Well, you gotta love developers. This past year, we’ve been hearing from a lot of teams that their games were taking advantage of more than four CPU cores. And you know what? We decided to put a lot of CPU-bound games to the test. The results are a bit disheartening – to say the least – proving that most PC games are unable to properly take advantage of multiple CPUs.

For this article, we used a stock i7 4930K (with Turbo Boost at 3.7Ghz enabled) with 8GB RAM, an Nvidia GTX690, an SSD (that resulted in no benefits apart from faster loading times), Windows 8.1 and the latest version of the GeForce ForceWare drivers.

So, let’s see how a high-end CPU performs on a variety of games. For obvious reasons, we tested the following games with and without Hyper Threading enabled. Hyper Threading is basically ‘cutting a CPU core in half’, meaning that two threads can run ideally via this method. Hyper Threading is ideal for games that take advantage of more than four-six CPU cores or for games that have more than eight threads enabled by default. In order to also avoid any possible GPU limitation, we disabled SLI in most titles and lowered our resolution to 1280×720 (we maintain ultra/max settings though as there are some Ultra options – in games like ArmA 3 and Hitman: Absolution – that have an impact on the CPU too).

Hyper-Threading Off

Tomb Raider: 72.8fps (specific scene)
Resident Evil 6: 14254 (no performance difference between quad-core and six-core)
Resident Evil 5: 125.4fps (no performance difference between quad-core and six-core)
Lost Planet 2: 97.6fps
The Last Remnant: 267.35fps (no performance difference between quad-core and six-core)
Cryostasis: 118.7fps
Sleeping Dogs: 144fps
Shadow Warrior: 55fps (specific scene, 2fps difference between quad-core and six-core)
Assassin’s Creed IV: 60fps (specific CPU-bound scene)
ArmA 3: 49fps (first showcase scenario, no difference between quad-core and six-core)
Batman Arkham City: 82fps (benchmark)
Hitman Absolution: 70fps (benchmark – 5fps difference between quad-core and six-core)
Saints Row IV: 86fps (sandbox area, no difference between quad-core and six-core)
Civilization V: 49fps (units stress test, 9-10fps difference between quad-core and six-core)
Total War Rome 2: 55fps (benchmark, no difference between quad-core and six-core)
Battlefield 4: 155fps (1280×720, Ultra, specific scene, difference between quad-core and five-core = 10fps)
Crysis 3: 77fps (1280×720, Ultra, specific scene, difference between quad-core and five-core = 10fps)
THIEF: 74.7fps (benchmark, 3fps difference between quad-core and six-core)

Hyper-Threading On

Tomb Raider: 73fps (no gains with HT)
Resident Evil 6: 13780 (performance hit with HT)
Resident Evil 5: 112.4 (performance hit with HT)
Lost Planet 2 DX11: 117.3 (significant gain with HT)
The Last Remnant: 269.10 (no gains with HT)
Cryostasis: 109.5fps (performance hit with HT)
Sleeping Dogs: 141.5fps (minimal gain with HT)
Shadow Warrior: 54fps (specific scene, no gains with HT)
Assassin’s Creed IV: 60fps (specific CPU-bound scene, no gains with HT)
ArmA 3: 48fps (first showcase scenario, no gains with HT)
Batman Arkham City: 82fps (benchmark, no gains with HT)
Hitman Absolution: 70fps (benchmark, no gains with HT)
Saints Row IV: 86fps (sandbox area, no gains with HT)
Civilization V: 49fps (units stress test, no gains with HT)
Total War Rome 2: 56fps (benchmark, no gains with HT)
Battlefield 4: 155fps (1280×720, Ultra, specific scene, no gains with HT)
Crysis 3: 77fps (1280×720, Ultra, specific scene, no gains with HT)
THIEF: 74.7fps (benchmark, no gains with HT)

As it’s pretty obvious from the tests, there is minimal differences between quad-cores and six-cores in most games. Even CPU-bound games like ArmA 3 and Total War: Rome 2 do not benefit when you throw at them additional CPU cores. And once again, we noticed how unoptimized Ubisoft’s games truly are. Watch_Dogs had significant framerate drops and while Assassin’s Creed IV: Black Flag ran with 60fps (without PhysX as those were bringing the game to 40fps), the game mainly stresses only one CPU core even on our i7 4930K.

Shadow Warrior showed minimal differences between a quad-core and a hexa-core, as did Hitman: Absolution and THIEF. Do note, however, that we are talking about only 3-5fps, meaning that those games – while they scale fine on CPUs with more than four cores – are  not really taking advantage of them.

Batman: Arkham City, Saints Row IV, Resident Evil 6, Cryostasis, Resident Evil 5, The Last Remnant and Sleeping Dogs did not benefit from the two additional CPU cores of our i7 4930K, as our performance was similar to our simulated quad-core system.

On the other hand, Battlefield 4, Crysis 3 and Civilization V surprised us as there was a 10fps performance difference between our hexa-core and our quad-core tests. Do note that there were no performance differences between our simulated penta-core and our six-core systems, indicating that there are currently no games that benefit from eight-core CPUs (hell, we couldn’t find one title able to push all of our six CPU cores to their limits).

But how about Hyper Threading? Well, our findings are somehow similar to our previous results. Enabling HT had a negative impact in games like Resident Evil 5, Resident Evil 6 and Cryostasis. Lost Planet 2, on the other hand, saw significant gains when HT was enabled. Crysis 3, Battlefield 4 and Civilization V did not see any performance gain whatsoever.

In conclusion, PC games are still unable to take advantage of more than 4 CPU cores (something we’ve been claiming this whole time). While there are currently a number of games that benefit from a penta-core system, there is no game that can push a hexa-core to its limits. Still, a lot of games are CPU-limited because they can’t put more than four CPU cores to good use. Therefore, PC gamers – with even high-end CPUs like ours – will have to resort to overclocking their machines in order to overcome these CPU optimization issues.