You've seen all the reviews, mine will be a test of the GTX680 in comparison to the GTX470 in various multiplayer maps/modes/settings.
Testing Methodology
This is not the standard bench. In this bench I join a 64 player server that has a few spots open. I make sure there are lots of tickets left, and begin testing. Using fraps, I begin recording the fps after I have deployed and have started driving away from the deployment. From there I continue playing for 10 full minutes. During this time I concentrate on converging on areas where there is action and engage combat as much as possible. I don't spend time capping flags, I just move from where the action is to the next, and keeping on moving between points to see most of the map. My next deployments happen at the busiest locations on the map. My benches are live multiplayer, so it does stress the framerate even more than what benches have shown from reviewers up to now. These results are not official, just a well recorded experience of my multiplayer gameplay on three different maps.
I would like to re-emphasize that these framerates were obtained while in frequent heavy combat. Explosions, more explosions, smoke, fire, tanks, helicopters, deaths and mainly playing facing the center of the map (no pointing to the sky or outwards from teh map where the fps shoots up) This is the result of an online heavy 64 player multiplayer experience. It is pretty much the worst case scenario for any GPU, so during average matches and single player, the framerate will average out to 5-10fps higher than my results on a similar system.
Test System:
i7 860 @ 2.8GHz
MSI GTX470 | EVGA GTX680
EVGA P55 Micro SLI mATX
16G Corsair Vengeance
Corsair H100
Silverstone Strider Plus 850
Crucial M4 256G SSD
Windows 7 64bit
High + 16AF
In this setting the GTX470 crept somewhat close to the GTX 680, considering the performance potential of the GTX 680. Rest assured, this is not the GPU's fault, my CPU is at stock speeds and is bottlenecking the GPU when on low settings. The GTX 470 however, is not being bottlenecked at all.
Ultra no=AA no=HBAO
When the Ultra settings were added along with blur and post processing, it didn't seem to have a very large impact on performance as I thought, but it was enough for the GTX 470 to start feeling slow during gaming. During parts with several explosions and strong interation, the fps dips would start to decrease the smoothness of gameplay on some occasions. The GTX 680 on the other hand, had no issues at all. Here I noticed that the GTX 680 still wasn't being heard... I was used to the GTX 470's "SHHHHHHHH" sound.. Some sound was coming from the 680, but that is because I noticed and paid attention to it. Otherwise, with some low volume gameplay and regular daily family life, I didn't notice it.
Ultra + 4xMSAA + HBAO
4xMSAA and HBAO really separrated the men from the boys. Here the CPU limitation is not affecting the 680 as much, which shows a better comparison between the two and completely destroys the 470 by pretty much doubling its performance during Caspian Border, the ultimate hog in BF3 multiplayer. Here the GTX 470 had some very noticeable slowdowns with dips that were unnacceptable to me. It was very easy to get killed with the 470 at those settings with rough framerates and only occasionnal smoothness during Caspian, and a slight better experience with Kharg. Metro didn't present much problems, other than when I walked outside or when in the subway where the fires are lit where there would be dips and stutters. The GTX680 remained smooth as butter during all gameplay. Hands down no issues at all with the 680.
Testing Methodology
This is not the standard bench. In this bench I join a 64 player server that has a few spots open. I make sure there are lots of tickets left, and begin testing. Using fraps, I begin recording the fps after I have deployed and have started driving away from the deployment. From there I continue playing for 10 full minutes. During this time I concentrate on converging on areas where there is action and engage combat as much as possible. I don't spend time capping flags, I just move from where the action is to the next, and keeping on moving between points to see most of the map. My next deployments happen at the busiest locations on the map. My benches are live multiplayer, so it does stress the framerate even more than what benches have shown from reviewers up to now. These results are not official, just a well recorded experience of my multiplayer gameplay on three different maps.
I would like to re-emphasize that these framerates were obtained while in frequent heavy combat. Explosions, more explosions, smoke, fire, tanks, helicopters, deaths and mainly playing facing the center of the map (no pointing to the sky or outwards from teh map where the fps shoots up) This is the result of an online heavy 64 player multiplayer experience. It is pretty much the worst case scenario for any GPU, so during average matches and single player, the framerate will average out to 5-10fps higher than my results on a similar system.
Test System:
i7 860 @ 2.8GHz
MSI GTX470 | EVGA GTX680
EVGA P55 Micro SLI mATX
16G Corsair Vengeance
Corsair H100
Silverstone Strider Plus 850
Crucial M4 256G SSD
Windows 7 64bit
High + 16AF
In this setting the GTX470 crept somewhat close to the GTX 680, considering the performance potential of the GTX 680. Rest assured, this is not the GPU's fault, my CPU is at stock speeds and is bottlenecking the GPU when on low settings. The GTX 470 however, is not being bottlenecked at all.
Ultra no=AA no=HBAO
When the Ultra settings were added along with blur and post processing, it didn't seem to have a very large impact on performance as I thought, but it was enough for the GTX 470 to start feeling slow during gaming. During parts with several explosions and strong interation, the fps dips would start to decrease the smoothness of gameplay on some occasions. The GTX 680 on the other hand, had no issues at all. Here I noticed that the GTX 680 still wasn't being heard... I was used to the GTX 470's "SHHHHHHHH" sound.. Some sound was coming from the 680, but that is because I noticed and paid attention to it. Otherwise, with some low volume gameplay and regular daily family life, I didn't notice it.
Ultra + 4xMSAA + HBAO
4xMSAA and HBAO really separrated the men from the boys. Here the CPU limitation is not affecting the 680 as much, which shows a better comparison between the two and completely destroys the 470 by pretty much doubling its performance during Caspian Border, the ultimate hog in BF3 multiplayer. Here the GTX 470 had some very noticeable slowdowns with dips that were unnacceptable to me. It was very easy to get killed with the 470 at those settings with rough framerates and only occasionnal smoothness during Caspian, and a slight better experience with Kharg. Metro didn't present much problems, other than when I walked outside or when in the subway where the fires are lit where there would be dips and stutters. The GTX680 remained smooth as butter during all gameplay. Hands down no issues at all with the 680.
Attachments
Last edited: