Battlefield 4 Video Card Performance and IQ Review @ [H]

FrgMstr

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,634
Battlefield 4 Video Card Performance and IQ Review - Battlefield 4 is this holiday season's blockbuster from the Battlefield series. It features the brand new Frostbite 3 game engine which provides a higher level of realism in the game. We strap 8 video cards to the test bench to see what kind of gameplay experience is delivered under Windows 8.1.
 
Very nice! Can't wait for the 780 Ti results and multi-gpu.

To me though (and to many others) 60fps is a minimum requirement for multiplayer shooters.

Interesting that the 280X and the 290 are pulling similar minimums and average frame rates...
 
I thought someone said [H]ard was going to do comparison between Win 7 and Win 8.1 in BF 4 multi-player. I hope that's going to happen eventually.

I fully understand how difficult it is to get similar runs in a multi-player environment. Some folks don't seem to though. They throw up results and speak about them like they just ran the Valley bench a few times.
 
did you notice the system ram usage while playing bf4? and, what was the speed of the system ram?
 
Blacklash said:
I thought someone said [H]ard was going to do comparison between Win 7 and Win 8.1 in BF 4 multi-player. I hope that's going to happen eventually.

I fully understand how difficult it is to get similar runs in a multi-player environment. Some folks don't seem to though. They throw up results and speak about them like they just ran the Valley bench a few times.
Click to expand...

Yes and yes. Conclusions are not drawn from the small amount of data that we see on the graphs.

trick0502 said:
did you notice the system ram usage while playing bf4? and, what was the speed of the system ram?
Click to expand...

1600MHz.
 
Great Sunday night read. So - with a 770 (680), could you still run at 2560x1600 with some lower settings? Is it really better to drop down to 1080p? It is sort of the same recommendation you made with Tomb Raider.
 
In all the high range cards you have a setting of 16xAF. How is that set?

EDIIT: Is hyperthreading enabled on your test rig?

Thanks for this review, there is a deal tonight of 10% off, I think a R9 290 will do a nice replacement for the GTX580SLI, especially (hopefully) once Mantle is enabled in BF4 (BF4 will be 90% of my gaming of the next year or so)
 
Last edited:
Very nice review! I have been waiting for someone to do something comprehensive like this and you guys hit the nail on the head.

Definitely getting the 280X. Best price/performance card for 19x10.

OT: Should I even wait for Black Friday / Cyber Monday to purchase my card?
 
A good tear down of Bf4, focusing on MP no less! Many sites would simply use the SP campaign as a benchmark. It is interesting to see Nvidia does really fluctuate in fps more than AMD, hopefully this can be addressed in one form or another by a driver or patch by both sides.
Overall the game is very new so hopefully there is lots of room for improvement.
With so many of us now enjoying 120hz monitors the term playable becomes different (imo).
Even once I upgrade I'm sure ill have to lower settings to keep my 100fps/100hz combo nice and smooth.
Thanks [H]!
 
With the recent price drops, the 7990 @ $599 also becomes an interesting player in single card configurations.
 
I'd like to know if there's a performance improvement in Windows 8.1 vs Windows 7.
 
mach1.9pants said:
In all the high range cards you have a setting of 16xAF. How is that set?

EDIIT: Is hyperthreading enabled on your test rig?

Thanks for this review, there is a deal tonight of 10% off, I think a R9 290 will do a nice replacement for the GTX580SLI, especially (hopefully) once Mantle is enabled in BF4 (BF4 will be 90% of my gaming of the next year or so)
Click to expand...


first question, you can set it in your drivers..

second question yes it was enabled. was written at least 3 times that i saw in the review.
 
Nice article. I would love to see some overclock results as well (be it only apples to apples).
Also, is there IQ difference between AMD or nVidia especially in terms of AA? I saw a noticeable improvement in IQ using nVidia technologies such as SGAA which was available at a huge cost on AMD.

This was when I compared Radeon 7970 to a GTX 680 and found AA to be superior - even MSAA on nVidia. Maybe I am too sensitive to these things.
 
KickAssCop said:
Nice article. I would love to see some overclock results as well (be it only apples to apples).
Also, is there IQ difference between AMD or nVidia especially in terms of AA? I saw a noticeable improvement in IQ using nVidia technologies such as SGAA which was available at a huge cost on AMD.

This was when I compared Radeon 7970 to a GTX 680 and found AA to be superior - even MSAA on nVidia. Maybe I am too sensitive to these things.
Click to expand...


probably will have to wait for the custom cooled/PCB 290/290x for apples to apples overclocks.
 
pandora's box said:
oh wow. i just tested this myself,

Low setting = Ultra)
Normal setting = High
High setting = Normal
Ultra setting = Low
Click to expand...

Interesting. That could definitely impact a bench. What's the difference in FPS for you between Low and Ultra Terrain Decoration?

Credit to Clubic.com. They're the ones that noticed it.
 
Man that is too bad no modern day video card can run Resolution scale at 200.

Maybe 2 290x's in crossfire @ 1080p?....LOL 3.5g of Vram....crazy shit.

Thanks for the great review. This honestly is making me think about sending back the lightnings for 2 290's.....Way cheaper and more performance out of the box.
 
1384476134-06829508-photo-details-du-terrain-faible-donc-ultra.jpg
 
"We also spent some time gaming with the R9 290X at 1920x1080 with resolution scale set to 200 and experienced average frame rates in the 30-40FPS range. Given its impact on performance as well as using about 3.5GB of VRAM, we believe that the overall gaming experience is better using MSAA while maintaining a higher frame rate."

That 3.5gb of VRAM used was at 1080p?
 
I took a 1080P video of the BF4 issue. It's uploading to youtube now. Compared all the terrain decoration settings.
 
DASHlT said:
"We also spent some time gaming with the R9 290X at 1920x1080 with resolution scale set to 200 and experienced average frame rates in the 30-40FPS range. Given its impact on performance as well as using about 3.5GB of VRAM, we believe that the overall gaming experience is better using MSAA while maintaining a higher frame rate."

That 3.5gb of VRAM used was at 1080p?
Click to expand...

Resolution scaling is OGSSAA - Of course it uses a ton of VRAM...that is to be expected.

Resolution scaling/OGSSAA is not a realistic setting for anyone to use in any game, I don't use SSAA except in games that are YEARS old, like from 2008 or beyond. OGSSAA is even more demanding than SGSSAA. You won't be using resolution scale in BF4 any time soon at 200%. ;) Maybe 2 years from now. Maybe.
 
xoleras said:
Resolution scaling is OGSSAA - Of course it uses a ton of VRAM...that is to be expected.

Resolution scaling/OGSSAA is not a realistic setting for anyone to use in any game, I don't use SSAA except in games that are YEARS old, like from 2008 or beyond. OGSSAA is even more demanding than SGSSAA. You won't be using resolution scale in BF4 any time soon at 200%. ;) Maybe 2 years from now. Maybe.
Click to expand...

Well if its using 3.5G VRAM at 1080p. Then I guess you can use 2 290x to run it....but for 1080p its almost LOL
 
Whelp, 280x or continue not having a GPU while waiting for the 290 Non-ref.

Really interesting to see how much better the 290 performed than the 780. I'm still expecting a price drop of the 780 to $400.
 
I would see when the tests were performed.

The client side patch that was released actually increased FPS. So anything that was tested before the client side patch should now be rendered invalid.

Also, Ram speed in BF4 does matter.

I've done my own testing with this and picked up 4 FPS(5760x1080) going from 1600 to 1866, most should be getting 12-15 FPS at 1080.

I would also test the 290 with the 13.11 Beta 9.2 driver to see the change in performance with the increase in fan speed.
 
Priller said:
I would see when the tests were performed.

The client side patch that was released actually increased FPS. So anything that was tested before the client side patch should now be rendered invalid.

Also, Ram speed in BF4 does matter.

I've done my own testing with this and picked up 4 FPS(5760x1080) going from 1600 to 1866, most should be getting 12-15 FPS at 1080.

I would also test the 290 with the 13.11 Beta 9.2 driver to see the change in performance with the increase in fan speed.
Click to expand...

The patch that dropped a few days ago was the first major patch that dealt with memory leaks, framerates, client and server stability, CPU Usage.

If this testing was performed prior to the patch (Which is probably the case) I would certainly take them with a grain of salt.

The 780ti results, if performed after the patch, are going to be very skewed.
 
Good read, you guys always go the extra mile to make sure your reviews are thorough. Did you do any SLI testing with the 331.70 drivers? I am hoping it helps the flickering issues that we are seeing now.
 
pandora's box said:
oh wow. i just tested this myself,

Low setting = Ultra
Medium setting = High
High setting = Medium
Ultra setting = Low
Click to expand...

I just tried this on my personal installation, and it is affecting me too. Had absolutely no idea. One would assume when you set ultra it would be ultra, but the values do seem to be reversed. Thing is, this is probably happening for everyone, and they don't know it and are also assuming ultra terrain decoration is the best terrain decoration detail.

We used the global setting to set Ultra in our review. I'm having David check it now to see if it affected us too.

Why does BF4 still feel like we are playing a Beta version? So many bugs.
 
Any comments on Corsair's claims that higher speed system RAM affects framerates dramatically?
 
Domingo said:
Any comments on Corsair's claims that higher speed system RAM affects framerates dramatically?
Click to expand...

I love their products but I question their intent behind that statement. It could be true but no other game has shown any type of benefit so I really doubt it - Corsair is just trying to sell their product and put it in a good light.

If i'm wrong, that's cool but that's basically how I see it. I have a hard time taking the claim from "corsair" seriously, even though I love their products.
 
Brent_Justice said:
I just tried this on my personal installation, and it is affecting me too. Had absolutely no idea. One would assume when you set ultra it would be ultra, but the values do seem to be reversed. Thing is, this is probably happening for everyone, and they don't know it and are also assuming ultra terrain decoration is the best terrain decoration detail.

