HD7850 + 1080p = BF on ultra?

[andybird123]

The problem with multiplayer is something different happens every time making it very hard to produce consistent results.

whilst this is true in terms of averages, the minimum frame rate you experience shouldn't be a million miles away from each other - unless on one run you get right in the thick of it and then next you stay at spawn starring at the floor

in total I logged hours of gameplay (as in logged to text file) when I was trying various GPU vs. memory overclocks on 1080p and never once did I see a dip to 30's or 40's in gameplay (alt tabbing out and back in will cause a dip to 16 fps but that is by the by)

in all of that time I find it hard to believe that I didn't experience the same type of situation that seems to be causing these dips on other users machines

 

[Rusty0611]

Its a real shame thee isnt a benchmark utility for BF3

Ive yet to see evidence of peformance gains in BF3 for the latest AMD drivers. Just because a few people say its true and jump on the bandwagon doesnt make it so. I'm not saying there arent improvements but we need to see them before we can call it fact.

Its also possible that nvidia drivers have improved and they are seeing better minimum frame rates than they used to.

+1 million

This kind of gets at what I was trying to say

in all of that time I find it hard to believe that I didn't experience the same type of situation that seems to be causing these dips on other users machines

Same here - I've ran FRAPS benchmarks while playing multiplayer for hours as well and I have also never seen drops like that.

 

[Havana_UK]

Matt, let's use the following in-game settings:
All detail at ultra
4xMSAA
FXAA High
16xAF
VSync off
Motion Blur on
HBAO Occlusion
(I think that all the above is default ultra settings)

Don't force any other settings in nVidia/AMD drivers.

Now, you'll have to tell me what I need to do in Afterburner to record the vid and capture the relevant info as I've never done that before.

 

[LtMatt]

I'll just leave this here.

Using AMD 12.8 driver. AMD have caught up.

http://gamegpu.ru/action-/-fps-/-tps/battlefield-3-armored-kill-test-gpu.html

 

[LtMatt]

Matt, let's use the following in-game settings:
All detail at ultra
4xMSAA
FXAA High
16xAF
VSync off
Motion Blur on
HBAO Occlusion
(I think that all the above is default ultra settings)

Don't force any other settings in nVidia/AMD drivers.

Now, you'll have to tell me what I need to do in Afterburner to record the vid and capture the relevant info as I've never done that before.

Ok mate, i think if we just use the ultra pre-set that will be easier. I will have to play around with some afterburner settings as ive never recorded using it either. The quality isn't really important.

EDIT

See my post above, no longer need to do it. The proof is in the pudding. A recent review using recent drivers.

 

[Spenty]

That graph says it all tbh. That's only at 1080p as well. I'd love to see a video of someone proving it doesn't drop below 60 though.

 

[Rusty0611]

I'll just leave this here.

That graph says it all tbh. That's only at 1080p as well. I'd love to see a video of someone proving it doesn't drop below 60 though.

Taking these results as gospel (which I have a huge problem with anyway as a principle) and with an overclock it's going to bump up to around the figures Andy and I have been saying. Not 30's...

Also 7970 GE shouldn't be compared to a stock 680. A factory BIOS OC isn't a like for like comparison.

So the original point which we responded to doesn't hold up just on the face of that graph.

 

[Havana_UK]

I think I mentioned in a previous post that the AK maps (probably due to their size) seem to hit the minimums harder than the originals, because I have seen lower minimums (42 from memory) on Bandar (sp?) Desert when I was right up in the crane and there were a bunch of exploding tanks I was looking at.

 

[LtMatt]

Taking these results as gospel (which I have a huge problem with anyway as a principle) and with an overclock it's going to bump up to around the figures Andy and I have been saying. Not 30's...

You can't write this one off rusters.

Its there for all to see and the results display as i expected.

All the cards there would benefit from an overclock of course.

http://forums.anandtech.com/showthread.php?p=33958417

 

[Gregster]

I don't wish to sound like I am going off on one but that chart is trash. I get far more than that with a single 680 not over clocked.

 

[LtMatt]

Interesting to note the vram usage.

 

[Rusty0611]

You can't write this one off rusters.

Its there for all to see and the results display as i expected.

See my edit - 7970 GE is not comparable directly to a 680.

Don't be stupid now .

