Battlefield 4 Beta Performance Preview @ [H]

I played the game for at least 5 minutes on Windows 8.1 and this is what my task manager looked like:

bf4_usage.png


CPU info is in the image (i7-950 w/HT), GPU is GTX 780 with 331.40 driver.

Don't know what all the CPU usage hubbub is about. Windows 8 isn't really that bad. The Start screen... IS A FULLSCREEN START MENU. Hit the Windows key -> BAM! DESKTOP!
 
Can't fix it unless DX 11.1 comes to Win7

I meant the overall performance on Windows 7. Not talking about missing features. Your not going to screw over your largest user base by not optimizing Windows 7 further...and not going to buy Windows 8 for one game...
 
I played the game for at least 5 minutes on Windows 8.1 and this is what my task manager looked like:

bf4_usage.png


CPU info is in the image (i7-950 w/HT), GPU is GTX 780 with 331.40 driver.

Don't know what all the CPU usage hubbub is about. Windows 8 isn't really that bad. The Start screen... IS A FULLSCREEN START MENU. Hit the Windows key -> BAM! DESKTOP!

that's basically maxing your proc out. hyperthreading is the equavalent of 8 half power cores. if you disabled it the graph would show 100%.

most people either don't have a proc with ht or have it disabled.
 
When mantle comes out you might be able to stay with windows 7 and get the same benefits and more. Have to see. It will bypass dx11.1.
 
I meant the overall performance on Windows 7. Not talking about missing features. Your not going to screw over your largest user base by not optimizing Windows 7 further...and not going to buy Windows 8 for one game...

Considering that BF3 sold nearly 10 million on the PC alone (a staggering figure for a PC version) and many of those 10 million upgraded their PC solely for BF3? I'd say a few will update to windows 8.1 for BF4. Windows 8.1 is actually a great product, i'm not sure why there's a negative stigma attached to it.

That said, i'm not a proponent of Windows 7 x64 performance being lower. Yet by the same token, I don't think that will be fixed unless MS allows DX11.1 in windows 7 as Brent mentioned. I can't see that happening. It is what it is. MS has consistently done this with DirectX over the years (to persuade those upgrading), I certainly am not a proponent of this practice.
 
Windows 8
GTX 770 4GB
i5-4670k
16GB RAM

No stuttering, everything looks and feels great, can play for hours. Glad I spent the extra $50 on the 4GB version of the GTX 770.
 
It looks like people in here forgot what Beta stands for.

Due to the fact that BF4 is optimized for AMD GPU's you need to give Nvidia some time to do the coding to smooth things up, I mean the game runs at least 30% faster than AMD cards.

This is clearly a driver issue, no need to panic folks, the game is not even out yet.

While I am sure they will get "some" performance increase (that usally happens in nearly every game) I think you are under estimating the amount of work AMD has put into this title with DICE..Nvidia isn't going to be able to close the gap with driver improvements alone, especially if Mantle ends up being ~50% as good as they (AMD and DICE) are claiming..

I just wish that Mantle would come before December..If it did and offers a ~30%+ improvement, I wouldn't have anywhere near as much need for a 2nd 7950..When you add in the fact that X-Fire/SLI usually limits your O/C'ing headroom vs a single GPU, that is gonna make one sad panda unless I can find another Golden Sample that will do 1.3Ghz:(:(..

IF Mantle doesn't get the performance increase up enough and I end up needing another 7950, then all the retail channels will have sold off their stock by December..That is going to make it hard to get a good deal on a used one..It would suck to have to pay $225-250 for a used one in December when you get buy a brand new one now with 4 Games for ~$185-225 after MIRs...

There won't be any difference between the two versions of windows.

While it remains to be see how much performance Nvidia can squeeze outta their drivers, this statement here is simply Wrong. Very Wrong. Numerous users have already pointed this out to you...DICE themselves recommended Win 8/8.1 from the beginning, and the beta shows why..

Direct X 11.1 is making a pretty big difference so far. Once you add in the rest of the performance killing features (higher res/quality textures etc) that are missing from the beta, I think it will make performance swing even MORE toward 8/8.1..Unless you are running an AMD GCN card and Mantle comes through to help reduce the draw calls/other CPU heavy loads that is..


*SNIP*

CPU info is in the image (i7-950 w/HT), GPU is GTX 780 with 331.40 driver.

Don't know what all the CPU usage hubbub is about. Windows 8 isn't really that bad. The Start screen... IS A FULLSCREEN START MENU. Hit the Windows key -> BAM! DESKTOP!

I think you have things backwards here my friend, which is very easy to do since the Win 8.X hate is strong here...What most are saying is that Win 8/8.1 performs BETTER then Win 7 due to DirectX 11.1 using WAY less CPU power..

