• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Interesting BF 4 CPU usage [GameGPU.Ru]

Page 5 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Communism said:
Their hand-crafting of the assembly is obviously not good enough or else it wouldn't show any improvements in hyperthreading over non-hyperthreading. It would probably be easier if the programmers coded to the actual RISC co-processors in the cores (they can't of course).
Click to expand...

You're talking complete nonsense. As soon as you need to pull data in from memory, you're not going to hit 100% CPU utilisation with a single thread. Apart from silly toy benchmarks which fit entirely in cache, your code is going to stall when it needs to go out to main memory to fetch data. Hyperthreading means that while one thread is stalled going out to main memory, the other thread can utilise the resources which would otherwise go idle.

Coding to the "RISC coprocessors" (I can only assume you mean decoded micro-ops) would make no difference. If your algorithm has a dataset larger than the CPU's cache, you're going to stall.
 
Shakabutt said:
Im pretty surprised by the numbers, didn't expect my FX 6100 to do this well, or the entire FX line for that matter. 😛 (look at the fx 6300 or fx 8350, insane performance in this game for the money)
Click to expand...

I wouldnt consider that to be a particularly good score, since it would still be handily beaten by a 3 year old i5-760 with a very modest and attainable overclock. Even if both the i5-760 and the FX-6100 were taken to their respective max on-air overclocks, the ancient i5 would still perform much better. The FX-6300 is kind of impressive though. I think that chip vs the old i5 would make for a very competitive matchup.
 
Carfax83 said:
The reviews are all over the place. Some show NVidia with a lead, and others show AMD with a lead..

This one benchmarked the single player campaign though, and the single player campaign isn't nearly as CPU bound as the multiplayer.
Click to expand...

There is no cloud of confusion that we can't clear up when working together. Lets get a guy with an FX cpu and another with an i7, both with similar cards. Lower BF4 details to low, go on same server and pop a squat in the corner somewhere and just stare at the action across the map. That actually makes for a good multiplayer comparison.
Oops, you were talking video cards. I was talking CPUs. Still, I say we test the CPUs like I mentioned.
 
moonbogg said:
There is no cloud of confusion that we can't clear up when working together. Lets get a guy with an FX cpu and another with an i7, both with similar cards. Lower BF4 details to low, go on same server and pop a squat in the corner somewhere and just stare at the action across the map. That actually makes for a good multiplayer comparison.
Oops, you were talking video cards. I was talking CPUs. Still, I say we test the CPUs like I mentioned.
Click to expand...

And that makes a good multiplayer comparison ?? 🙄
 
AtenRa said:
And that makes a good multiplayer comparison ?? 🙄
Click to expand...

Yes it does. You'll find that if you go to a corner of the map with all the action visible, that FPS will drop and remain stable, taxing the CPU a lot. This is a good way to get a reliable comparison between CPU's.
 
moonbogg said:
Yes it does. You'll find that if you go to a corner of the map with all the action visible, that FPS will drop and remain stable, taxing the CPU a lot. This is a good way to get a reliable comparison between CPU's.
Click to expand...

It does if you will like to view the scenery, light your i love you and have a smoke , not to see how the system or CPU will handle multiplayer gameplay. 😛
 
Would show perfectly how well the cpu can handle it, not the system, but we shouldn't be interested in build performance at this level.

8/8.1 seem to be the key for AMD, I'd help out but I'm not wasting money on another EA Madden series.
 
BallaTheFeared said:
Would show perfectly how well the cpu can handle it, not the system, but we shouldn't be interested in build performance at this level.

8/8.1 seem to be the key for AMD, I'd help out but I'm not wasting money on another EA Madden series.
Click to expand...

The Battlefield series is the only way to get Halo multiplayer for the PC after they changed architectures away from x86 for the Xbox(es). Halo 5 will probably come out half arsed for the PC as well. Btw, you are Master Chief for the single player in Battlefield 4, lulz.
 
USER8000 said:
Sweclockers tested the R9 290X in BF4 multiplayer:

http://translate.google.com/translate?sl=auto&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fartikel%2F17810-prestandaanalys-battlefield-4%2F4%23pagehead

They are using the same 32 player map as pclab and at the same Ultra settings used in both revews,the Core i5 4670K and FX8350 are within 10% of each other.

However,if you drop the settings to medium,it looks like the way the CPU is loaded changes.
Click to expand...

These sites really need to detail their full settings list.

BF4 scales processor intensity very fast when going from the default 70 FOV to the normal 16:9 FOV of 90, and gets crazy when you go to 120 FOV (the gamegpu did 120 FOV in their single player tests, showing large differences CPU dependency even there.)

I think they really need to test at at least 90 FOV because any less than that in 16:9 is way too much consolitis.

Also CPU tests need to be done in 60+ player servers.

Do 32 player servers if you want in supplement but 64 player servers is the main draw of battlefield.
 
