Once we get consoles with better multi-threaded support this kinna stuff won't happen. As long as games continue to run a majority of their tasks on one or two cores you're stuck watching your CPU power go to waste. CPU's are plenty fast enough, it's the game.
Well, like I said earlier, in 64 player maps I can see the frame rate dip under 50. At (and on top) the hotel in Oman, looking over the rest of the map is one problem area.
I'll take some new FRAP's benches of a game when I get a chance.
Even without the hard figures to hand at the moment I just wanted to post that I'm seeing similar to what the OP is seeing.
Works for me. If your frame rates drop on an empty server, why would they be better on a full one? Too bad BF3 doesn't have a meaningful benchmark for this scenario built in.
I love seeing stuff like this. One game that could of been optimized a little better causes so many people to harp about bottleneck this or bottleneck that and end up spending cash upgrading shit just to play one game. It really is hilarious
Heheh
Am finished testing for the day!
I did want to do a run on an EMPTY gulf of oman server just for giggles to see how sli scaling worked when there was no one around.
So what you get below is the exact same 3 benches which are duplicated for SLI off and SLI on.
I spawned at Hotel, ran to construction site then ran back. Exciting or what!!!!!!!????
1a SLI OFF Ultra - Min 51 Max 79 Average 63.593
1b SLI ON Ultra -- Min 77 Max 153 Average 106.786
2a SLI OFF no MSAA - Min 64 Max 105 Average 81.993
2b SLI ON no MSAA--- Min 82 Max 164 Average 115.78
3a SLI OFF HIGH - Min 68 Max 120 Average 84.786
3b SLI ON HIGH -- Min 85 Max 160 Average 115.98
Couple of interesting things.
1. In SLI get an almost identical frame rate with everything set to HIGH as I do with everything set to ULTRA with no MSAA (see 2b vs 3b)
2. SLI scales just dandy, when the server is empty
3. CPU usage is still around 70% even on an empty Gulf of Oman server
You may be able to draw other conclusions?
Don't get me wrong, I'm delighted with my PC, and on the whole BF3 runs superbly in SLI, nice and smooth with a high framerate and full eyecandy. SLI does help a lot to keep the average framerates nice and high and helps the mins. However, there are certain maps that have areas that are tough on me even with the SLI setup and the CPU OC'd to 4.5ghz.
Hope you found the above useful, waiting for the first comment along the lines of 'oh man you spent a bunch are your system is PANTS!' :thumbsup:
In reality though, my SLI benches do appear to match up to what the reviews have out there and the Unigine scores are in line etc.
I just haven't a clue what's going on with BF3 in regards to my system, and obviously, everyone's system is different.
Thanks to you guys who have actually contributed to the numbers here. I find it ridiculous that an empty Oman server still destroys even the fastest CPU's around. An Ivy @ 4.5 getting pegged at 70% on an EMPTY server? What the hell did Dice do? My only guess would be the destruction system, but if nothing is happening then I just don't get it.
BF3 runs fine, but I find that obsessing over its performance is very entertaining. I just hope that any companies releasing games in the near future don't use BF3 as a reason or excuse to tax CPUs even more in future titles, because that would render them unplayable for like 80% of us.
Cheers Moonbog. It is odd though. I know someone with the same CPU as me running a single 680 and they reckon their CPU never goes above 55%. I can't verify that they are using the same conditions I used today, and I guess that's part of the problem, too many variances. Differences in motherboards, hard drives, drivers, you name it.
But it's very good to know that I'm not the only one seeing such an absurd CPU usage on an empty map.
Thanks to both of you for the numbers.
I think one thing that you both need to keep in mind, however, is that with 670SLI, you are no longer GPU-bottlenecked. There is no way the CPU won't work harder under these conditions than with a single card. BF3 was designed at a time when the fastest card out there was the 580. Dual 580s might have just barely been able to push the bottleneck to the CPU, but with the 600 series, we are seeing it in full effect.
There's nothing wrong with this, and nothing wrong with BF3. There are people playing it on much weaker CPUs (by that I mean weaker quads) that are having no trouble playing the game on a 7850, 570, etc.
The curious thing is why is it so damn CPU intensive? Even with 4 modern cores heavily overclocked and on an empty map? If it were better optimized I imagine it may improve everyone's experience, but I admit complete ignorance when it comes to making games.
I'm by no means an expert in game development, but from my understanding, even games that are well threaded like BF3 still rely on one main worker thread and if that main thread is CPU bottlenecked, then you're CPU bottlenecked. I could certainly be wrong about this, but it's what I've read and it seemed to make sense to me.
I doubt that's the case, Bulldozer seems to perform surprisingly well in BF3 if I remember correctly.
disable HT as it doesn't help at all
It does help, been proven in this thread few times. For dualcores (i3) HT is the difference between unplayable and playable, for i7 it does help tiny bit in the most stressful situations.
also in a lot of benchmarks i saw the i7 with HT enabled got lower FPS than with it disabled ?
Just wanted to report I ended up upgrading out of necessity due to my last computer literally going up in smoke. New setup includes i7 3770k @ 4.5GHz, GTX 670 SLI & Z77X UD3H.
Haven't dropped under 60 on Oman, even overlooking from hotel on 64 player server... but I'm still yet to try TV Station on Sharqi which I remember was the biggest culprit for me.