Take your exact same setup with Win 7 instead of 8, and you will go from using ~54-55% to 85~100% CPU usage..That is a pretty HUGE difference, especially for those with older CPUs, sort of like yourself...

BTW, that is a pretty decent O/C (4.2Ghz) on that 950 (assuming that is all cores and not a "Turbo" of 1-2 cores)..Most of the 950's were extremely crappy @ O/C'ing unless you fed them a TON of voltage..You appeared to have lucked out.
 
Why is it just because people don't want to upgrade from a perfectly working Windows 7 they are all labeled Windows 8 haters now? So much insecurity...:rolleyes:
 
Erm, I certainly didn't say that. I'm just saying that MS doesn't do free DirectX upgrades when a new Windows version is released - they didn't for XP and they didn't for Vista. As I mentioned i'm not a proponent of this, and that seems to be the entire issue with regards to BF4 beta performance (optimized for DX 11.1).

I think most here don't care whether you get Windows 8 or not ;)
 
While I am sure they will get "some" performance increase (that usally happens in nearly every game) I think you are under estimating the amount of work AMD has put into this title with DICE..Nvidia isn't going to be able to close the gap with driver improvements alone, especially if Mantle ends up being ~50% as good as they (AMD and DICE) are claiming..

I just wish that Mantle would come before December..If it did and offers a ~30%+ improvement, I wouldn't have anywhere near as much need for a 2nd 7950..When you add in the fact that X-Fire/SLI usually limits your O/C'ing headroom vs a single GPU, that is gonna make one sad panda unless I can find another Golden Sample that will do 1.3Ghz:(:(..

IF Mantle doesn't get the performance increase up enough and I end up needing another 7950, then all the retail channels will have sold off their stock by December..That is going to make it hard to get a good deal on a used one..It would suck to have to pay $225-250 for a used one in December when you get buy a brand new one now with 4 Games for ~$185-225 after MIRs...



While it remains to be see how much performance Nvidia can squeeze outta their drivers, this statement here is simply Wrong. Very Wrong. Numerous users have already pointed this out to you...DICE themselves recommended Win 8/8.1 from the beginning, and the beta shows why..

Direct X 11.1 is making a pretty big difference so far. Once you add in the rest of the performance killing features (higher res/quality textures etc) that are missing from the beta, I think it will make performance swing even MORE toward 8/8.1..Unless you are running an AMD GCN card and Mantle comes through to help reduce the draw calls/other CPU heavy loads that is.

I guess you missed the part where Nvidia Cards beat AMD cards in BF4 in terms of performace, lol.
 
Erm, I certainly didn't say that. I'm just saying that MS doesn't do free DirectX upgrades when a new Windows version is released - they didn't for XP and they didn't for Vista. As I mentioned i'm not a proponent of this, and that seems to be the entire issue with regards to BF4 beta performance (optimized for DX 11.1).

I think most here don't care whether you get Windows 8 or not ;)

That wasn't really directed at you and I don't care what DX version is on which Windows version. They need to optimize the game for their biggest user base...is that so hard to understand?
 
I guess you missed the part where Nvidia Cards beat AMD cards in BF4 in terms of performace, lol.

I don't think you read the conclusion of the article. That certainly isn't what "really" happened. ;) Go back and read the concluding comments. The author notes that the game ran better on AMD hardware despite the FPS difference (stuttering on NV cards). With that said, I re-iterate that this is due to being played on Windows 7. The performance would be more normalized on Windows 8.x and DX 11.1.

After logging hours of play time in Battlefield 4 Beta using both the Radeon R9 280X and GeForce GTX 770, the AMD Radeon R9 280X appears to deliver a far superior gaming experience compared to the NVIDIA GeForce 770 GTX no matter what the framerate graphs show. Since the game is so new, there might be a significant amount of performance that could be unlocked with driver updates from both NVIDIA and AMD that could quickly change the observations that we have today. The full-version game could also change this game in big ways, as all the graphics options come into play.

As mentioned though, this is a beta. The performance will certainly change, and NV will ensure that everything runs 100% on their hardware. The only question in my mind is whether Windows 7 performance will suffer as a result of the game using DX 11.1. I would hope not.
 
I think you have things backwards here my friend, which is very easy to do since the Win 8.X hate is strong here...What most are saying is that Win 8/8.1 performs BETTER then Win 7 due to DirectX 11.1 using WAY less CPU power..

Take your exact same setup with Win 7 instead of 8, and you will go from using ~54-55% to 85~100% CPU usage..That is a pretty HUGE difference, especially for those with older CPUs, sort of like yourself...

