Electronic Arts had PC gamers talking when it released Battlefield 3 for PC on October 25, 2011. Battlefield’s Frostbite 2 not only promised great graphics, exciting game play, and a host of a DLC and expansion packs, but was also touted as perhaps the first FPS shooter to take true advantage of modern multicore processors when running its impressive 64 player online modes.
Battlefield, and its Frostbite 2 engine, are now over a year old. The Frostbite engine has now found its way into other shooters and the Need for Speed franchise. Frostbite is a popular choice – the engine’s ability to shoot out fast paced, multifaceted gameplay combined with great looking DX11 effects is fantastic. Frostbite is only available on Windows Vista and newer, to take advantage of the latest DirectX technologies, and Battlefield’s recommend system specs also call for a 64 bit operating system – improvements over the days of Windows XP, DirectX 9 gaming.
Frostbite, and the game it powers, were immediately touted as the next “system crusher,” and perhaps the first game to leverage Intel’s Core i7 processors and Hyper-Threading technology. Ever since Battlefield’s release, the internet has filled with discussion and debate over just how much CPU you need to have a great experience in multiplayer Battlefield 3.
The answer, depending on your expectations, can be quite a bit. I am here to add fuel to an old fire. I will be revisiting Battlefield 3, but now under Microsoft’s latest Windows 8 x64 operating system using Intel’s Core i3 3220 and Core i7 3770k processors and pairing them with two nVidia GTX 680 video cards in factory overclocked mode from EVGA’s Superclocked line. The twin 680s in SLI mode should work to eliminate GPU bottlenecks, and help to test Frostbite’s ability to hog CPU cores. I will be testing Battlefield 3 on a new map, Talah Marketplace, from its new Aftermath expansion, with the “Premium” add-on content enabled all in 64 player Death Match. I will be comparing the i3′s performance with and without Hyper-Threading technology, and doing the same for the 3770k. I will also be downclocking and overclocking the 3770k to explore performance benefits.
My goals for revisiting EA’s Battlefield 3, and its new expansion packs and add-ons, are to test Ivy Bridge dual and quad core performance when paired with powerful GTX 680 video cards from EVGA running in SLI mode, all under Microsoft’s new Windows 8 operating system. I will be asking questions like: “What performance benefits does a quad core Ivy Bridge offer over a dual core with Hyper-Threading (HT)?” “Is Hyper-Threading effective at improving multiplayer performance?” “Is it overclocking worthwhile?” I will be attempting to obtain strong answers to these questions in the form of performance trends recorded on a Battlefield 3 server with as close to the maximum of 64 players as possible.
I specifically mean to test this equipment on a modern Battlefield 3 map from the Aftermath expansion pack – Talah Market. Although it would have been an ideal stress test to select a modern expansion pack map with vehicles, I was forced to select a server that regularly ran the Talah Market map and dependably hosts at least 50 players to insure consistent testing (in so far as that is possible in a multiplayer environment).
I will be testing performance at both 1920×1200 and 2560×1600 resolutions. I will be revisiting Eyefinity/Surround performance in a future article, but I feel that these two resolutions should provide convenient benchmarks for the majority of single display users. I am opting for 1920×1200 instead of 1080p because it is a 16:10 ratio like 2560×1600, and I would prefer to keep the same viewable area. EVGA’s Superclocked GTX 680s in SLI mode should go a long way toward eliminating video card bottleneck. The power target for these video cards is manually set to the highest point EVGA Precision software will allow. I will be testing Battlefield at the Ultra settings described on the next page.
In order to test Ivy Bridge dual and quad core performance in Battlefield 3 I will be using the Intel i7 3770k chip and the Intel i3 3220. I do not have an Intel i5 3570k on hand, so I will be testing the i7 3770k with Hyper-Threading disabled. The 3770k with Hyper-Threading disabled should be a good indicator of the 3570k’s performance (HT is the major difference between the two), and should give 3770k owners a good idea of HT’s performance.
I will be testing the 3770k at speeds ranging from 3,000 to 4,500 mhz. This range of speeds is designed to give users a good idea as to the benefit of overclocking their processor. Although it is not the express purpose of this test, I will be testing the 3770k at slower than stock speeds to give legacy quad core owners (Sandy Bridge and before) an idea of performance. I do not have information to directly relate the performance of a 3770k to a 2700k, or an i7-920, or a Q6600, etc – but relative performance should be indicative of a need to upgrade, overclock your legacy CPU or make no changes. If the performance of the i7 3770k at 3,000 MHz is less than desired, then it is reasonable to assume that a legacy Intel quad core will provide equal or lower performance at this same speed.
I will be testing the Intel i3 3220 at stock speed with and without HT enabled – this is a much more straightforward test as this processor does not have an unlocked multiplier for over and down clocking.
This is a true multiplayer hardware test. Multiplayer gaming means unpredictable events, and performance spikes. I will be using my information gathered in the following tests to look at performance trends, not exact numbers. I will not max frames, avg frames, etc, but I am more interested in the “area under the curve.” Numbers gathered in this test cannot be taken and directly applied to another system gaming on Battlefield 3 – there is just no way to duplicate every variable of multiplayer gaming. The trends, and “area under the curve” performance information gathered in this test, on the other hand, will be useful in analyzing Battlefield performance.