Best CPU For Gaming – 9 Processors And 8 Games Tested Milos Milosevic September 28, 2014 Benchmarks We decided to conduct thorough testing and try to establish an answer to this never ending dilemma: What is the Best CPU for Gaming ? “Which CPU should I buy? Intel or AMD? What about overclocking, is there any use of it at all and if there is, is it worth giving extra money for a more expensive overclockable CPU?” So many questions that need answering, and answering only one of them is a difficult task. PC gaming is on its turning point as next-gen consoles have been released, previous generation consoles are still here and will probably stick around for a year or two at least. How does this affect PC gaming you ask , since game developers usually do not bother optimizing games when porting them from consoles, this usually means that games only use 2 CPU cores/threads, or in the best case scenario 4 cores/threads. And yes next-gen consoles have an eight core CPU however, if you have read our previous benchmarks you should know that optimization isn’t perfect with next-gen consoles and games. As long as old consoles are still here and make the majority of the market, improvements when it comes to optimizing console ports is not in sight. Most users are confused when looking to get a new CPU since it’s hard to pick a long lasting solution without the fear of wasting money, which is quite valuable in these troubling times. Not an easy time to pick a CPU, then again it never was. Testing Methodology and List Of Games Choosing a list of games to benchmark and the pre-sets to use wasn’t exactly an easy thing since some games, like Assassin’s Creed 4: Black Flag or COD: Ghosts are capped at 60 or 90 FPS meaning that we have to max out the settings in order to decrease the number of frames and avoid that retched cap. This changed our initial idea of testing the games at a 1080p resolution at medium settings, since we tried to simulate the real life situation as best we could. 1080p monitors are a standard for some time now as well as most gamers not maxing out their settings due to incapable hardware. The same thing goes for 1440p since it would only decrease the number of frames per second, and therefore decreasing or even diminishing the margin between processors due to heavier GPU rendering, which is something we wanted to avoid. Starcraft 2 ARMA 3 Call Of Duty: Ghosts Metro: Last Light Assassin’s Creed 4: Black Flag Sleeping Dogs Tomb Raider Battlefield 4 Best CPU For The Money In order to determine which CPU is the best for the money we included a novelty into our testing procedure called Performance Value Index. After careful examination of the data we found that the i3 4130 is the cheapest CPU in our benchmark that gives the most stable and consistent performance, so we decided to give that CPU a 100% performance value index. We calculated index performance for the benchmarked CPUs while overclocked at 4.5GHz (only for those where overclocking is possible of course), since if you are buying Intel’s K or AMD’s FX CPU you will want to overclock otherwise you are wasting your money. But we also tested the CPUs at stock speeds in order to find out whether overclocking gives benefits in games and to which extent. The only CPU that could have been overclocked but wasn’t, is AMD’s FX 8320. Our sample could not reach more than 4.2GHz which is slightly better than the default frequency of the CPU running at 3.5GHz (4GHz turbo). Since the FX 8350 runs at 4.2GHz in Turbo mode there wasn’t really any reason in overclocking the FX 8320, since a FX 8350 gives exactly the same performance at stock. All of the other indexes for CPUs have been tested and calculated in comparison with the i3 4130 index performance value. Of course we chose the best prices for each CPU that we could find on the market. These are the prices that have been used when calculating Performance Value Index: i7 4770k – $320 i5 4670k- $222 i5 4570- $195 i3 4340 – $160 i3 4310- $120 FX 8350- $199 FX 8320- $140 FX 6300- $110 FX 4300- $97 You will find a detail description of pre-sets and the testing methodology for each game we have benchmarked. Also we added a Performance Value Index for each game which was calculated based on the performance in-game and the prices in online stores. Test Systems Test System Intel CPUs :i7 4770K - i5 4670K - i5 4570 - i3 4340 - i3 4310 AMD CPUs:FX 8350 - FX 8320 - FX6300 - FX4300 RAM :2x 8GB@ 2133 Mhz Kingston HyperX Beast : KHX21C11T3K2/16X SSD :2 x Intel 520 Series 240 GB GPU :GTX770 4GB MONITOR :Shimian QH270-IPSMS 2560x1440p 27" PSU :Thermaltake ToughPower XT 775W OS :Windows 8.1 Pro x64 Starcraft 2 CPU Benchmark For benching Starcraft 2 we used Unit Tester 0.97. This mod is very useful for players who want to perfect their in-game combat skills but it is also useful for benchmarking since you can put a large amount of units to collide. We decided to test one of the most extreme situations that you will ever encounter in Starcraft 2, 125 Zealots and 25 Medivacs on each side which gives a total of 250 active units. Of course all of that was run at Extreme settings as you can see from the image. Most of the time Starcraft 2 is not a demanding game and can run on pretty much any CPU locked at 60 FPS, however we wanted to investigate how CPUs behave in some of the most stressful situations possible and we believe we found the right measure. This scenario is likely to happen in 4 vs 4 large scale battles. Starcraft 