Can you justify upgrading from a 2600k for gaming?

[atavax]

I've seen people comparing the 2600k to the 8700k or the 7700k and there is always small improvements going with the newer CPU, but I think a better test is comparing it to lower end CPUs. If my overclocked CPU is outperforming a non overclocked CPU with a retail price of around $250, then don't i already have a very high end modern gaming CPU? So like if my 2600k outperforms a 7600, then upgrading from my 2600k makes less sense than upgrading from a 7600, right?

[WereCat]

unless you play at high refresh rates like 144Hz+ then no

[GirlFromYonder]

1 minute ago, atavax said:

I've seen people comparing the 2600k to the 8700k or the 7700k and there is always small improvements going with the newer CPU, but I think a better test is comparing it to lower end CPUs. If my overclocked CPU is outperforming a non overclocked CPU with a retail price of around $250, then don't i already have a very high end modern gaming CPU? So like if my 2600k outperforms a 7600, then upgrading from my 2600k makes less sense than upgrading from a 7600, right?

yes. But also the 7600 variety is i3 now. If your 2600k is giving you problems upgrades. If not and you are fine. Dont worry about it, Pascal and Polaris/Vega are very good.

[VictimRAID]

Nope, if ur only gaming then the 2600K is still a great chip

[draginator]

You could upgrade to a used 3770k and still be on the same platform, then you would have as much performance as possible without having to upgrade anything else.

[GirlFromYonder]

2 minutes ago, draginator said:

You could upgrade to a used 3770k and still be on the same platform, then you would have as much performance as possible without having to upgrade anything else.

/but why...if they are at 5 giggawiggles then Y.

[Rolling Potatoe]

For 60 fps gaming there is no point in upgrading in terms of performance.

You would get lower power consumption, m.2, usb 3.1 gen 2 etc. though.

[dizmo]

If you're aiming for 60 fps, then no there's no point. If you're aiming for 144hz? Then yeah, there will be a difference.

[ARikozuM]

Don't bother, we're still a few years away from engines using more than four cores. 

[Vroooom]

Only if you have a high refresh rate monitor and a GPU powerful enough to draw that many frames at your chosen resolution and settings

[Zmax]

I have a I5 2500, still going strong. Update my GPU still low level gtx960 Not to great but got 6 years out of it,

Next one is I7 8700k  soon as chip is available

[atavax]

1 hour ago, dizmo said:

If you're aiming for 60 fps, then no there's no point. If you're aiming for 144hz? Then yeah, there will be a difference.

do you have any proof that a 2600k prevents high fps for 144hz gaming? hardware Canucks tested the 2600k vs the 8700k and with both overclocked there was maybe a 2-5fps difference.

[Zmax]

Most games used GPU power . Games in the future will go CPU cores and GPU. Star Citizen is going that way. The more core you have the better the game will be. 

[dizmo]

5 minutes ago, atavax said:

do you have any proof that a 2600k prevents high fps for 144hz gaming? hardware Canucks tested the 2600k vs the 8700k and with both overclocked there was maybe a 2-5fps difference.

They did, yes. DigitalFoundry did testing and it saw much larger gaps in a lot of titles. I trust DigitalFoundry's testing a lot more than Hardware Canucks.

1 hour ago, ARikozuM said:

Don't bother, we're still a few years away from engines using more than four cores. 

They already use more than four cores..

[atavax]

16 minutes ago, dizmo said:

They did, yes. DigitalFoundry did testing and it saw much larger gaps in a lot of titles. I trust DigitalFoundry's testing a lot more than Hardware Canucks.

They already use more than four cores..

can you link the digital foundry testing? i'm not seeing it.

 

and yes, utilizing more than four cores is very prevalent and is why a 2600k benchmarks better than the 2500k on virtually every game, and an overclocked 2600k beats most i5s in alot of games.

[ARikozuM]

5 minutes ago, dizmo said:

They did, yes. DigitalFoundry did testing and it saw much larger gaps in a lot of titles. I trust DigitalFoundry's testing a lot more than Hardware Canucks.

They already use more than four cores..

HT is assigned by the OS, not the engine. Not many engines are using more than four cores and those that "do" aren't showing the improvements that would be expected. 

 

Cryengine shows an improvement as I would have expected. I don't have Cryengine nor do I want to pay the $600 for it. 

CDProjekt Red's engine shows a good improvement, but I wouldn't have expected a 10% increase only. 

 

The majority of engines aren't using more cores. I'd give it a couple more years (maybe 2019). 

[dizmo]

1 hour ago, atavax said:

can you link the digital foundry testing? i'm not seeing it.