And as Gregster said - I personally see higher results than those.

 

[LtMatt]

See my edit - 7970 GE is not comparable directly to a 680.

Don't be stupid now .

And as Gregster said - I personally see higher results than those.

That's because your card is overclocked, this just gives a representation on stock clocks for all the cards listed.

Looks like a good review to me, i see no mention in anand forums that its a terrible inaccurate review.

If its that bad post your findings over there.

Thanks for calling me stupid.

 

[Rusty0611]

That's because your card is overclocked, this just gives a representation on stock clocks for all the cards listed.

Looks like a good review to me, i see no mention in anand forums that its a terrible inaccurate review.

If its that bad post your findings over there.

Thanks for calling me stupid.

Well obviously I think that kind of goes without saying.

On stock settings I still see higher. After all you're drawing a general conclusion from only one set of results and basing a 7970 GE vs stock 680.

As I said, I have no interest in debating which card is faster but I will question dodgy conclusions drawn from only one set of results. I have no problem with the results themselves on their own only your statement above which in your haste to declare victory you have overlooked some fundamental points.

And actually the minimums are the same but a stock 680 is still faster overall than a 7970 GE. If you think that is "caught up" then I'll let you think that .

 

[LtMatt]

Well obviously I think that kind of goes without saying.

On stock settings I still see higher. After all you're drawing a general conclusion from only one set of results and basing a 7970 GE vs stock 680.

As I said, I have no interest in debating which card is faster but I will question dodgy conclusions drawn from only one set of results. I have no problem with the results themselves on their own only your statement above which in your haste to declare victory you have overlooked some fundamental points.

And actually the minimums are the same but a stock 680 is still faster overall than a 7970 GE. If you think that is "caught up" then I'll let you think that .

I haven't been debating which one is faster, i said nvidia are, and probably were still faster. You keep bringing that up.

I' m saying and have been saying since the start that both sides suffer fps drops and this review has just proved that fact.

I'm still waiting for the video from you to prove you never dip below 60fps on Armoured Kill.

AMD have caught up a lot from where they were, give or take a few fps i think its safe to say they've caught up, or almost caught up. Not bad for TWIMTBP game.

 

[Gregster]

I didn't realise you could speak Russian Matt You spend too much time as the ruskies on BF3 me thinks....

 

[Rusty0611]

I haven't been debating which one is faster, i said nvidia are, and probably were still faster. You keep bringing that up.

I'm still waiting for the video from you to prove you never dip below 60fps on Armoured Kill.

AMD have caught up a lot from where they were, give or take a few fps i think its safe to say they've caught up, or almost caught up. Not bad for TWIMTBP game.

True - it's good they've caught up as nobody in the right mind wants other people to have bad performance but you haven't addressed my 7970 GE vs 680 point.

I'm not stating that (the faster bit) for your benefit more in case somebody else chimes in .

You might be waiting a while - I've provided FRAPS results. If you think they're false then man up and call them out rather than asking for a video of me sitting there playing BF3. I am pretty peng but that is a rather odd request . *joking*

I' m saying and have been saying since the start that both sides suffer fps drops and this review has just proved that fact.

I'm not surprised AK provides lower minimums but I've already given FRAPS results from the vanilla game clearly showing it doesn't go under 60.

Stock vs stock comparisons are pretty boring anyway. As you've asked so nicely I will play an AK map for 15 mins tonight and post the results but it will be overclocked 680 results as I don't see much point of stock benchmarks when there's a plethora of results already out there.

 

[Spenty]

Tbh Rusty fraps results could easily be altered when posting them like that. Now it just sounds like you're avoiding the situation as you know it will probably dip under 60 a few times and you don't want to be proved wrong.

 

[Gregster]

FRAPS recording gives a hit on performance quite abit I have found.

 

[LtMatt]

You provided fraps results which although nice prove nothing, we just have to take your word.

I've provided many reviews from different sites, but they're all wrong because you get different results.

Myabe you should run a test on AK Death Shield and see what you're minimums are there.

The 7970GH is abit faster than the 680, is that your point? I don't disagree but i don'th think theres too much difference between them is there? The stock 7970 was a bit slower than the 680 with boost, but amd released the ghz edition to catch up or even overtake slightly.

The main point is im sure before 12.8's the ghz edition was no way near the 680 on bf3.