[H] Battlefield 3 Open Beta Performance and Image Quality

[RobertR1]

Here my benchmarks compared to HARDOCP:

PLAYABLE SETTINGS AS PER HARDOCP FOR 2560x1600. GTX 580 OC'D TO 950/1900/2225. Intel 2600k @ 4.6ghz.

The benchmark were conducted defending the first M-COM stations in the same game. The server was full and game was action packed. All the action takes place top side which is the by the far most taxing part of the level:

Run 1:
Frames, Time (ms), Min, Max, Avg
6540, 120000, 43, 70, 54.500

Run 2:
Frames, Time (ms), Min, Max, Avg
6644, 120000, 43, 72, 55.367



SAME SETTINGS AS HARDOCP BUT WITH HBAO/SSAO DISABLED (restarted the game to ensure settings took and waiting for the same game scenario as above)

Run 1:
Frames, Time (ms), Min, Max, Avg
8016, 120000, 54, 88, 66.800

Frames, Time (ms), Min, Max, Avg
8104, 120000, 51, 93, 67.533

 

[happy medium]

Here my benchmarks compared to HARDOCP:

PLAYABLE SETTINGS AS PER HARDOCP FOR 2560x1600. GTX 580 OC'D TO 950/1900/2225. Intel 2600k @ 4.6ghz.

The benchmark were conducted defending the first M-COM stations in the same game. The server was full and game was action packed. All the action takes place top side which is the by the far most taxing part of the level:

Run 1:
Frames, Time (ms), Min, Max, Avg
6540, 120000, 43, 70, 54.500

Run 2:
Frames, Time (ms), Min, Max, Avg
6644, 120000, 43, 72, 55.367



SAME SETTINGS AS HARDOCP BUT WITH HBAO/SSAO DISABLED (restarted the game to ensure settings took and waiting for the same game scenario as above)

Run 1:
Frames, Time (ms), Min, Max, Avg
8016, 120000, 54, 88, 66.800

Frames, Time (ms), Min, Max, Avg
8104, 120000, 51, 93, 67.533

Now try it with no overclock on your gtx580.
Do you run out of memory @ 2500x1600 4xaa max settings?

 

[Silverforce11]

^ 2600K @ 4.6ghz is a bit different to i920 ish @ 3.6ghz. Also factor that in, and the game is seriously multithreaded, user commented it has 7 active threads: http://hardforum.com/showpost.php?p=1037829118&postcount=10

Without the OC on the gtx580, i suspect fps will be around the 42 mark.

 

[toyota]

^ 2600K @ 4.6ghz is a bit different to i920 ish @ 3.6ghz. Also factor that in, and the game is seriously multithreaded, user commented it has 7 active threads: http://hardforum.com/showpost.php?p=1037829118&postcount=10

Without the OC on the gtx580, i suspect fps will be around the 42 mark.

it may use many threads but all it needs is 4. even the i3 with HT is not getting pushed.

 

[Grooveriding]

Here my benchmarks compared to HARDOCP:

PLAYABLE SETTINGS AS PER HARDOCP FOR 2560x1600. GTX 580 OC'D TO 950/1900/2225. Intel 2600k @ 4.6ghz.

The benchmark were conducted defending the first M-COM stations in the same game. The server was full and game was action packed. All the action takes place top side which is the by the far most taxing part of the level:

Run 1:
Frames, Time (ms), Min, Max, Avg
6540, 120000, 43, 70, 54.500

Run 2:
Frames, Time (ms), Min, Max, Avg
6644, 120000, 43, 72, 55.367

Nice. Hard's card is not overclocked though. Although I don't think you'd lose 15 frames over a 180MhZ overclock, maybe 10.

Thanks for the numbers.

 

[AtenRa]

My system at 2.8GHz(133x 21), single HD6950 1GB at 810MHz default, Catalyst 11.10

1920x1200 ultra, 2x MSAA, Post OFF, Blur OFF, Blur Amount 0, AF 16x, HBAO ON

Is playable with 31fps Minimum and 39fps Average (Open map).

With 4x MSAA and Blur ON, minimum fps dives down to 10-12fps and its unplayable.