2 Min-Avg-Max FPS Starcraft 2 Average Frametimes As you can see from the frame rates and frametimings, AMD’s CPUs are sluggish, even the overclocked ones. It is interesting that the i3 CPUs handle the game quite well, not as well as a more expensive i5 or i7 CPU but, the performance is stable. Of course Intel’s i7 4770k and i5 4670k overclocked at 4.5GHz give the best performance and the gameplay with these CPUs is as smooth as butter. Starcraft 2 CPU Performance Index Value However, when we turn to the performance index and what you are getting for your money the situation is different. The i3 4340, FX 4300 4.5GHz and i5 4670k are all at a 78% index performance value in terms of frames/price. But as you can see horrible frame timings with AMD’s CPUs make them less attractive. ARMA 3 CPU Benchmark ARMA 3 is one of the most CPU intensive games on the market since it features 2 islands, Altis (270 square kilometers) and Stratis (20 square kilometers). Having huge maps with a viewing distance of up to 20 km is a massive hit on CPU performance. Testing ARMA 3 in MP surroundings was not possible due to inconsistent frame rates with all CPUs. In that light we went for free strolling and running across the island of Stratis. Even in that scenario Arma showed to be a very demanding piece of software. ARMA 3 Min-Avg-Max FPS ARMA 3 Average Frametimes Intel’s CPUs take the crown again when it comes to performance and the difference is substantial. This is one more game where it’s clearly shown that although being very CPU intensive, having more cores or threads doesn’t make much of a difference (if any). AMD’s FX 8350 and Intel’s i7s do a better job than their slower and less expensive counterparts.When it comes to frame rating Intel again does it better, although 10ms average frametimes for AMD is not a bad result. ARMA 3 CPU Performance Index Value In our performance value index only the FX 6300 and FX 4300 both overclocked at 4.5GHz can compete with the i3 4130 in terms of price/performance. If you want the best performance and you have enough money an i5 4670k is as good as it gets and buying an i7 CPU will not give you many benefits. Call Of Duty: Ghosts CPU Benchmark Call of Duty: Ghosts is one of the games that come with capped FPS. Around 90 FPS is the most you will get out of this game. This is strange since this is a competitive game, where people use 120Hz monitors and 120 FPS is a must. Our initial idea was to test all of the first person shooter games at medium settings in order to get the performance that is most widely used amongst gamers. However with COD: Ghosts we had to make an exception and test on the highest possible settings in order to avoid that unfortunate FPS cap. Also check out our COD:Ghosts Benchmark . Call Of Duty : Ghosts Min-Avg-Max FPS Call Of Duty : Ghosts Average Frametimes The results are more or less identical, proving that you really don’t need an expensive CPU to run this game and get maximum performance. As a matter of fact you don’t really need to overclock. Call of Duty Ghosts is one of those games that can run on pretty much any modern CPU. COD Ghosts CPU Performance Index Value In this case, performance value index is just here to state the obvious. The FX 4300 is the best CPU you can buy for this game when it comes to the price/performance ratio. You won’t be wrong if you buy a FX 6300 or i3 4130. Buying an expensive CPU for just this game would be a clear waste of money. Metro: Last Light CPU Benchmark As we mentioned before, FPS games are tested at Medium settings in order to get the highest frame count per second as possible but still keep good image quality. Metro: Last Light is also one of the most demanding games on the market and it’s here to set an example on how CPU power hungry a single player FPS game can get. Metro: Last Light Min-Avg-Max FPS Metro: Last Light Average Frametimes This is one of the rare instances where the i3 CPUs didn’t perform very well. The FX 6300 beats them on stock clocks as well as when overclocked. The optimization in this game is good, but isn’t perfect. The i7 4770k does better than any other CPU, especially when overclocked but the difference is marginal vs the i5 4670k. Metro: Last Light CPU Performance Index Value The numbers speak for themselves. The FX 6300 gives the best value for money on Metro Last Light. The FX 4300 also behaves well in terms of frames but just plain horribly when it comes to frame rating, so it is best to skip this CPU. The i3 and FX 8320 (on stock) give the best ratio in terms of price and performance on one hand and in-game smoothness on the other. For those who want the best performance, an i5 4670k is all you need, since once again the performance index values indicate that the i7 4770k is a waste of money. Justin interesting article…good for those needing a budget build to get the most out of games etc…. However imo the i7 isn’t a waste of money for gaming…running at 1440p the i7 is nice esp on large multiplayer games…I have had and used just about every chip since 2009 for gaming and the 4770k is hands down a monster 4 core cpu….yes the 4930k is a bit better but totally different budget. Also the 4770k can be found for around 250 or even less so…Anyways thanks! Milos M As you go up with the resolution CPU dependency in terms of performance drops and the difference wouldn’t be that noticeable. Thats why fe when people want to test the true CPU power they test it at 720p or even lower, on,lowest