6990 Poor Performance on BF3

F

Flekx

Limp Gawd
Joined
Apr 29, 2012
Messages
229
Over the past week, I've done just about everything I could think of to resolve this 6990 performance issue:

-Reformatted entire SSD
-Tried out the following CC drivers: 12.2, 12.3, 12.4, 12.6, 12.7
-Swapped out PSU's for two others
-Swapped out RAM modules
-Run 2500k at stock and at 4.5 ghz OC
-Run it on AMD Phenom II X4 955

It's still under performing my GTX 670 which doesn't make sense at all.

My Furmark benchmarks are running at 78 FPS on the Burn-In test which is pretty normal.

Thanks in advance for your help!
 
Last edited:
TFchris said:
it's your CPU, BF3 is very CPU intensive
Click to expand...

Dont think it's his processor. He is saying his 6990 is getting worse performance than the 670 in his system (Using the same processor). Of course a better processor would be better but not what he is going for.

What kind of fps are you getting and what resolution?

http://www.anandtech.com/bench/Product/598?vs=515

According to their benches the 670gtx is in the ballpark and sometimes faster than the 6990 in BF3...
 
I'm getting huge variation between 200 and 97 depending on the maps @ 1920x1080
 
BF3 can run on 4 core, and actually likes more. BF3 runs best with 2600k. it really wants the hyper threading. My 6 core AMD get almost fully used in BF3.
 
doug_7506 said:
Dont think it's his processor. He is saying his 6990 is getting worse performance than the 670 in his system (Using the same processor). Of course a better processor would be better but not what he is going for.

What kind of fps are you getting and what resolution?

http://www.anandtech.com/bench/Product/598?vs=515

According to their benches the 670gtx is in the ballpark and sometimes faster than the 6990 in BF3...
Click to expand...

http://www.tomshardware.com/reviews/geforce-gtx-670-review,3200-5.html

Toms is showing 120+ on ultra. I'm running everything low and I'm not seeing that =/

And according to this data, I should be out performing both the 680 and 670
 
Multiplayer. I've seen videos of other members with similar rigs as mine posting well over 120 on even the new Close Quarters maps.

I also want to note that hyper threading has been shown to produce micro-stutterers which has been confirmed by the competitive community on BF3. I also wanted to cite two benchmark tests by Toms and Bit-Tech which show that CPU cores doesn't have much of an impact on frame performance:
http://www.tomshardware.com/reviews/battlefield-3-graphics-performance,3063-13.html
http://www.bit-tech.net/hardware/2011/11/10/battlefield-3-technical-analysis/7
 
Last edited:
Just my two cents, but I would think the 6990 would shine at a much higher resolution or more demanding settings, and I believe this has been bourne out in a number of reviews.

I also believe that is why the 670 is holding it's own against it.

I assume you have the latest AMD CAPs installed along with the drivers? For the 6900 series of catds I don't think the 2012 sets of drivers offer much advantage, in other words I don't think anything past 12.3 will change a thing.....I would back up to the last 2011 driver set and use the CAPs 12.7 and see what you get.:)
 
featsdontfailme said:
Just my two cents, but I would think the 6990 would shine at a much higher resolution or more demanding settings, and I believe this has been bourne out in a number of reviews.

I also believe that is why the 670 is holding it's own against it.

I assume you have the latest AMD CAPs installed along with the drivers? For the 6900 series of catds I don't think the 2012 sets of drivers offer much advantage, in other words I don't think anything past 12.3 will change a thing.....I would back up to the last 2011 driver set and use the CAPs 12.7 and see what you get.:)
Click to expand...

By the last 2011 driver set you mean 12.3? Thanks for the tip, I'll give it a try when I have the time (will probably report back on Tuesday).

I'm not even getting 120 fps on low so there isn't much room for it to shine on higher settings, seeing as how it can't even handle the lower ones... lol
 
Flekx said:
Multiplayer. I've seen videos of other members with similar rigs as mine posting well over 120 on even the new Close Quarters maps.