 

[Grooveriding]

My system at 2.8GHz(133x 21), single HD6950 1GB at 810MHz default, Catalyst 11.10

1920x1200 ultra, 2x MSAA, Post OFF, Blur OFF, Blur Amount 0, AF 16x, HBAO ON

Is playable with 31fps Minimum and 39fps Average (Open map).

With 4x MSAA and Blur ON, minimum fps dives down to 10-12fps and its unplayable.

:thumbsup:

 

[lehtv]

Does anyone else think that post-processing AA looks absolutely terrible? It's a downgrade in image quality at the cost of FPS ;<

 

[Grooveriding]

Does anyone else think that post-processing AA looks absolutely terrible? It's a downgrade in image quality at the cost of FPS ;<

I can't stand it. It blurs textures and does indeed reduce IQ. I don't like it in Red Orchestra 2 either, I didn't find it so bad in DE:HR though.

 

[houe]

If you could be so kind as to give us some high 1920x1080 and ultra 1920x1080 numbers that would be nice of you Single display of course i'm not as rich as you

Appears a 2gb 6970 could be in my future but a 3gb gtx580 has crossed my mind.

Would be happy playing 1920x1080 on high with no aa of any kind just like i do in BC2 .

I don't care how pretty that bush over there /\is in ultra while some camping f*ck is blasting me from over in that bush>>

Only good thing about Ultra is how much money you can throw at it,i'm not legally blind there isn't worlds difference for me to say oh yeah i need dual gtx580s for that i know when to quit :thumbsup:

Sure. I'll probably be doing a fair bit of testing when it comes. I'll post my results here.

 

[3DVagabond]

Hurray! I'm glad I'm not the only one that sees it.

2 conclusions

Hardocp preview sucks and you dont need more than 1gb for 1080p ultra 4xaa. The game somehow uses what it has avalable.

Discrepancies between 560ti's could be from different models. It's difficult with nVidia cards because they are offered at so many different clocks and reference designs are often the least represented model in online reviews. Everyone seems to have some O/C'd model that they were sent when the cards were released instead of a reference card. I don't know which models have been used in the benches, but I've seen sites use the 1GHz SOC and not note it in the charts before.

Besides, in my experience benches need to be compared across the web and, more or less, averaged. I don't know if it's differences in samples or in benching techniques, but results vary a lot. It's not uncommon. Either we have to think there are a lot of crooked sites, and not just the ones who's results don't make our favorite vendor look good, or we have to accept that it's not an exact science. I think that it's mostly the latter. Also you can't follow both vendors review guides. I've seen sites follow one or the others and, surprise surprise o), that vendor's cards look better.

 

[WMD]

Does anyone else think that post-processing AA looks absolutely terrible? It's a downgrade in image quality at the cost of FPS ;<

It looks fine. The blur is nowhere as noticeable as AMD driver MLAA. For midrange cards below gtx570 level using msaa is out of the question if you want 60fps performance.

 

[3DVagabond]

:thumbsup:

You can believe hardocp and the rest of us will believe the rest of the world.
Good luck with that.
Now back to our technical discussion.:thumbsup:

Don't use Fud to put someone else down. You do not speak for anyone else, never mind the rest of us. You are entitled to your opinion, but it is no more than that... your opinion.

I'm discussing the HardOCP evaluation and believe in their tests, the differences in frame-rate is not based on memory limitations. DO I believe this title can be memory limited? Absolutely, especially with surround.

I agree that there was no evidence of memory limitations in [H]'s review. It would be easy to see because they actually post the realtime graphs. A lot of people only want to see minimum/avg./max frame rates. Not very useful, IMO. Real time graphs tell the whole story so much better.

 

[3DVagabond]

It scales it based on distance and FoV. There is no LoD setting you can change.

Whats your take then? Why is it appearing to be using more vram the more you available, and less if you don't have it?

It's not using more VRAM, it's just holding it hostage.

 

[Silverforce11]

It's not using more VRAM, it's just holding it hostage.

Are you suggesting its the chuck norris of games?!

