2500K vs 4670K for cpu limited games?

[Techhog]

cool. Well it's settled then i am getting that combo.

I have a gtx 760. 1080P

games like, ARMA 3, ARMA 2, Rome 2, Civ 5 HoI series. Strategy games. etc.

I can say with certainty that you'll get a consistent 60FPS in Civ 5 without overclocking. Better question, what's your current CPU?

 

[BallaTheFeared]

I get similar performance with my 4.8GHz Haswell as I did my 5.4GHz SB chip.

Biggest difference for me was power consumption. Haswell is considerably better on this front despite the continued blunder from Intel on the IHS gap issue (mine doesn't have it though).

 

[tweakboy]

Its a K , meant for enthusiast OCing.

 

[Techhog]

I get similar performance with my 4.8GHz Haswell as I did my 5.4GHz SB chip.

Biggest difference for me was power consumption. Haswell is considerably better on this front despite the continued blunder from Intel on the IHS gap issue (mine doesn't have it though).

Balla and the god chips...

 

[poohbear]

Its very unlikely ure gonna be seeing 4.8ghz on a sandy bridge. Most reach 4.5ish, dont be fooled by people saying 4.7~4.8ghz is easy cause its rare (despite what u might have seen).

For gaming u wont notice any performance increase upgrading from a SB to a haswell, performance was never the focus of haswell, power efficiency was. It was developed with laptops in mind, not desktops. And the SB as desktop cpu was already super efficient, so ure not gonna notice much of a difference in ur power bill.

 

[Headfoot]

4.8 on Sandy is for sure not guaranteed on every chip. 4.8 is probably only realistic for the upper 25% of chips, and that's with at least All-in-one watercooling/high end air. 4.8 will be pretty much impossible for you if this is your first time overclocking unless you get one of the best damn chips made. The last few hundred MHz are the most difficult to get.

So its good you're going with the 4670k. Better than 50% chance you can get to 4.2-4.3

 

[Yuriman]

Its very unlikely ure gonna be seeing 4.8ghz on a sandy bridge. Most reach 4.5ish, dont be fooled by people saying 4.7~4.8ghz is easy cause its rare (despite what u might have seen).

For gaming u wont notice any performance increase upgrading from a SB to a haswell, performance was never the focus of haswell, power efficiency was. It was developed with laptops in mind, not desktops. And the SB as desktop cpu was already super efficient, so ure not gonna notice much of a difference in ur power bill.

Yep, no improvement.

It's closer when comparing average max overclocks, granted.

 

[24601]

Lower power CPUs don't require as beefy mobos to OC.

That's why you could OC Ivy to it's max on the cheapest 4 phase boards.

Haswell is the same way, but be wary of bios quality, some boards manufacturers were really phoning it in on the bios on some z87 boards.

Also the whole "Hyper 212 is enough for 4.8ghz stable" lie needs to stop. Like seriously.

 

[2is]

Lower power CPUs don't require as beefy mobos to OC.

That's why you could OC Ivy to it's max on the cheapest 4 phase boards.

Haswell is the same way, but be wary of bios quality, some boards manufacturers were really phoning it in on the bios on some z87 boards.

Also the whole "Hyper 212 is enough for 4.8ghz stable" lie needs to stop. Like seriously.

The fact that their "max" is also 2-400MHz lower than SB might play a part also... I'd say that's a bigger factor that a slightly more efficient design.

As to the OP, the 2500k is a great CPU, I have one in one of my gaming machines and I've never felt underserved by it even in the most CPU demanding games like BF4. That said, unless you already own it, you're far better off with the Haswell processor. It's better at pretty much everything. It may not OC quite as high, but it really doesn't need to. Not to mention, the OC you think you're going to get out of a 2500k you likely won't, at least not without some exotic cooling, which IMO is a waste of money for an additional couple hundred MHz. I have a Hyper 212 on mine and 4.5GHz is about as high as I feel comfortable pushing it.

 

[JimmiG]

4.8 on Sandy is for sure not guaranteed on every chip. 4.8 is probably only realistic for the upper 25% of chips, and that's with at least All-in-one watercooling/high end air. 4.8 will be pretty much impossible for you if this is your first time overclocking unless you get one of the best damn chips made. The last few hundred MHz are the most difficult to get.

So its good you're going with the 4670k. Better than 50% chance you can get to 4.2-4.3

You may get a 4670K that does 4.8 GHz or an SB that only does 4.2 GHz - Overclocking is not an exact science and there are no guarantees.

 

[poohbear]

Yep, no improvement.

It's closer when comparing average max overclocks, granted.

lol I don't know how they got those numbers, maybe in Russia they're all liquored up on Vodka when they do their tests? Maybe that's the Beta? Its no secret that any 4 core CPU will perform similar to others in the frostbite engine, BF4 is no different. This has been tested to death and i simply don't trust the methodology of that Russian site. Here's another benchmark from a more trusted source that shows what other benchmarks have shown, BF4 RETAIL doesn't CARE about CPU, its GPU limited.

article link:

http://www.techspot.com/review/734-battlefield-4-benchmarks/page6.html

i noticed no difference going from my PhenomII X6 @ 3.9ghz to my 2500k @ 4.5ghz (in BF3 @ 1080), and that's supposed to be a much bigger performance gap than a 2500k & a 4670k.

