[H] Battlefield 3 Open Beta Performance and Image Quality

[tviceman]

Humm, it is possible to publish articles just to troll readers

"stands up and cheers this post" :thumbsup:

Whats the excuse now why Hardocp's benchmarks once again are the total opposite of every other site out there. The old excuse was canned benchmarks vs real world gameplay.
...I said it once and I'll say it again, Hardocp's biased, controversial, and more often then not, crap benchmarks are not worth reading, as is this thread.:thumbsdown:

I am now officially convinced that HardOCP is biased in favour of AMD. HD6950 is 10% faster than a GTX570 at 1920x1200? Ya, sure...In every new game, his website is the only one that continues to show AMD in a much better light than anywhere else. Coincidence?...

I agree with all these posts and have felt the same way about [H] for a long time. Their results are always skewed vs. every other website out there. Further more, they continue to test games at settings and resolutions that I personally feel are not playable settings.

 

[happy medium]

So wait - the 2GB Radeon 6950 is faster than the 1GB 560Ti at high resolutions where VRAM becomes a bottleneck.

Holy crap, next youa re going to tell me the Pope is Catholic and water is wet. Stunning revelation there.

If you want to do a side by side comparison, either 1) Compare a 1GB Radeon to the Ti or 2) Compare the 2GB Ti to the 2GB Radeon, so you can neutralize the VRAM discrepency. Otehrwise, you're doing a VRAM bench for a game that is obsviously demanding it whenever you enable every visual option.

Next, cut the fanboi crap like "sets XXX packing" which #1 is simply not true, and #2 is ridiculous.

Also, that 2GB Radeon costs a good deal more on average than a 1GB Ti. Not surprisingly, the 2Gb versions are within about $10 of each other.

Finally, please find me a stock 560Ti 822MHz that is still regularly sold by a prominent vendor such as Newegg. You can't. So this test of [H]'s, besides being setup to be ridiculously biased towards AMD (I have just about written that site off att his point - they've lost all intergity), is usueless to someone like me, because it doesn't test my gaming conditions.

Here's what I do know - I have 10 hours in the beta so far and I can, and have, played entire matches @ ultra everything on my rig and never dipped below 47 FPS per Afterburner. However, I prefer to shut HBAO off because the performance hit is ridiculous compared to the visual quality. There's the real story - what settings are worth it and what are not? that's the evaluation I am trying to make. Last night I played with 16xAF and no AA or HBAO and was getting about 115 fps in the subway. that's cute, but since I never dropped anywhere close to 60 fps, even in a fire fight, I probably should turn 2xAA back on and settle in. My goal is to find the best mix of settings that still delivers an average framerate without stuttering/huge dips/etc.

Can you please just for shits and giggles run @ 1080p ultra with 4xaa and run gpuz in the backround and tell us you max video memory used?
after you are done playing you just click on memory in gpuz and find the max.

Thanks so much.

edit: link to gpuz
http://www.techpowerup.com/downloads/2039/TechPowerUp_GPU-Z_v0.5.5.html

 

[Grooveriding]

If your memory limitation theory made sense -- then why is the GTX 570 offering more frame-rate than a 6950 with 2 gigs of ram based on Techspots findings? It is hardocp's findings that are out of whack compared to everyone else's -- the 6950 hands the GTX 570 a new one based on HardOCP evaluation as well and in that 15 percent range, when the GTX 570 is faster over-all than a Radeon 6950.

Well you can discuss the 570 and ignore the MIA 560 in that graph I was addressing if you want :sneaky: I have been addressing 1GB cards at 1080P/1200P. The 570 is not 1GB, correct ?

And my examples I gave of memory usage in my personal experience are @ 2560x1600, not the 1920x1200 in that graph, 2560x1600 is about 80% more pixels to push than 1920x1200.

 

[happy medium]

Well you can discuss the 570 and ignore the MIA 560 in that graph I was addressing if you want

If "the Advocate" comes through, you will be once again proven wrong.
Whats the difference between my gtx 460 1gb overclocked and a gtx560ti 1gb?
I just told you I only use 981mb of video memory at 1080p/ultra 4xaa.

Why not try to find out why instead of continuing to argue that it needs more than 1gb.
How much evidence do you need?

