textures, and they both work. Ati and NV dont use the exact same techniques for anything, so even stuff like SM3 branching, shader optimizations, AA sampling patters, AF algorithms, color compression, hidden surface removal - all those are different beween Ati and NV. In fact, I cant really think of anything that's identical between Nv cards and their equivalent Ati cards, so expecting features to work exactly the same way is unrealistic, but comparing the image quality and performance of features that have the same goal is completely reasonable. IMO, most sites just dont test these high end features because they're lazy - just like most sites dont test image quality either.

exactly. 7800 also supports HDR and AA, but their results doesn't show that

correct, unbiased way of reporting a invalid comparison

We can't run the test with anti-aliasing and get scores for both cards to compare, since Nvidia's cards don't generate a score.

its unprofessional/biased to just post a 0 score on SM3.0 for 7800 as the FS review.

 

[blckgrffn]

Well, darn the higher default res. My projector can only scale down from 1024*768

Guess I'll just to wait a few days for the pro features to magically appear

Nat

 

[Munky]

Originally posted by: beggerking
correct, unbiased way of reporting a invalid comparison

We can't run the test with anti-aliasing and get scores for both cards to compare, since Nvidia's cards don't generate a score.

its unprofessional/biased to just post a 0 score on SM3.0 for 7800 as the FS review.

So should the reviewers have refused to run games in SM3 last year because the x800 series did not support it? The fact that the hardware does not support a feature should be noted when reviewing it. Maybe it would have been more "politically correct" to just exclude Nv totally when running the HDR+AA tests, but that does not change the fact that Nv card can not do FP HDR with AA, while the competitor's cards can.

 

[xtknight]

Holy slow downloads batman.

 

[xelpmoc]

6971 with my X2 at 2.6 and dual 7800 GTs at 450/1150.

My processor will do 2.8 with more voltage and I know the video cards will go a bit higher, so topping 7000 is no problem.

 

[Matt2]

6214

Good for first on the ORB.

Although xelpmoc raped me. Time to get dual core I guess.

 

[otispunkmeyer]

Originally posted by: Ackmed

Originally posted by: keysplayr2003

Originally posted by: Ackmed

Originally posted by: Rage187

Originally posted by: Frostwake
I guess Ati is getting low scores because it was tested without AA... and we all know nvidia > ati without AA so.. lets wait for some AA results and check it after

yes, lets keep changing the default test until ATi wins and then it's a valid test.

When synthetic tests do not have the same results as real games, its a problem with the benchmark. Also, who buys a high end card, such as a X1800 or GTX and doesnt use AA/AF? Honestly. Most reviews of high end cards, dont show low resolutions, or tests without AA/AF for a reason.

I dont have a problem with synthetic benches in themselves, but as I said, when results dont match real world gaming, I dont see how anyone can count them as valid. Just as Sisoft Sandra shows a huge (about 1000 point or about 20%) increase from 2T to 1T, games show nothing, but perhaps a 1-2fps gain. While I do like the graphics in 3dmarks, I just dont see how anyone can take the results over real games.

If the ATI card did better, you would not be making a peep. You're unhappy with the results, so automatically there must be something wrong with the testing methods, without a doubt. ATI may have a better AA implementation above 4x, but everyone runs 3dmarkXX with default setting for a good baseline to compare to. Don't worry, with a few optimizations on ATI's part, I'm sure they'll be ahead.

Um, ATi isnt ahead with 2x or 4xAA, with 8xAF. The 256MB GTX is faster in every test, except 2 I think, and thats at 1600x1200 and less than a FPS difference. Which goes against the vast majority of video cards reviews, when they use games. Making it an pretty unvalid "test". It doesnt reflect real world performance.

I would be making "a peep" if the 512MB GTX scored less than a X1800XT too. Because that too would not match real world performance with games. Dont assume you know everything, or what I would/wouldnt do, because you dont.

As for your request for a download link, this is what I used; http://www.bjorn3d.com/download/index.php?dlid=2 Took a while to get it to actually download, but when it did, I was at 750k all the way.

Originally posted by: otispunkmeyer

Originally posted by: BroadbandGamer
Yeah, turn on 4x AA and I'll bet the X1800 XT wins.

sure

but then NV cant run the HDR. ATI can do HDR and AA, NV can do HDR or AA.

so if they put AA on.....does that mean HDR wont run? or the bench will skip those tests with HDR?

NV can do HDR+AA, just look at HL2 for proof of this. It just cant do the type of HDR+AA that Farcry uses, when ATi can.

yeah thats valves HDR, it looks good but its not the proper way of doing it. kudos to them for making something everyone can use though, but its just not proper.

big thumbs up to ATI for implementing the hardware in their desgin to allow AA and proper HDR.

 

[BUrassler]

1508