How many cores do you think is necessary for gaming?

Hey there guys, hope you are doing well.
I am just wondering what your guys opinion and/or knowledge is about CPU cores for gaming.
Most people say 4 is the "sweetspot" or what you should go for but I realized not everybody has the money for it and will
go for a dual core,, albeit with hyperthreading atleast. If you have a dual core, I was wondering how it performs for you.
Thanks for your input.


Albeit LTT, they make some good points. 4 is nice but 6 would be ideal.

Saw that already but thanks for that. I was just curious what other people think.

I actually just tested this out the other day with a GTX 770. I'll be making a thread with graphs and everything either tonight or tomorrow, so keep an eye out for that.

Anyways, the basic result is that with what I tested, a 2.3 GHz dual core is actually pretty adequate, and a 3.3 GHz dual core up to 4 core 8 thread doesn't have much of a difference. I even tested a single core/thread, but that wasn't a pretty sight.

EDIT: Here's a peek at Unigine Valley results:

At this moment basically you almost never have games that use more than 1-2 threads, so a game will only use 1-2 core at the most. You see it does not really matter how many cores you have on your machine, if the software itself is not coded in a way that actually uses them. Developers avoid doing that though because parallel programming (as it is called) is vastly more difficult to code.

So theoretically speaking even 4 cores is too much at the moment. Of course the single core is a modern 4-core CPU is usually faster than the single core on a dual core that are older, so a modern CPU will still perform better.

Until people manage to create APIs and other tools that make parallel programing easier this will be the case. If this ever happens in the future then it will change things significantly...

2 Likes

This is the only answer. The idea of throwing more cores or indeed physical CPU's at games or software that aren't programmed to take advantage of the additional resources is a waste of money. Save money on a 6 or 8 core CPU and get a better graphics card instead. That will get you the most bang for you buck.

Either 4 or 2 with ht is necessary. Some games will not run without seeing that. Other games turn the settings to potato.

I don't like it, especially since a dual core no ht Intel chip is more than powerful enough to run whatever, but I don't have to like it. Its still a thing.

PC Perspective has also done a more in depth evaluation on what combination of CPU and GPU is more effective for games. From my experience games with a low amount of characters on the screen usually don't need a lot of horse power to run at decent framerates and a good GPU can really make the difference. But if you play a game with hundreds or thousands of characters on the screen a dual core CPU will struggle for sure.

1

more is nice though...

CPU Custom 1.75 GHz AMD 8-core APU (2 quad-core Jaguar modules)
Memory 8 GB DDR3 (5 GB available to games)
Graphics 853 MHz AMD Radeon GCN architecture 768 cores (inside of APU)

OMG...I just saw the LTT video.

The analysis is really bogus. They do not really check if the extra performance comes from the the fact that the 4-core and 8-core have better single thread performance because they are faster than the dual core. The benefit comes from the fact that the single cores on the 4/8-cores are faster than the single cores on the dual core. They give no Specs for the CPUs to help to make this clear.

There is nothing to suggest that multiple threads were used by the games. Just enabling the hyperthreading does not mean that the game will use it. The slim difference when hyperthreading is active is because the CPU will run other background applications in other cores more efficiently, not because the game is executed on multiple cores.

Unless the games tested are explicitly declared by the developers that they can use more than 2 threads, this video is misinformation coming from complete ignorance of computer architecture.

EDIT: If that is the case and the games do not support multi-core execution this video is really pissing me off.

Valley is a graphics benchmark so having a better cpu won't help much with that.

4 cores is pretty much all you need for gaming right now. And if going from an i7 to an i5 means you can go one tier up on your gpu that is almost always worth the trade.
http://www.pcper.com/reviews/Systems/Quad-Core-Gaming-Roundup-How-Much-CPU-Do-You-Really-Need

Depends on the game.
But as a whole.

2 BARE MINIMUM
4 SWEET SPOT
6+ ICING ON THE CAKE

Of course Hyperthreads always helps too :)

Hmm.. okay let's say the most recent triple AAA titles like Star Wars: Battlefront, Witcher 3, The Division (I guess), Far Cry 4 (even though LTT said dual cores are locked, then I would ask for Far Cry 3 since that was the last release, granted its a few years old) ... would these games be playable on a dual core granted we get a great GPU (Titan X or 980Ti, Fury X, whatever the card to avoid bottlenecks)

1 Like

Theoretically yes.

In practice the newer 4-8 cores have faster individual cores inside than the older dual core CPUs. So you will get better performance with them, but not because you are using the extra cores.

Now individual games, that strictly depends on how heavy the game is and if the single core inside a CPU is fast enough to run them.

If companies were making modern gaming dual cores they would be absolutely fine for gaming and equally well perfroming as the 4-8 core CPUs. The do not do it often because they use the extra cores as a marketing buzzword.

Almost no game uses more than 1-2 cores of a CPU.

for most games, 2 cores.

but others I say 6 is the sweet spot.

since I've found games either to be EXTREMELY single threaded, GPU bound or eat every CPU core you can give it.

Depends on your graphics card... Most of the time I'd say 4 but if you're getting up into the 980/fury range an i7 (4 cores 8 threads) would be more ideal.

4 is necessary, there lacks optimization for 2, so the quad core remains

I agree with what you said but those average numbers are incorrect.
15.27 + 83.13 ÷ 2 = 49.2 not 40.9 the same goes for
12.43 + 78.33 ÷ 2 = 45.38 not 41.57
unless you have added some elements which are not displayed on the graph.

Those are the average FPS across the benchmark rather than just the average of the max and min.

Having said that, I'm retesting some Bioshock Infinite because the results don't really make sense. I don't think close to 400 FPS max makes a whole lot of sense.