We used the global setting to set Ultra in our review. I'm having David check it now to see if it affected us too.

Why does BF4 still feel like we are playing a Beta version? So many bugs.
Click to expand...

Unfortunately, the suits have figured out that many folks will grumble about a bugged product but still pay for it. So they take a release it now patch it later approach.
 
Keep in mind also, we tested in the Shanghai map, which is a city map, with little to no grass vegetation that would be impacted by the Terrain Decoration option.
 
Volkum said:
Same here. I have absolutely zero reason to "upgrade" if there isn't a performance benefit to be had.
Click to expand...

Some demonstrated meaningful benefits in the Beta and I've not see reports of large differences between Win 7 and Win 8.1 in the live patched version.

It seems to largely depend on who you ask. Reports are mixed.

http://www.overclock.net/t/1375478/...rmation-discussion-thread/10620#post_21165645

If there were a 10 FPS variation in mins and avgs, I'd think about it. Personally, I wouldn't change my OS for 3-5 FPS.

Right now, my advice to people on Win 7 would be to disable core parking via the registry and call it a day.

For people who aren't aware of how to do that; run regedit as an admin and use find to locate specify the minimum number of unparked cores/packages
allowed (in percentage), match the ValueMax to the ValueMin. It should be set to 0. Use F3 or find next to adjust all other instances that appear in the
same manner. Restart your computer.

I'm waiting for [H]ard's verdict on the issue. Then I may change my mind.
 
Last edited:
nice review David and Brent, it was a good read for me on a lazy sunday night. i'm really curious to hear what you find with the terrain decoration wonkiness. i set mine on Ultra assuming that's as good as it gets.
 
Domingo said:
Any comments on Corsair's claims that higher speed system RAM affects framerates dramatically?
Click to expand...

My next article will focus on CPU and memory configuration differences in performances within the game, so hopefully there will be some sort of conclusion that can be reached with regards to those claims.

twzTechman said:
Great Sunday night read. So - with a 770 (680), could you still run at 2560x1600 with some lower settings? Is it really better to drop down to 1080p? It is sort of the same recommendation you made with Tomb Raider.
Click to expand...

With a 770 at 2560x1600 using Low FXAA, you're looking at performance that will be a couple FPS lower than that of it running 1920x1080 at 4X MSAA, thus, ~45 FPS on average. At that frame rate, you're going to have valleys that dip down into the 20's when things go boom. If you can live with that (it might be bearable as it at least doesn't feel like a slideshow), then you can run the higher resolution. I'd say it depends on your overall preference, but those dips can put you at a competitive disadvantage in a MP environment.

KickAssCop said:
Nice article. I would love to see some overclock results as well (be it only apples to apples).
Also, is there IQ difference between AMD or nVidia especially in terms of AA? I saw a noticeable improvement in IQ using nVidia technologies such as SGAA which was available at a huge cost on AMD.

This was when I compared Radeon 7970 to a GTX 680 and found AA to be superior - even MSAA on nVidia. Maybe I am too sensitive to these things.
Click to expand...

We were only using the game provided options for FXAA, MSAA and SSAA. Overall, I did not see a noticeable difference in IQ across the red and green teams. The screen shots that you see within the article were done with AMD hardware.

Priller said:
I would also test the 290 with the 13.11 Beta 9.2 driver to see the change in performance with the increase in fan speed.
Click to expand...

The 13.8 driver that we used for testing includes the 47% fan limit for the R9 290 and I manually verified that the limit was set there prior to testing. The change to 9.2 changes the driver of the fan from being a percent based throttle to an RPM based throttle.

WinSomeLoseNone said:
The patch that dropped a few days ago was the first major patch that dealt with memory leaks, framerates, client and server stability, CPU Usage.

If this testing was performed prior to the patch (Which is probably the case) I would certainly take them with a grain of salt.

The 780ti results, if performed after the patch, are going to be very skewed.
Click to expand...

The data within this evaluation was collected prior to the 11/14/13 client side patch, thus, it should all match up with itself. Going forward, we will use fresh datasets to make sure that an equal patch level is in play - thus, when we do collect 780Ti data, it will be compared against equally patched data from a few days ago.

Though, for many users, the latest patch has made it a rather difficult venture to even play the game...
 
David_Schroth said:
The data within this evaluation was collected prior to the 11/14/13 client side patch, thus, it should all match up with itself. Going forward, we will use fresh datasets to make sure that an equal patch level is in play - thus, when we do collect 780Ti data, it will be compared against equally patched data from a few days ago.

Though, for many users, the latest patch has made it a rather difficult venture to even play the game...
Click to expand...

This is great to hear!

Re-benchmarking all those cards in future reviews to provide accurate results is much appreciated!
 
David_Schroth said:
The 13.8 driver that we used for testing includes the 47% fan limit for the R9 290 and I manually verified that the limit was set there prior to testing. The change to 9.2 changes the driver of the fan from being a percent based throttle to an RPM based throttle.
Click to expand...

Which is a difference of 2-300 rpm and that could make a difference.
 
You must log in or register to reply here.
Back
Top