• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

***THE BF4 BENCHMARKS THREAD***

LtMatt said:
If it was irrelevant then the 770 SLI minimums would not be so low and neither would the 690 minimums. After speaking with a 690 user from another forum he confirmed that he also had fps drops that stopped once he removed AA and then he was able to maintain 60 fps give or take. The main thing to note is the drops do not happen instantly. Only as you play the game more and move to different parts of the map. Something that is not going to be picked up in a 40 second bench.
Click to expand...

The minimum is the same minimum as the 7970 GE. You're kind of going off point there:

a) VRAM isn't affected by action on screen but frame rate is. In fact I've seen usage decrease sharply while in battle so it's kind of independent of that rather than being linked

b) the previous bench being referred to has not been consistently measured - online gameplay by definition is varied

c) GameGPU in your words are trusted enough and their actual FPS readings are in line with what others are getting so there is no reason to take their results above or below what others have got

I think the problem is you've dug way too deep on this subject and you can't see the wood for the trees. Kaap wasn't limited by VRAM on his 690s at 1600p and the results from GameGPU suggest the same. The sweclockers one also isn't even showing the 2GB cards running out of VRAM. That's your interpretation based on your own bias.

Having bias isn't a criticism but you have to make sure you're seeing everything and not just seeing the results as you want to see them. Scientific bias and all that :p.

Edit: I agree that mem usage will increase over time but we're not talking a massive amount here.
 
Last edited:
Kaap can you run a whole map with Fraps on if you get the chance? Two 690s will be fine. And report if you get any page file / system RAM accesses from a lack of VRAM?
 
LtMatt said:
Mate that was just for that benchmark. If you look at my started threads ive run many benchmarks before that. I know that mission better than any man to have graced the universe. Nothing to write home about granted. That's why i now use a different part of the game. Its a bit less demanding and shorter but its very consistent. ;)
Click to expand...

I figured, was only kidding :D.



Own logos you say...will be importing all my old time favourites from TF2 lol.
 
Rusty0611 said:
The minimum is the same minimum as the 7970 GE. You're kind of going off point there:

a) VRAM isn't affected by action on screen but frame rate is. In fact I've seen usage decrease sharply while in battle so it's kind of independent of that rather than being linked

b) the previous bench being referred to has not been consistently measured - online gameplay by definition is varied

c) GameGPU in your words are trusted enough and their actual FPS readings are in line with what others are getting so there is no reason to take their results above or below what others have got

I think the problem is you've dug way too deep on this subject and you can't see the wood for the trees. Kaap wasn't limited by VRAM on his 690s at 1600p and the results from GameGPU suggest the same. The sweclockers one also isn't even showing the 2GB cards running out of VRAM. That's your interpretation based on your own bias.

Having bias isn't a criticism but you have to make sure you're seeing everything and not just seeing the results as you want to see them. Scientific bias and all that :p.
Click to expand...

No that's their (swe clockers) interpretation backed up with evidence vs your opinion. Your opinion not having done any benchmarks, and going off kaap's results where he used 4 gpu's instead of 2. He also admitted he was rubbish and therefore didn't play for long as its not his type of game.

I've had another 690 user who plays a lot seeing similar performance to what swe clockers say. Only by lowering AA did he stop the frame rate dips to low teens. Something which is not experienced on other 3gb+ cards included in the bench. I do not see this on my 2x7950's.

As previously stated this will not show itself in a short 40 second benchmark. Its something that will only be experienced by playing the game and exploring the level as the buffer fills up. Its not something that will show itself in a short bench or playing for a few minutes.

I think the problem is you've also dug yourself in too deep that you will just argue the point despite there being good evidence that what im saying is correct. To use a famous quote Rusty, you're going round the houses.

I don't mind admitting im wrong as and when there is some clear evidence of it. So far the only clear evidence we have are the swe clockers benchmarks where only 2gb SLI cards drop to low teens.

Rusty0611 said:
Kaap can you run a whole map with Fraps on if you get the chance? Two 690s will be fine. And report if you get any page file / system RAM accesses from a lack of VRAM?
Click to expand...

No you copy their settings and setup or its pointless. No ifs or buts. Either do it the same or not at all. :)

If Kaap plays for a full round on 64 man Conquest large at the highest settings with x4 AA we'll see what his frap results are using a single 690 at stock.
 
Last edited:
We are seeing benchmarks all over the internet showing when given the chance this game is using easily over 2GB. Why - seriously why is it so hard to believe that you're going to have performance related issues with 2GB cards. If anything you're just complementing DICE's abilities :D
 
LtMatt said:
No that's their (swe clockers) interpretation backed up with evidence vs your opinion. Your opinion not having done any benchmarks, and going off kaap's results where he used 4 gpu's instead of 2. He also admitted he was rubbish and therefore didn't play for long as its not his type of game.

I've had another 690 user who plays a lot seeing similar performance to what swe clockers say. Only by lowering AA did he stop the frame rate dips to low teens. Something which is not experienced on other 3gb+ cards included in the bench. I do not see this on my 2x7950's.
Click to expand...