 

and yes, utilizing more than four cores is very prevalent and is why a 2600k benchmarks better than the 2500k on virtually every game, and an overclocked 2600k beats most i5s in alot of games.

It was this one. It's using the 3770k for comparison, which is a little quicker than the 2700k so it still makes sense.

The tests in the first half there's more of a difference, but that just goes to show that while such things are game dependent, in some titles and possibly more in the future it's very much true.

 

52 minutes ago, ARikozuM said:

HT is assigned by the OS, not the engine. Not many engines are using more than four cores and those that "do" aren't showing the improvements that would be expected. 

 

Cryengine shows an improvement as I would have expected. I don't have Cryengine nor do I want to pay the $600 for it. 

CDProjekt Red's engine shows a good improvement, but I wouldn't have expected a 10% increase only. 

 

The majority of engines aren't using more cores. I'd give it a couple more years (maybe 2019). 

I think you're mistaking "using cores" for "can't perform well without them."

[ARikozuM]

Just now, dizmo said:

I think you're mistaking "using cores" for "can't perform well without them."

I'm not. One would usually think that a 50% in core count would yield a 0-50% increase in performance. Actual performance difference seems to be around the 10% mark; this means that either drivers are off-loading/scheduling work across more cores than assigned or that the engine doesn't have much more to ask for in terms of threads (actual threads, not the logical cores). 

[dizmo]

1 minute ago, ARikozuM said:

I'm not. One would usually think that a 50% in core count would yield a 0-50% increase in performance. Actual performance difference seems to be around the 10% mark; this means that either drivers are off-loading/scheduling work across more cores than assigned or that the engine doesn't have much more to ask for in terms of threads (actual threads, not the logical cores). 

Yes but my point is your statement of "games don't use more than 4 cores" is simply incorrect. There's an improvement, regardless of how marginal that may be.

[ARikozuM]

2 minutes ago, dizmo said:

Yes but my point is your statement of "games don't use more than 4 cores" is simply incorrect. There's an improvement, regardless of how marginal that may be.

Using HT isn't "using a core" since that isn't dependent upon the engine. While my statement may incorrect as an absolute, it is not from a broader picture. Very few engines are using more than the four cores leaving the rest to either catch up or give better performance for what they ask for. 

[atavax]

35 minutes ago, dizmo said:

It was this one. It's using the 3770k for comparison, which is a little quicker than the 2700k so it still makes sense.

The tests in the first half there's more of a difference, but that just goes to show that while such things are game dependent, in some titles and possibly more in the future it's very much true.

its also a super conservative overclock. In the hardware Canucks testing they had the 2600k at 4.8Ghz. In the digital foundry testing they had it at 4.5Ghz and higher clock speeds is probably the largest factor in increased game performance for CPUs, so it makes sense there is a large discrepancy between the two tests.

[atavax]

as for the topic of the threads game engines take advantage of. The xbone and ps4 have 8 cores on the x64 architecture; i think you're in denial if you think game engines aren't largely taking advantage of 8 threads and the benchmarks tend to back that up with the 2600k outperforming the 2500k in gaming benchmarks. I would expect game engines to start taking advantage of more than 8 cpu threads on a large scale once there are game consoles with more than 8 cpu threads.

[sputnik13]

Not intending to offend the OP, but I feel like questions like these are a bit pointless...  if your system is struggling with games that you want to play, then you should be asking what to upgrade to...  if your system isn't struggling, then why upgrade?  Better performance on benchmarks is meaningless if the things you do with it doesn't take advantage of the performance, other than maybe bragging rights (hey my system is 10% faster than yours ha hah...).

[Zmax]

I aggree . Go with the flow 

[AlTech]

5 hours ago, ARikozuM said:

Cryengine shows an improvement as I would have expected. I don't have Cryengine nor do I want to pay the $600 for it. 

CryEngine is pay what you want. You can get it 100% for free.

3 hours ago, atavax said:

as for the topic of the threads game engines take advantage of. The xbone and ps4 have 8 cores on the x64 architecture; i think you're in denial if you think game engines aren't largely taking advantage of 8 threads and the benchmarks tend to back that up with the 2600k outperforming the 2500k in gaming benchmarks. I would expect game engines to start taking advantage of more than 8 cpu threads on a large scale once there are game consoles with more than 8 cpu threads.

OP.

 

If you're interested in 144Hz gaming then upgrading to an 8700K is a no brainer.

 

If you're interested in sticking with 60fps but want better performance in CPU demanding games and software then you could upgrade to an 8400 or a 8700.

 

IPC has improved quite a bit since Sandy Bridge. Sure, improvement per gen isn't great but over time the performance increase is quite huge. Kaby Lake (and Coffee Lake for that matter) are over 40% faster than Sandy Bridge in terms of IPC.