 

[Grooveriding]

Last night I played with 16xAF and no AA or HBAO and was getting about 115 fps in the subway. That's cute, but since I never dropped anywhere close to 60 fps, even in a fire fight, I probably should turn 2xAA back on and settle in. My goal is to find the best mix of settings that still delivers a 60 fps average framerate without stuttering/huge dips/etc.

Thank you. All I've been saying is that BF3 is very demanding on VRAM completely maxed out. Turning off AA or HBAO, which by the way I have to disagree - it adds a lot to IQ, allows you to get playable frames on 1GB cards.

115fps is impressive considering there are now two benches showing the 560ti averaging about 35fps on ultra without AA @ 1920x1200. I wouldn't of thought disabling HBAO would see 200% gains in FPS.

edit: ah, I see you are running 1680x1050 - not 1080P or 1920x1200, so that is going to be more forgiving on your FPS.

 

[TheAdvocate]

Can you please just for shits and giggles run @ 1080p ultra with 4xaa and run gpuz in the backround and tell us you max video memory used?
after you are done playing you just click on memory in gpuz and find the max.

Thanks so much.

edit: link to gpuz
http://www.techpowerup.com/downloads/2039/TechPowerUp_GPU-Z_v0.5.5.html

I've got GPU-z. I'll see if I have time tonight.

I suspect that I use every bit of my VRAM, even at 1680x1050. Every anecdote and bench I have seen suggests that the game will use what it can, but to your earlier discussion about the 570 vs the 6950, the game taxes the GPU chips as well, so a really fast chip can probably partially overcome a VRAM bottleneck, or so I suspect.

 

[happy medium]

I've got GPU-z. I'll see if I have time tonight.

I suspect that I use every bit of my VRAM, even at 1680x1050. Every anecdote and bench I have seen suggests that the game will use what it can, but to your earlier discussion about the 570 vs the 6950, the game taxes the GPU chips as well, so a really fast chip can probably partially overcome a VRAM bottleneck, or so I suspect.

Really if you look here it seems I am right.
How on earth would a gtx 570 tie a 6970 @ 2500x1600/ultra/4xaa?
Groover himself said he uses 1.45gb of memory at this setting.
The gtx 570 would definitly run out of memory if I was wrong.

On a side note look at sli scaling go!!!!!!!

 

[SirPauly]

If you want to do a side by side comparison, either 1) Compare a 1GB Radeon to the Ti or 2)

Hardocp did. They compared a 1 gig 6870 vs a 1 gig GTX 560 TI, and reduced the settings of the GTX 560 TI and the 6870 came on top by around 15 percent and it sent the GTX 560 TI packing and begging for frame-rate.

 

[TheAdvocate]

Thank you. All I've been saying is that BF3 is very demanding on VRAM completely maxed out. Turning off AA or HBAO, which by the way I have to disagree - it adds a lot to IQ, allows you to get playable frames on 1GB cards.

115fps is impressive considering there are now two benches showing the 560ti averaging about 35fps on ultra without AA. I wouldn't of thought disabling HBAO would see 200% gains in FPS.

I must be blind then - re: HBAO. I honestly cannot tell it's on. Is there a video capture somewhere than shows it?

I'm done with using screencaps for reference to make these decisions. It's not a still life, it's a game. If it's not something that I'll see running around getting rockets shot up my ass, it's probably not worth the performance hit. My friend swears max AA helps him snipe. I dunno. But point is, most of this is childish chest thumping to justify ridiculous uses of discretionary income on GPUs that will be in the clearance bin in 18-24 mos. from my standpoint, I just want to evaluate the value perspective - give me the cheapest way to run a brand new game @ constantly playable framerates. Eventually, the hardware will catch up to the game, and a $100 card will run it at 100 fps with all the eye candy turned on.

In the meantime, I don't get why people are arguing or benching mid range GPUs at high end resolutions, and then arguing because one value card does 5 fps better than the other but neither can really handle the 2560x1440 or whatever. Once it's obvious that you're going to be stuck in the 20s/30s/40s for fremerates, the discussion is over. If you really want to game at that resolution, buy a more powerful card. I don't care how fast my Ti runs the game at high res. I didn't buy it for that. I bought it for 1680x1050 and it runs the beta at that resolution quite capably, even easily.