BTW, that is a pretty decent O/C (4.2Ghz) on that 950 (assuming that is all cores and not a "Turbo" of 1-2 cores)..Most of the 950's were extremely crappy @ O/C'ing unless you fed them a TON of voltage..You appeared to have lucked out.
I tried to OC to 4.0GHz but it kept crashing. I wanted good round numbers so I wanted my RAM to run at 1600MHz and x20 CPU multiplier (or was it bclk?) just wasn't working. Bump to x21 and adjust whatever numbers (it's been a while) and it's stable at 4.2GHz. Apparently some of these don't like the x20 multi at all, bump to x21 and it shuts up and runs.

I was being a bit sarcastic about the CPU usage since mine didn't look bad. Just trying to poke fun at the people that are scared of Windows 8.
 
I can attest to the high ram usage in Beta.

I have 8 gigs currently and unless I close everything running in background, BF4 beta will eventually crash saying I ran out of available memory.

8 Gigs clearly not enough I'd say.
 
I can attest to the high ram usage in Beta.

I have 8 gigs currently and unless I close everything running in background, BF4 beta will eventually crash saying I ran out of available memory.

8 Gigs clearly not enough I'd say.

are you running 8 or 7?
 
Something very odd about the results here, 280X should be doing much better than what is shown in the article. It almost looks as if you benchmarked a 7950/plain R9 280 as the results line up better that way.

hzGsoPU.jpg
HKoig19.png


QNiFGSd.jpg
ssOA0Ij.png
 
Last edited:
Ram usage sounds like a classic memory leak scenario, slowly getting worse the longer you play. I'm sure they will iron that one out during the beta.

"However, in reality, some of the slowdowns that we experienced on the GTX 770 (that we did not feel on the R9 280X) felt like the issue was related to an insufficient amount of video memory"

I'm not sure exactly what you mean here, you're saying the card (770) "felt" like it was running out of memory but it wasn't maxing out the VRAM? How could you determine that performance issues are caused by lack of memory not something else entirely (like frametime!), especially when the data says that the 2gb memory is not maxed out? Seems very speculative (I noted the needs to be fleshed out part).

"Considering that NVIDIA started its entire frame pacing / frametime investigation based on what HardOCP "said,"

Wow, impressive I was under the impression that Scott Wasson was the driving force behind quantifying these issues, any chance of more info on this / links?

I know you hate hearing this but I agree that frametime data would be most useful, it could very well clarify or at least narrow down what is going on with the 770. Maybe if it was used very specifically and sparingly you could avoid adding unreasonably to your benching workload?

Thanks for the preview.
 
Memory leak or not, now I really wish I bought two sets of the Samsung kit haha :/
 
Something very odd about the results here, 280X should be doing much better than what is shown in the article. It almost looks as if you benchmarked a 7950/plain R9 280 as the results line up better that way.

I don't think you can compare results from benchmarks with a different methodology. There is not a particularly strong correlation between the 770 results from [H] and the ones you linked either. I have enough faith to assume they used the correct cards :p
 
Something very odd about the results here, 280X should be doing much better than what is shown in the article. It almost looks as if you benchmarked a 7950/plain R9 280 as the results line up better that way.


test systems are fairly different, 3970x @ 4.9 with DDR3 2400 vs 3770k @ 4.6 and who knows what ram speed, 300mhz + 2 extra cores with quad channel ddr3 2400 =p
 
Since tonight's BF4 update performance on my machine has doubled and all choppiness/stutter is no longer apparent. I was running about 40fps average with constant spikes to 3-5fps every few seconds, now that's about 70fps average with pure smooth gameplay on ultra/1080P.

Running two 2GB 660Ti's in SLI, both MSI power ed running with stock OC, Vram usage is about 1650MB per GPU.
 
Something very odd about the results here, 280X should be doing much better than what is shown in the article. It almost looks as if you benchmarked a 7950/plain R9 280 as the results line up better that way.
Multiplayer benchmarks tend to be all over the place. Different sites will have different results, depending on the areas of the map they conduct their tests.

The gtx 770 not feeling as fluid in these tests, is probably the lack of v-ram or windows 7. My 7950 and 770 4gb, run this game buttery smooth in both of my systems (both have windows 8).
 
Something very odd about the results here, 280X should be doing much better than what is shown in the article. It almost looks as if you benchmarked a 7950/plain R9 280 as the results line up better that way.

[SNIP]http://i.imgur.com/hzGsoPU.jpg[/IMG] [SNIP]http://i.imgur.com/HKoig19.png[/IMG]

[SNIP]http://i.imgur.com/QNiFGSd.jpg[/IMG] [SNIP]http://i.imgur.com/ssOA0Ij.png[/IMG]

The sticker on the card said it was a R9 280X Engineering Sample and GPUz confirmed the card type, clock speed and memory.

I'm not able to read Russian, but at a quick glance, I'm not able to confirm that their methodology and base test rigs match ours. Given how CPU intensive this game is, their hexcore processor alone could be the reason for their higher numbers.

Ram usage sounds like a classic memory leak scenario, slowly getting worse the longer you play. I'm sure they will iron that one out during the beta.

"However, in reality, some of the slowdowns that we experienced on the GTX 770 (that we did not feel on the R9 280X) felt like the issue was related to an insufficient amount of video memory"

I'm not sure exactly what you mean here, you're saying the card (770) "felt" like it was running out of memory but it wasn't maxing out the VRAM? How could you determine that performance issues are caused by lack of memory not something else entirely (like frametime!), especially when the data says that the 2gb memory is not maxed out? Seems very speculative (I noted the needs to be fleshed out part).

Thanks for the preview.

It IS very speculative and it is based upon my experiences reviewing games and video cards during my tenure here. The nature of the performance drops were quite similar to other games where the issue was directly related to VRAM size (specifically, quickly turning around a few times causing major drops). CPU, drivers, game engine optimization and the current phase of the moon could also be contributing factors to the performance drop. However, at this time, we don't have the ability to do a full investigation until the retail game ships (which may have already resolved when it ships).

Keep in mind that frametime is NOT a root cause of poor GPU performance. I do not think that frametime is a good indicator of a game's playability. Poor frametimes can happen as a result of the variety of thingsabove.
 
test systems are fairly different, 3970x @ 4.9 with DDR3 2400 vs 3770k @ 4.6 and who knows what ram speed, 300mhz + 2 extra cores with quad channel ddr3 2400 =p

For the record, its 2x4GB 1600MHz memory at 10-10-10-27-2T timings.
 
The sticker on the card said it was a R9 280X Engineering Sample and GPUz confirmed the card type, clock speed and memory.

I'm not able to read Russian, but at a quick glance, I'm not able to confirm that their methodology and base test rigs match ours. Given how CPU intensive this game is, their hexcore processor alone could be the reason for their higher numbers.
Yeah that makes sense, forgot that this is a multiplayer run through and not a baked demo.
 
The sticker on the card said it was a R9 280X Engineering Sample and GPUz confirmed the card type, clock speed and memory.

I'm not able to read Russian, but at a quick glance, I'm not able to confirm that their methodology and base test rigs match ours. Given how CPU intensive this game is, their hexcore processor alone could be the reason for their higher numbers.

If you can post a picture of said Russian writing, I can translate.
 
...and the current phase of the moon could also be contributing factors to the performance drop.
Frames go in, frames go out, never a miscommunication :D


Keep in mind that frametime is NOT a root cause of poor GPU performance. I do not think that frametime is a good indicator of a game's playability. Poor frametimes can happen as a result of the variety of thingsabove.

Not sure I understand the importance of the distinction. I understand that many factors can cause high frame latency (Drivers, Engine, Architecture, etc, etc.) I think in certain situations it can be as good if not better metric than framerate. Anyway thanks for the response and I respect you guys for always doing things the way you see fit.
 
Something very odd about the results here, 280X should be doing much better than what is shown in the article. It almost looks as if you benchmarked a 7950/plain R9 280 as the results line up better that way.

http://i.imgur.com/hzGsoPU.jpg[/IMG] http://i.imgur.com/HKoig19.png

http://i.imgur.com/QNiFGSd.jpg http://i.imgur.com/ssOA0Ij.png[/QUOTE]

So many different variables here that you cannot compare. Driver, system, game Beta build version, user playability and gameplay style, how much time tested in run-through, path and nature of run-through, objects and players in run-through, map area, number of players, types of gameplay partaken in run-through, and many many more, it is impossible to compare, even more-so when doing multiplayer game testing, so many different and dynamic situations are in play.

I can tell you we used R9 280X with Cat 13.11 Beta, our system specs are in the article, we tested in 5 min run-through, and the test procedure is outlined in the article.
 
Playing with the rig in my sig, mostly at medium-high settings, my system ram usage is about 4Gb on 32 player maps.....
 
Did Hardocp read about all the Win 7 issues with Nvidia for this game? Try the game on Windows 8. The game runs flawlessly on Nvidia hardware with Windows 8.

I don't have any sort of framerate dips during action scenes.
 
Last edited:
I am running an i7-4770k
ASUS Z87-C
Windows 8 64bit
16GB RAM
Nvidia 770 4GB

I can tell you without reservation that performance of the beta is pretty much flawlessly smooth no matter what on my rig at 1920x1080 at maximum game settings.

So whether it's my current cpu/mobo, windows 8, the 4GB video card or a combination thereof, the point is that it's running perfectly.
 
Do you guys have 4GB nvidia card to test with ?

Would be interesting to know if i made a mistake with 2GB or not.
 
I'm going to wait and see what happens. If it ends up running much more poorly on Win 7after its official release, I'm going to skip it. I'm perfectly happy with Win 7, I'm not buying Win 8 for this title.
 
Back
Top