possible settings. Games are now days are mostly optimized to use quad core efficiently (such as BF4) while the games such as Metro: Last light drain every CPU to its limits most of the time. protoo BF4 -fx 6300 stock performance just like a fx 8320 stock o.0, that´s not right, Milos M The data are valid developers at DICE should explain this laziness. Similar story is with i7 vs i5… ARIF TOGOO A very well researched and comprehensive article although I hope the author would consider the fact that games are still not optimized for 8 cores and the true potential of FX 8350 cant be judged by most of todays games. Also why the OC was capped at 4.5 GHz? FX 8350 can do much better at higher clock speeds with proper cooling.. Still a very informative reading. Milos M It has been considered and acknowledged but that is the performance of the current CPUs. When Mantle hopes along situation will probably get better for FX CPU’s but not in all of the games and it will take some time before mantle has been accepted as a standard or at least as an alternative worth while for the developers. When we decided the OC frequency we went for the most viable option meaning that almost anyone can reach 4,5 Ghz on all of the CPU’s. Our test sample can go up to 4.7 but it requires much higher voltage so its not really worth while same thing goes for i7 4770k whose limit is 4.6 GHz (our sample). In real life situation you can hardly squeeze more out of FX and Haswell CPUs… cubs223425 Yes, but by the time that 8 cores start getting used, Piledriver will be obsolete. It could even be argued that it already is, given that AMD is launching Steamroller in a few days (but not an FX CPU). Matt AMD performance is skewed… an 8 core should not perform less than a 6 core. Nikolas Nikolaou It solely depends on game optimization. Most games can’t utilize all cores. cubs223425 I’m about to start reading the article, but the general test bothers me enough to point out what I think is a meaningful flaw in the system. The premise of testing a bunch of these CPUs is great, and I like it. However, I think that most folks aren’t interested in seeing the 4570. I think that more would prefer to see the Ivy Bridge i5-3570K tested against the i5-4670K, to see what kind of difference the generations offer at stock. The same could be said for the i3 choices, since the difference between those two is just a slight clock increase (200 MHz). I’d also be intrigued if you had taken out the FX-8320, and instead offered up benchmarks of the high-end Richland CPU or APU, since the 8320 should be able to perform closely to the 8350, since their difference is again just clock speed (though 500 MHz this time). I think that an i3, i5, and i7 from both Ivy and Haswell should represent Intel, while AMD should offer up a high-end APU, a high-end Richland-based CPU, the FX-6300, and the FX-8350. Maybe, instead of the two Richland chips, you could have gone with one of them and one of the FX-9000 chips. I just don’t see $13 being a meaningful-enough barrier for gaming purposes, and so the FX-4300 is pretty useless here. Then again, that is all just personal opinion based on generational intrigue. Nikolas Nikolaou Thanks for pointing out a few of your personal views cubs223425. We are slowly trying to add more systems into the mix so people can see everything side by side and have a better picture on what to get. My own personal opinion in terms of testing Intels Ivy vs Haswell or in general any very closely related generation of CPUs is pointless. I have swapped from Sandy to Ivy to Haswell and havent seen anything that great in terms of fps jumps to even document. Most consumers stick to the same system for about 2-5 years. We have added the 4570 non K as some people prefer to build a budget system with a cheaper mobo and add a better GPU. The New APU CPUs need to be tested as thats a whole different story, good pointer on that. We try to add as many price ranges as possible as most consumers tend to not spend alot on hardware. Nikolas Nikolaou Hey Sam , I understand your concerns. We have nothing to do with Dice nor EA we due not get “paid” for reviews like most of the sites. Our hardware and games are from our own pockets so I guess thats as unbiased as you can get ( especially when you shell out yourself). With the new Mantle Api 6 core and 8 core optimization will become a reality. When ? we will have to wait and see. I do not understand your rage can you please explain alittle better ? JAGUARCD32X There seems to be rampant paranoia and conspiracy theories with gamers recently as I keep reading similar posts on lots of different websites about people being biased . In any case I have just found this website today and love it so have bookmarked it in my browser tab. This was a good test but I would like you to do another test where you benchmark older I7’s to see if there is any benefit to upgrading if you have older I7’s. I have an I7 2600 non K sadly so it’s running at 3.8ghz and since owning it I have upgraded my GPU three times. I bought a prebuilt which was my first ever PC with the 2600 and a gtx590. I then upgraded to the 690 and now I own a 980 but the CPU seems to be good to go for some time yet but I would like to have seen benchmarks just to see the difference anyway. Is that something you could do as I noticed that you have various brands of AMD CPU?. You could start with the first I7 right up to newest Nikolas Nikolaou I don’t know why we should