 

[happy medium]

:thumbsup:

Hardocp did. They compared a 1 gig 6870 vs a 1 gig GTX 560 TI, and reduced the settings of the GTX 560 TI and the 6870 came on top by around 15 percent and it sent the GTX 560 TI packing and begging for frame-rate.

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:

 

[SirPauly]

Well you can discuss the 570 and ignore the MIA 560 in that graph I was addressing if you want :sneaky: I have been addressing 1GB cards at 1080P/1200P. The 570 is not 1GB, correct ?

And my examples I gave of memory usage in my personal experience are @ 2560x1600, not the 1920x1200 in that graph, 2560x1600 is about 80% more pixels to push than 1920x1200.

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.

 

[Grooveriding]

Really if you look here it seems I am right.
How on earth would a gtx 570 tie a 6970 @ 2500x1600/ultra/4xaa?
Groover himself said he uses 1.45gb of memory at this setting.
The gtx 570 would definitly run out of memory if I was wrong.

Unfortunately there are no minimums in that graph. As per that reviewer's comments. No single card can handle the resolution so they were all dying.

'There is no denying that dual-GPU configurations are a must at 2560x1600 using Ultra settings. The slowest single GPU card we tested, the Radeon HD 6950 averaged just 20fps. The GeForce GTX 570 also managed 21fps while the GeForce GTX 580 was slightly faster with 23fps. Even the Radeon HD 6990 scored a mere 34fps while the Radeon HD 6970 Crossfire cards and GeForce GTX 590 were matched at 36fps.'


Considering the usage of memory I've seen, it probably explains why the multi-gpu configurations shown are 580SLI and 6970CF. The 570 would not cut it unless they were the 2.5GB versions.

Also the 580SLI is doing fairly poorly. My cards are overclocked to stock 580 performance and I never drop below 50 or so as a minimum, my average is about 75. I do have a third card, but I don't run 4xAA because of running out of memory. The 580s may have had their score nerfed down because of poor minimums trying to run 4xAA.

 

[happy medium]

I'm done with using screencaps for reference to make these decisions. It's not a still life, it's a game. If it's not something that I'll see running around getting rockets shot up my ass, it's probably not worth the performance hit. My friend swears max AA helps him snipe.

I totally agree with you here, I dont see much difference between high and ultra to be honest.

 

[SirPauly]

: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:

I like Kyle's site. I'm trying to gauge why their findings differ than the others. I don't think it is bias.

 

[happy medium]

Unfortunately there are no minimums in that graph. As per that reviewer's comments. No single card can handle the resolution so they were all dying.

'There is no denying that dual-GPU configurations are a must at 2560x1600 using Ultra settings. The slowest single GPU card we tested, the Radeon HD 6950 averaged just 20fps. The GeForce GTX 570 also managed 21fps while the GeForce GTX 580 was slightly faster with 23fps. Even the Radeon HD 6990 scored a mere 34fps while the Radeon HD 6970 Crossfire cards and GeForce GTX 590 were matched at 36fps.'


Considering the usage of memory I've seen, it probably explains why the multi-gpu configurations shown are 580SLI and 6970CF. The 570 would not cut it unless they were the 2.5GB versions.

Also the 580SLI is doing fairly poorly. My cards are overclocked to stock 580 performance and I never drop below 50 or so as a minimum, my average is about 75. I do have a third card, but I don't run 4xAA because of running out of memory. The 580s may have had their score nerfed down because of poor minimums trying to run 4xAA.

There are no memory limitations with this game engine ,plain and simple.
It uses resources avalable.

 

[TheAdvocate]

Hardocp did.

No. They did not test the 6950 2GB against and 900Mhz 2GB 560Ti (standard clock speed on Ti's now - My overclock is to 1GHz) or a 1 GB Radeon 6950 against a 1GB 560Ti, whcih is the test i want to see, preferably from Anandtech.

I'm done arguing with you now.

 

[happy medium]

I like Kyle's site. I'm trying to gauge why their findings differ than the others. I don't think it is bias.

Well I dont like thier site, and I fully believe they purposly give totally opposite reviews to gain page hits through contriversy. Bias? mabe not, accurate? no way in hell.

 

[TheAdvocate]

There are no memory limitations with this game engine ,plain and simple.
It uses resources avalable.

I think that's too simplistic. Wouldn't it just use less compressed textures if it had the memory? Or vice versa?