I also want to note that hyper threading has been shown to produce micro-stutterers which has been confirmed by the competitive community on BF3. I also wanted to cite two benchmark tests by Toms and Bit-Tech which show that CPU cores doesn't have much of an impact on frame performance:
http://www.tomshardware.com/reviews/battlefield-3-graphics-performance,3063-13.html
http://www.bit-tech.net/hardware/2011/11/10/battlefield-3-technical-analysis/7
Click to expand...

Those are singleplayer benchmarks and they are GPU-limited.
I would suggest you lower clocks by 10% and look how your fps change. If they go down by 10% too, your CPU is the bottleneck.
 
boxleitnerb said:
Those are singleplayer benchmarks and they are GPU-limited.
I would suggest you lower clocks by 10% and look how your fps change. If they go down by 10% too, your CPU is the bottleneck.
Click to expand...

Would you agree with featsdontfailme and also recommend running catalyst 12.3?

Also, based on my limited understanding, GPU scaling is partially architecture and partially clock speed. That's why overclocking suffers from diminishing returns correct?
 
Sorry if it was not clear: I meant changing CPU clocks.

So two possibilities:
a) lower/increase CPU clock by at least 10% -> check if fps change according to that (also about 10%)
OR
b) overclock the GPU and see if you get a respective gain

Overclocking should - if the limit is near absolute - yield a 1:1 return.

I would check all this first before installing new drivers or software.
 
Perhaps you are dealing with dreaded microstutter and crossfire driver issues.. I had dual 6870's that supposedly ran the game over 60 FPS at nearly all times, but it felt like it ran like crap.. It wasn't till I popped in the 670 that I really experienced smooth and flawless gameplay.
 
I've had the same issues. Feels like my 6990 was underperfoming on that game specifically. Everything else, a charm
 
Compared to what you should get (120fps) it's kind of bad yeah. But I see what you mean
 
There is no such thing as a 120fps minimum in a 64 player BF3 MP map on any modern consumer processor and that includes LGA 2011
 
Bumrush said:
Perhaps you are dealing with dreaded microstutter and crossfire driver issues.. I had dual 6870's that supposedly ran the game over 60 FPS at nearly all times, but it felt like it ran like crap.. It wasn't till I popped in the 670 that I really experienced smooth and flawless gameplay.
Click to expand...

No microstutters it feels smooth but not rendering max frames and i can feel the difference
 
Flekx said:
Even all low?
Click to expand...

Yes even with low, becuase all you're doing by going low is redcuing GPU load. CPU load will stay largly the same. If you want 120 minimums in a full BF3 server, you're going to have to wait because the hardware to accomplish that does not exist. I doubt you're even getting 97 minimums, it's likely lower than that. If you are noticing the fluctuations in fps, simply enable vsync and it will lock it in to your monitors refresh rate.
 
I upgraded from a 955be to a 2600k and it made a huge difference in Battlefield 3. BF3 was running my 955 at 4.3ghz around 90%+ usage. My i7 at 4.4 is around 30%. The max didnt really go up by that much but my minimum frames jumped up by 10 - 15fps.
 
RamonGTP said:
Yes even with low, becuase all you're doing by going low is redcuing GPU load. CPU load will stay largly the same. If you want 120 minimums in a full BF3 server, you're going to have to wait because the hardware to accomplish that does not exist. I doubt you're even getting 97 minimums, it's likely lower than that. If you are noticing the fluctuations in fps, simply enable vsync and it will lock it in to your monitors refresh rate.
Click to expand...

Sorry, I also failed to mention that I was testing on an empty 32 man server. Does that affect what you're saying at all?
 
So u are geting 90 fps +?and u say the game runs smooth?well if the game runs butter smoth at 90 fps than the solution is simple: turn off fraps and enjoy the game LOL.
To be helpfull the problem is the low rezolution,u dont buy a 6990 to play in full hd,try on a monitor at 1600 or some eyefinity setup and u will see 670 left behind. Ofc they perform the same at lower rezolutions...
 
I highlighted some benchmarks earlier in the thread showing the difference between the cards at my resolution. I should be getting 120+ all the time but it's frustrating that I'm not and that they're performing roughly the same.
 