be biased towards any manufacturer or publisher when we pay full price on all hardware and software like every consumer out there. People should get similar results on the same systems and drivers with very little variance. Its all in the data. I have answered some concerns below. JAGUARCD32X I agree, I really like this site and don’t really see what people are basing these accusations on but it seems to be a common theme running across many different websites to squeal “Teh bias” any time someone doesn’t like the results. Yashtir Gopee By CPU Performance Value Index, lower is better or higher? Nikolas Nikolaou Higher is better. You basically get better bang for your buck. Daniel I have an inspiron 570 with alomost every thank upgraded but the processer, I need know wich processer (Gaming) will work with my computer and motherboard. Here is a list of parts I have upgraded so far: 1. Power Supply Upgraded 2. GPU Upgradeed ( NVIDIA ) 3. Hard Drive Upgraded ( Mutable ) 4. Memory Modules Upgraded (3 out of 4) 5. Internet Card Upgraded ( T-Link) 6. Still deciding ( CPU ) I Need To Know! Milos M You need to be much more specific. What is your budget? What PSU and GPu do you have? Monitor resolution? And how much memory (in GB) do you have? Ahmed AL-Jaber of course no game can utilize too many core cuz it just slows the game down, its different story with consoles since the consoles is HIGHLY optimized with spesific hardware Nikolas Nikolaou If the game is optimized to use more cores it doesnt slow the game down, it helps get better and smoother performance on more powerful hardware. You can’t change resolutions and settings on consoles as they are limited by the type of hardware thats available when creating a game. JAGUARCD32X I just found this website today and I love it so am going to use this site a lot from now on. I liked this article but what I would really like to see is benchmarks of older I7 CPU’s as I own an I7 2600 Sandy and I would like to see how much the new Devil’s Canyon has improved. I am not looking to upgrade my I7 right now of course as there is no game that gives it any issue. In fact highest recorded CPU usage I have seen is 60% and only for little spikes during Crysis3. On average my CPU is only used around 40% and this is when running games with at least 60fps. I think it’s great that CPU’s have long lives which means we only need a GPU upgrade to keep things going Nikolas Nikolaou As mentioned previously I swapped from a 2600K to a 3770k then a 4770K, currently on a 3930K due to running multiple monitors/gpus and now 4K . At stock they all provided similar performance.There might be a slight difference vs K processors but only with games that utilize hyperthreading, with a GTX980 it shouldnt be noticeable at all. The higher you go on resolution and settings the less CPU dependant you will be , going to two cards or more you might need to change your cpu though. Its always recommended to run a balanced system (older hardware is not ideal running with newer or more powerful). I would stick with your current CPU though as it still has some life in it , you can always overclock the base clock if you have a decent motherboard when and if you feel that its not enough. JAGUARCD32X Thanks for the reply. I was just wondering how long a life my CPU should have as I have went through three GPU upgrades in this time but am not noticing GPU performance lower than website benchmarks using the same GPU’s and settings. Obviously that’s a good thing as it’s one less component and motherboard to upgrade but I am just wondering how long it’s going to last. talal Hello, I’m into casual gaming on PC since a year and half and I’m still having troubles to understand all the PC hardwares. Although I’m having a bit of a better idea. What I would like to know is how much difference is really significant in FPS between a i3 and a i5 (to keep it simple, I won’t ask about AMD). Few games adopt a multicore utilization so I feel like there’s no need to get a i5, let alone a i7, which would be a waste if only used for gaming. The i3 does a great job for the purpose. However, do you think that the difference is anything of significant? Directx12 is coming very soon (in a year if I’m not mistaken), how big of an impact will it have on CPUs? On multicore gaming? Thanks for reading. Nikolas Nikolaou “Really significant” can’t be gauged without knowing your own personal FPS preference current system and games that you play or are going to playing. If your looking into new game releases most are ports (multiplatform) meaning your going to need those extra cores. If you look at current benchmarks on the site, the XboxOne-PS4 to PC ports use multithreading most of the time.(Not perfectly ofcourse) talal Well smooth 60 FPS ofc is what I need. Anything above is extra. But yeah, I play mostly BF4. But something odd is that the G3258 which is a dual core, once overclocked manages to handle BF4 with no issues at all. It’s probably a particular case and more cores are surely needed with the upcoming games. Hell Y. Hansen I have not read the full article but i have one observation to make concerning battlefield 4 that i have benchmarked myself. You must run the game att max detail setting (= ultra). At medium the cpu load is very much reduced. So run at ultra setting but with anti aliasing turned off och and low screen resolution then you get true numbers. Or run it at high setting which does not use AA. Still great work.