Past a certain resolution, you've got frame buffer VRAM issues, no matter what tricks you use to enable better scaling. I think it can run on 1GB RAM at absurdly high resolution, but I don't think it's ideal.

God, I swear I feel like I'm at a race track arguing whose Honda Accord 4 door is faster. None of these cards being discussed is high end. If you want performance, get a 580 SLI or 6990 rig. Quadfire your SLI and all that rot. I'd rather not spend $2k on GPUs just to play one game. Lotta other things my wif... err I'd rather spend it on.

 

[Grooveriding]

There are no memory limitations with this game engine ,plain and simple.
It uses resources avalable.

Well you can choose to believe that I don't want to stand in your way.

You can ignore the several sites not even bothering to bench 1GB cards with 4xAA @ 1080P.

And you can ignore the results of a 1GB card like the 560 getting about 35FPS without AA at 1920x1200 on several sites. I guess you think adding the 4xAA will make frames go up.

And ignore the proof that games do not just 'use up' whatever memory is there as I showed many other demanding games just don't do that. They use what they need.

Also you can ignore my examples from BF3 showing it not using all available VRAM dependent on quality or AA settings. Ranging from using 1.25GB to maxing out the full 1.5GB on my system.

Not trying to change what you're posting or think I can. But if you want to keep replying to my posts refuting what I'm saying, please offer something to show why, instead of just saying it's not so.

 

[SirPauly]

No. They did not test the 6950 2GB against and 900Mhz 2GB 560Ti (standard clock speed on Ti's now - My overclock is to 1GHz) or a 1 GB Radeon 6950 against a 1GB 560Ti, whcih is the test i want to see, preferably from Anandtech.

I'm done arguing with you now. It's pretty obvious what you are.

What am I going to do with you?

No, HardOCP tested a much slower 1 gig 6870 product and it came up on top by a very noticeable margin and still reduced the settings of the 1 gig GTX 560 TI. Isn't this odd?

The GTX 560 TI is around 15 percent faster over-all than a HD 6870, both with 1 gig of memory and yet, HardOCP's findings are 15 percent faster for the 6870 --- very odd findings, one may imagine. Why? What can possibly be the reason for this?

 

[happy medium]

It uses resources avalable

Care to comment? You seem to be ignoring the obvious.
Whether playable or not, it will not use more than avalable memory.
Again there are no memory limitations with this game engine.

The game engine seems to adjust to the highest avalable memory use with no performance hit.
Like I said mabe it suffers from a image quality hit but I have not seen any evidence of this.

 

[TheAdvocate]

That's probably because the final textures/features/etc are not in the game yet.

If you're going to make such a fine line decision happymedium, then you'd probably need to wait for the release. All the beta does is show basic trends. I expect final performance numbers to be very different once all the effects are in. It doesn't appear that there is any tesselation in right now, and the nVidia cards do that better, so it will probably translate to a framerate hit to the Radeon cards, as one example.

 

[RobertR1]

I'm going to test my OC'd 580 at 2560x1600 using hardocp's recommended settings.

That will help us see how of a difference over clocking makes in this game.

Should I just record using fraps? Any benchmarking tips to come close to hard's methods?

 

[SirPauly]

That's probably because the final textures/features/etc are not in the game yet.

If you're going to make such a fine line decision happymedium, then you'd probably need to wait for the release. All the beta does is show basic trends. I expect final performance numbers to be very different once all the effects are in. It doesn't appear that there is any tesselation in right now, and the nVidia cards do that better, so it will probably translate to a framerate hit to the Radeon cards, as one example.

There is already a frame-rate hit on Radeons with every other site except for HardOCP, which there is a dramatic difference and dramatic turnaround.

 

[TheAdvocate]

Let's do a fanboy post to see how much value is added to the thread...

2.1 average fps faster than the 2GB Radeon 6950! ZOMFG THE 560TI STOMPS A F'ING HOLE IN THE RADEONS CRANIUM ALL THE WAY DOWN TO IT'S ANAL CAVITY AND THEN RIPS OUT IT'S ENTRAILS AND STRINGS UP ITS MOTHER SO THE JACKALS CAN SERVICE AND THEN CONSUME HER !11111!!

24.48979591836734693877510204082 % faster than the 6870 OMFGGAGGG!@!!