Those are average FPS numbers, not minimums and they are also single player not MP benches. You system is running just fine no reason to be frustrated, you're not comparing the same thing Toms is. Take the other guys advise and play the game. Can't fix what isn't broken.
 
cracknub said:
When you have a 120hz monitor.. yeah that's poor.
Click to expand...

Its not poor when he's getting the performance he should be. I really don't see anything wrong with these numbers. MP always has lower FPS numbers than SP. You can't compare the two and say "omg, there is something wrong with my system!"
 
My point is that my 6990 and my 670 are performing roughly the same which is just unacceptable...
 
You still need to compare them accurately if you're going to troubleshoot the issue. You've ran other benchmarks and it performs where it should. If you're going to test BF3 and compare it to what toms got, you have to test like toms did. It could be a limitation of the engine for all you know. Until you compare your hardware accurately to whatever reference you want to use, there's every chance you're barking up the wrong tree. You have the game, you have the card, and you have your reference data. Fire up a single player game, use the same testing methodology they used and see if you get what they got. Until you do that, everything is just guess work and a potential waste of time.
 
Flekx said:
My point is that my 6990 and my 670 are performing roughly the same which is just unacceptable...
Click to expand...

Here's some real experience for you: I went to a single GTX670 from a pair of HD6950 2GB cards, running at 2560x1600; it's the same amount of work as trying to get 120FPS at 1080p.

The result? Night and day. Crossfire is awful for BF3, for whatever reason. The AMD cards absolutely should have been faster, and yet they felt like they were running at half the speed.

Now, I'd say that a single HD79x0 would be similar, it's not an AMD/Nvidia thing, it's a CFX thing, and it's just as bad with the HD79x0 series in CFX as well as reported in the [H] reviews.
 
That's very good news; hell, I wonder if my HD6950's would smooth out with those drivers too, as it is probably an application-specific thing.

Of course, we're still going to have to wait for independent confirmation of sorts, or more in-depth observations from Vega. We need at least to see frame-time measurements that will expose micro-stuttering, if it's still there at all.
 
IdiotInCharge said:
That's very good news; hell, I wonder if my HD6950's would smooth out with those drivers too, as it is probably an application-specific thing.

Of course, we're still going to have to wait for independent confirmation of sorts, or more in-depth observations from Vega. We need at least to see frame-time measurements that will expose micro-stuttering, if it's still there at all.
Click to expand...

2 GPUs in CF give the highest micro-stuttering potential, tri and quad give the least, also the 6 series has the worse micro-stuttering potential between the 5,6,7 series, they got the scaling up with the 6xxx but increased the micro-stuttering potential with it, somewhat resolved with the 7 series.

Also some would say if takes frame-time measurements to notice it then its not worth worrying about and is nothing more than factual analytical data which my be important to some but not for others.
 
Last edited:
DarkTower said:
I upgraded from a 955be to a 2600k and it made a huge difference in Battlefield 3. BF3 was running my 955 at 4.3ghz around 90%+ usage. My i7 at 4.4 is around 30%. The max didnt really go up by that much but my minimum frames jumped up by 10 - 15fps.
Click to expand...

BF3 loves the CPU and when i had a 965 4Ghz it was pegged at 100% because of my quadfire on multi player.
 
BF3 performance does depends largely on the CPU. Benchmarks to the contrary are testing in singleplayer. You need to do a 64 player multiplayer benchmark to see the large difference.

Also as to the 670 performing the same/better than the 6990. Here is an experience I had.

I used to have a single GTX 480 and then I upgraded to SLi GTX 480. When I moved to SLi, my highs in BF3 went up but my lows went down. It's because SLi has additional CPU overhead. I then sold off both 480s and went to a single GTX 680 and saw my lows increase dramatically and my highs pretty much stayed the same as SLi GTX 480s.

The 6990 is a crossfire card so I would imagine it also has additional CPU overhead compared to a single 670.

http://www.sweclockers.com/image/diagram/2506?k=142b45af179c625ccd8f53fea7385155

You can see how much even the additional threads from hyper-threading help. I have not experienced micro stuttering on my i7 930 in BF3 personally. I have noticed higher FPS with it on too.
 
You must log in or register to reply here.
Back
Top