 

[Keysplayr]

I think that both are right, you can't use more memory than you don't have. If you have a game that's constantly pushing the RAM limits, means that the game is VRAM hungry. If you play a game on a GTX 560 Ti and uses 1011MB, and the same game uses nearly 1,4xx something on the GTX 480, means that the game is VRAM hungry. With Crysis 1 and several mods, AT 1280x1024 it used 1,2GB of my 2GB of VRAM, at 1920x1080, it increased to 1,5GB of my 2GB VRAM.

I believe the game just allocates all the available memory a video card has onboard, but I don't believe it actually uses it all. There would be some sort of evidence of performance degradation otherwise.

 

[blackened23]

Nice. Hard's card is not overclocked though. Although I don't think you'd lose 15 frames over a 180MhZ overclock, maybe 10.

Thanks for the numbers.

Its plausible. Nvidia hardware scales ridiculously well with overclocking, the performance jump on my 580 from stock (780ish) to 950 - MSI lightning 580 is tremendous. AMD doesn't scale nearly as well with OC'ing.

 

[RobertR1]

Nice. Hard's card is not overclocked though. Although I don't think you'd lose 15 frames over a 180MhZ overclock, maybe 10.

Thanks for the numbers.

It was my intention to show how much performance you gain with overclocking. Surprisingly, a lot. I might do additional testing tonight by turning off hyper threading and seeing what does, if anything.

 

[Grooveriding]

My system at 2.8GHz(133x 21), single HD6950 1GB at 810MHz default, Catalyst 11.10

1920x1200 ultra, 2x MSAA, Post OFF, Blur OFF, Blur Amount 0, AF 16x, HBAO ON

Is playable with 31fps Minimum and 39fps Average (Open map).

With 4x MSAA and Blur ON, minimum fps dives down to 10-12fps and its unplayable.

Well this is confirmation you can't expect to run this game on ultra with 4xAA on a 1GB card @ 1920x1200 & 1920x1080.

Fits in line with the benchmarks being seen of ultra settings with 4xAA not using 1GB cards.

 

[WMD]

My system at 2.8GHz(133x 21), single HD6950 1GB at 810MHz default, Catalyst 11.10

1920x1200 ultra, 2x MSAA, Post OFF, Blur OFF, Blur Amount 0, AF 16x, HBAO ON

Is playable with 31fps Minimum and 39fps Average (Open map).

With 4x MSAA and Blur ON, minimum fps dives down to 10-12fps and its unplayable.

Try turning off windows aero? It should free up enough to run 4xaa smoothly.

 

[TheAdvocate]

I am just not seeing the value of post processing AA. Especially on mid range cards where the peformance hit is more noticable.

Please feel free to convince me otherwise. Unless the retail release is vastly different, I'm fairly certain I'm going to play at 2xMSAA, w/ HBAO off. Seems to be my sweet spot.

 

[The Ultimate]

^ 2600K @ 4.6ghz is a bit different to i920 ish @ 3.6ghz. Also factor that in, and the game is seriously multithreaded, user commented it has 7 active threads: http://hardforum.com/showpost.php?p=1037829118&postcount=10

Without the OC on the gtx580, i suspect fps will be around the 42 mark.

AFAIK the game scales up to 4 threads, but Sandy Bridge has some serious gains in gaming performance compared to previous i7, let alone when overclocked. So it seems that HardOCP results aren't far from the truth regarding the GTX 580.

 

[imaheadcase]

Try turning off windows aero? It should free up enough to run 4xaa smoothly.

Aero turns off anyways when you enter full screen mode in games.

 

[AtenRa]

Battlefield 3 Beta Performance Test

Ehehe, i would say BD will be even faster than Phenom II X4 980.
It seams AMDs Gaming Evolved initiative helps their CPUs as well.

Dual cores (no HT) starting to be unsuitable in every new Game release i have seen so far.

 

[Skurge]

Battlefield 3 Beta Performance Test

Ehehe, i would say BD will be even faster than Phenom II X4 980.
It seams AMDs Gaming Evolved initiative helps their CPUs as well.

Dual cores (no HT) starting to be unsuitable in every new Game release i have seen so far.

Would be intersting to know why the AMD CPU's are doing so well. If gaming evolved as anything to do with it. Should we be excited that AMD can squeeze perfomace out of their CPU's by helping the devs or should be worried AMD is doing something sinister to the Intel CPUs