 

[Essence_of_War]

That's a pretty impressive "You must have this many cores to play" graph!

 

[john3850]

When oc all the i7 2600k to 4770k seem to do the same
amount of work as ipc goes up the speed comes down.

There were a quite a few people here that had 2500k at
1.4v at over 5000Hmz with water cooling which still ran
hot with IBT.

I do very little gaminig but I noticed that both my i7s
will run a game with backround programs with a 30% less
load then my 2500k does.

That 30% less load may be a benifit if your using air.
If you can get a good price at MC go with a 4770k.
I would use a 4670k if I were going to do a delid.

 

[Yuriman]

lol I don't know how they got those numbers, maybe in Russia they're all liquored up on Vodka when they do their tests? Maybe that's the Beta? Its no secret that any 4 core CPU will perform similar to others in the frostbite engine, BF4 is no different. This has been tested to death and i simply don't trust the methodology of that Russian site. Here's another benchmark from a more trusted source that shows what other benchmarks have shown, BF4 RETAIL doesn't CARE about CPU, its GPU limited.

IMAGE REMOVED

article link:

http://www.techspot.com/review/734-battlefield-4-benchmarks/page6.html

i noticed no difference going from my PhenomII X6 @ 3.9ghz to my 2500k @ 4.5ghz (in BF3 @ 1080), and that's supposed to be a much bigger performance gap than a 2500k & a 4670k.

The Russian graph is, as far as I can tell, from the retail version too. I expect they used a different test routine that involved multiplayer with lots of players instead of singleplayer (which was showing BF3 as running fine on old core2quads - but we know that not to be true). As I don't yet own BF4 I can't yet speak from personal experience.

 

[poohbear]

The Russian graph is, as far as I can tell, from the retail version too. I expect they used a different test routine that involved multiplayer with lots of players instead of singleplayer (which was showing BF3 as running fine on old core2quads - but we know that not to be true). As I don't yet own BF4 I can't yet speak from personal experience.

As i said i went from a phenomII X6 to a 2500k and noticed no diff in performance or fps in bf3 in either SP or MP. Ill trust all the other sites that emphasized frostbite engine is GPU dependant, so long as u have a decent quadcore.

 

[Yuriman]

This is what techspot had to say about BF3's CPU needs:

^ They do all of their benchmarks in singleplayer. I'm not about to believe a 1.5-module A6 does just fine in 64player maps.

This is closer to what the CPU scaling is like with 64 players:

Singleplayer benchmarks in a multiplayer game that is known to run more slowly as player count increases mean very little to me.

I refer you to this thread.

 

[Headfoot]

You may get a 4670K that does 4.8 GHz or an SB that only does 4.2 GHz - Overclocking is not an exact science and there are no guarantees.

Yeah exactly, so its unwise to purchase a cpu banking on getting a god chip

 

[nx02nx02]

Well after this thread I was definitely wanting the 4670K but I have a bit less money than I thought I did for this upgrade. To be honest, I like all three. 2500k, 3570k, 4670k.
I am ordering in probably about a week and will probably just get the one that I can make fit my budget or what is available in the forsale /trade forums at the time.

If I get the 3570k or 4670k I want to delid it. good idea? or is it even necessary?

 

[BallaTheFeared]

Really depends on your chip and what you want to do with it.

I planned to delid mine before I got it, plans change though as I had no reason to once I had it up and running.

 

[nx02nx02]

Really depends on your chip and what you want to do with it.

I planned to delid mine before I got it, plans change though as I had no reason to once I had it up and running.

I did the same thing years ago with my current build. I designed the whole thing for overclocking but when I got it all together I didn't really see any need because it ran almost everything perfectly. core 2 duo's sure were awesome for their time.

 

[BallaTheFeared]

Well mine runs 4.8GHz, I just didn't have a heat problem so there was no reason to void my warranty

 

[nx02nx02]

Well mine runs 4.8GHz, I just didn't have a heat problem so there was no reason to void my warranty

So is there some 4670k's? that don't have the heat problems? How did you get it to 4.8 without heat problems? That makes me want one even more.

 

[BallaTheFeared]

Intel gave it to me personally to spread false hope.

































Just pure luck in the silicon lottery. Even if you win the clock speed lottery, you still have to enter the IHS gap lottery.

I can't complain of course, I went from a really fast water cooled i5-2500k to a really fast mid-range air cooler i5-4670k.

 

[Barfo]

Get the 2500K if you can overclock it to 6HZ.

 

[2is]

lol I don't know how they got those numbers, maybe in Russia they're all liquored up on Vodka when they do their tests? Maybe that's the Beta? Its no secret that any 4 core CPU will perform similar to others in the frostbite engine, BF4 is no different. This has been tested to death and i simply don't trust the methodology of that Russian site. Here's another benchmark from a more trusted source that shows what other benchmarks have shown, BF4 RETAIL doesn't CARE about CPU, its GPU limited.

article link:

http://www.techspot.com/review/734-battlefield-4-benchmarks/page6.html

i noticed no difference going from my PhenomII X6 @ 3.9ghz to my 2500k @ 4.5ghz (in BF3 @ 1080), and that's supposed to be a much bigger performance gap than a 2500k & a 4670k.

Unless you're buying BF4 to only play the SP campaign, it VERY MUCH cares about CPU. More so than any other FPS before it and more so than most any other game out there.