No it's not their interpretation - they guessed based on what they thought was happening without any technical explantion.

Not surprising the 690 is struggling more at that res and with MSAA. Memory bandwidth will be a factor there. We're not talking about 690 and 7950 CF though: we're talking about 770 vs 7970 GE.

No offence but I don't really take anecdotal reports as really that useful. Sure they're not useless or anything but they aren't gospel.

LtMatt said:
As previously stated this will not show itself in a short 40 second benchmark. Its something that will only be experienced by playing the game and exploring the level as the buffer fills up. Its not something that will show itself in a short bench or playing for a few minutes.
Click to expand...

As previously stated, I agree, but not to the extent being implied. My memory usage was pretty static throughout the game.

LtMatt said:
I think the problem is you've also dug yourself in too deep that you will just argue the point despite there being good evidence that what im saying is correct. To use a famous quote Rusty, you're going round the houses.

I don't mind admitting im wrong as and when there is some clear evidence of it. So far the only clear evidence we have are the swe clockers benchmarks where only 2gb SLI cards drop to low teens.
Click to expand...

You're ignoring all of the other evidence in favour of evidence which allegedly backs up your own bias. And the evidence doesn't even back this up. It could do but it is far from proven. You're finding fault with the other evidence where no fault exists due to that same bias.

I'm not dug in at all: I'm not the one going into every thread and creating threads saying 2GB won't be enough :p. And let's be honest, multi-GPU and 2560 resolution is a fairly niche market, even at the high end so it is all a bit of a wet fart. In truth, I couldn't care less is 2GB is limiting in this scenario other than for the purpose of debating.


LtMatt said:
No you copy their settings and setup or its pointless. No ifs or buts. Either do it the same or not at all. :)

If Kaap plays for a full round on 64 man Conquest large at the highest settings with x4 AA we'll see what his results are.
Click to expand...

The point being if he is truly VRAM limited the extra GPUs won't help his minimums so infact it is a better test.

MjFrosty said:
We are seeing benchmarks all over the internet showing when given the chance this game is using easily over 2GB. Why - seriously why is it so hard to believe that you're going to have performance related issues with 2GB cards. If anything you're just complementing DICE's abilities :D
Click to expand...

BF3 used 2.4GB at 1080p on 580 3GBs yet used 1.7GB on 580 1.5GBs and had the same framerate. Usage is just one barometer not the be all and end all.
 
Last edited:
MjFrosty said:
We are seeing benchmarks all over the internet showing when given the chance this game is using easily over 2GB. Why - seriously why is it so hard to believe that you're going to have performance related issues with 2GB cards. If anything you're just complementing DICE's abilities :D
Click to expand...

What do you expect with such graphical settings!

I'm running medium-high and it's only doing just over 1GB VRAM

Oh and these performance problems most people are having are NOT related
 
LtMatt said:
Swe clockers >> Rusty. End of.
Click to expand...

Mines bigger than yours :p.

Got you before your edit.

LtMatt said:
Still going round the houses Rusty.

To summarise...

Swe clockers opinion >> Rusty's opinion.
Click to expand...

Why is that? They're just guys benchmarking harware the same as anybody else here. They aren't some kind of god's just because they're under the umbrella of a popular website.
 
It's not size that matters it's what you do with it that matters.

I think that applies to both the size and VRAM sides of this debate. :D :D

I'll get me coat
 
Rusty0611 said:
It's not size that matters it's what you do with it that matters.

I think that applies to both the size and VRAM sides of this debate. :D :D

I'll get me coat
Click to expand...

Only people with small ones say that. :D

When the final build arrives scores will be settled and debts will have to be paid. Although i still think im right i do agree that we need to wait for the final version because the game is just too buggy at the moment.

Good to see the Alpha vram benchmarks i posted ages back were 100% accurate. Man i got some grief for posting those but they were spot on. Now we wait to see what effect the final build will have on things. Given that some ultra settings and lighting effects are missing, i anticipate a further possible climb in vram figures.

Stay tuned people for next exiting vram debate, coming to a forum near you. :p
 
Cmon guys, are you kidding us here, back to the '2b is enough' statements despite it being a single beta map that doesn't have all it's assets -seriously?

Before I get flack for something I haven't said, the above sentence does not read '2Gb isn't enough' as I don't know for absolute certainty what's what.

There will be people taking a punt on 2Gb cards based on what very opinionated absolute statemants are being made in this thread without knowing for sure what's needed despite DICE making it very clear there is more assets going into the finished game.

Has anyone actually watched any other maps for comparison?

For example Flood Zone is a whole different game when the dam breaks and floods the park/streets, the lighting is incredible(3m10s)-I haven't seen any compute based lighting like that on the Shanghai map but admittedly, I'm **** at the best of times to notice as I'm trying not to get killed all the time when on foot.

 
You must log in or register to reply here.
Back
Top Bottom