Which CPU is the right one for ArmA 2? PC Games Hardware tests 18 processors, including Core i7, Phenom II, Core 2 and Athlon 64 X2.
ArmA 2 CPU benchmarks: Introduction
In the military simulation Armed Assault 2 you are sent to the fictional conflict area Chernarus where you have to fight tactically challenging battles as an infantry soldier or with vehicles and aircrafts. The base for this game is the third generation of the Real Virtuality Engine. With maximal details the DirectX 9 renderer proved to be too challenging even for high-end graphics cards (ArmA 2: Graphics card benchmarks
), but also is able to deliver impressive visual quality
Besides the graphics card the processor of a system is also stressed by the Virtuality Engine 3. Especially when there are many vehicles and NPCs (Non Player Characters) visible on the screen the CPU has to do a lot of work.
ArmA 2: Our settings for the CPU benchmarks [Source: view picture gallery]
ArmA 2 CPU benchmarks: Test setup and results
ArmA 2: CPU usage for 8, 4 and 2 cores [Source: view picture gallery]
For our benchmarks we use the initial sequence of the third scenario (Counterattack). With six A.I. soldiers, a tank and a lot of geometry (high draw distance and many buildings) this scene primarily stresses the processor. With practical graphics settings combined with an overclocked Radeon HD 4890/1G - the best single GPU card for ArmA 2 - we try avoid graphics limitation and give the CPU enough space to display its capabilities.
On top of our CPU charts stand the Core i7-920 running at 3.5 GHz (the setup we use for our graphics card benchmarks) while a QX9770 which is set to 3.5 GHz, too, is about 20 percent slower and thus equally fast as an i7-940. The smallest Core i7, the 920 with its default frequency of 2.67 GHz, noticeably beats the Q9650 (4x 3.0 GHz). ArmA 2 likes cores better than megahertz and thus an E8400 (2x 3.0 GHz) is beaten by a C2D Q6600 (4x 2.4 GHz) which itself is only slightly faster than a Phenom X4 9950 BE. The difference between a C2D E6600 (2x 2.4 GHz) and the C2Q Q6600 equals about 19 percent, while the Athlon X2 7850 BE and the Athlon 64 X2 6400+ are on the same level as the C2D E6600. With a single-core CPU ArmA 2 is unplayable - two cores are about 70 percent faster than a single one.
AMD's second generation Phenom CPUs are a little problematic though: An X4 955 BE overclocked to 3.5 GHz is 700 MHz (equal 25 percent) faster than an X4 920 (4x 2.8 GHz) but reaches only 1.1 frames per second more than the smaller CPU - thus the fastest Phenom II is still much slower than the C2Q Q9650. When rechecking the system with Cinebench R10 and Far Cry 2 we received the common, plausible results - the overclcoked X4 955 Be was more than 20 percent faster than the X4 920. Therefore we estimate that the abnormal behavior of the Phenom II CPUs is related to an incompatibility of ArmA 2. Possibly this problem (and several others) are fixed with patches.
ArmA 2 CPU benchmarks: Conclusion
ArmA 2 needs almost more CPU than GPU power and is scaling surprisingly well with higher frequencies and additional cores - nevertheless a single core is too slow. Intel's Core i7 family is dominating the competition with the Core 2 Quad models about 20 percent behind them. The latter ones are about 20 percent faster than equal Core 2 Duos running at the same clock speed. While AMD's Phenom and Athlon 64 processors are able to compete with the Core 2, the Phenom II line is too slow in comparison - although this is obviously caused by the game itself.