Last edited:
USER8000 said:
Sweclockers tested the R9 290X in BF4 multiplayer:

http://translate.google.com/translate?sl=auto&tl=en&js=n&prev=_t&hl=en&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fartikel%2F17810-prestandaanalys-battlefield-4%2F4%23pagehead

They are using the same 32 player map as pclab and at the same Ultra settings used in both revews,the Core i5 4670K and FX8350 are within 10% of each other.

However,if you drop the settings to medium,it looks like the way the CPU is loaded changes.
Click to expand...

That is because at the 1440p, ultra, the test is basically gpu limited for the faster processors. At 1440p medium or 1080p, pretty much any setting, the i5 or i7 is about 15 to 25 percent faster.
 
frozentundra123456 said:
That is because at the 1440p, ultra, the test is basically gpu limited for the faster processors. At 1440p medium or 1080p, pretty much any setting, the i5 or i7 is about 15 to 25 percent faster.
Click to expand...

Pclab used the same settings(Ultra) at 1920X1080 and the results profile looked very different. The same map,the same number of players and the same settings.

Moreover,at medium settings I suspect as the card is less pushed,the drivers are probably showing a different CPU load in this scenario. If the test was GPU limited at higher resolutions there would be no difference at all in any of the tested CPUs at 1920X1080 and this is not happening. It is beyond a 5% variance.

It will be interesting to see thread loading at both settings.
 
Last edited:
Communism said:
These sites really need to detail their full settings list.

BF4 scales processor intensity very fast when going from the default 70 FOV to the normal 16:9 FOV of 90, and gets crazy when you go to 120 FOV (the gamegpu did 120 FOV in their single player tests, showing large differences CPU dependency even there.)

I think they really need to test at at least 90 FOV because any less than that in 16:9 is way too much consolitis.

Also CPU tests need to be done in 60+ player servers.

Do 32 player servers if you want in supplement but 64 player servers is the main draw of battlefield.
Click to expand...

Agreed,especially on the 64 player maps.
 
USER8000 said:
Pclab used the same settings(Ultra) and the results looked very different. Moreover,at medium settings I suspect as the card is less pushed,the drivers are probably showing a different CPU load in this scenario. If the test was GPU limited at higher resolutions there would be no difference at all in any of the tested CPUs at 1920X1080 and this is not happening.
Click to expand...

Huh?? Your very last statement and the results in this test are classic symptoms of being gpu limited. That is, very close results at high resolution and image quality, while differences between cpus show up at lower image quality or resolutions.
 
Communism said:
These sites really need to detail their full settings list.

BF4 scales processor intensity very fast when going from the default 70 FOV to the normal 16:9 FOV of 90, and gets crazy when you go to 120 FOV (the gamegpu did 120 FOV in their single player tests, showing large differences CPU dependency even there.)

I think they really need to test at at least 90 FOV because any less than that in 16:9 is way too much consolitis.

Also CPU tests need to be done in 60+ player servers.

Do 32 player servers if you want in supplement but 64 player servers is the main draw of battlefield.
Click to expand...

USER8000 said:
Agreed,especially on the 64 player maps.
Click to expand...

AtenRa said:
Where does it say they benched on a 32 player server ??

As for the settings, they used the preset Medium and Ultra.
http://translate.googleusercontent....g=ALkJrhjjlBHp6KxtmWiycpGkQthn1StaDg#pagehead
Click to expand...
I am one of the main contributors to the SweClockers article (I hate Siege of Shanghai now..), and without trying to hijack the thread I just want to clarify some things. There is always a lot of information lost in translation, especially as Google Translate is less than stellar sometimes..
😉

* All our tests took place on the 64p version of "Siege of Shanghai" during real multiplayer on real servers. I can't swear we had 60+ players all the time, but I can almost guarantee we played on servers with 50+ almost every benchmark run.

* We outline our benchmark methodology and our settings as detailed as we can (with screenshots!). Of course we couldn't do identical multiplayer playthroughs, but we tried to compensate by collecting data during pretty long timeframes (at least 3x 180 sec). If our three runs didn't give a plausible value, we did a few more until satisfied.
http://www.sweclockers.com/artikel/17810-prestandaanalys-battlefield-4/2

* If you read the comments below the graphs (I know, translations..), we actually doesn't call "a winner" when the FPS difference is to narrow. The nature of multiplayer is to random, so a few FPS must be considered within margin of error. We also try to point out the CPU and GPU bottlenecks in the different scenarios.

Again, I am sorry for "hijacking" the thread a bit, I just wanted to clear some things up. Happy hunting in BF4!
🙂
 
Last edited:
I'm sorry but I have to laugh at that Gamegpu benchmark, yes I do think they're fake, as are Tek Syndicate's. Between my Intel and AMD system BF4 runs significantly better on my Intel system. I'm a long time AMD fain since the mid 90's, so if you're into gaming you're probably better off with an Intel for modern games.

Also lol at that FX 8350 @ 5GHz barely matching an i7 4770K at stock clocks... wth?
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top