Xeon for gaming question.

[RespawnRedemption]

This questions has probably been answered somewhere and I appoligize preemtively if so. I scrolled for a while and seen lots of posts about them but not quite answering my question. 

 

We all know there are no games that need a 20 core beast for gaming as most games do not support more than 4-8 cores, even than few on the higher core values. I was wondering though, if I was to dedicate cores to individual prosseses, would this improve anything? IE 4 cores running nothing but my game, 4 cores running nothing but my video software, 4 cores recording video, 4 cores running stream and the rest running everything else. Its been bugging me a while now. Would this effect communication between everything creating bottlenecks? 

 

It might be annoying to have to organize the cores everytime you need to reset but the hardcore would likely not care about this if it give them even 10 more fps.

 

This has been bothering me since I had to disable cores on my old 6 core phenom 2 to let my nephew play Spore.

 

I'm saving for a new PC now and while I'm not likely to get a 20 core xeon, I might consider a 12 core prossesor to split 3 ways between game, video recording and everything else.

[TheRandomness]

You can't choose what cores run what (well, you can set affinity, but that'll just prioritises one thread). The CPU works on all tasks at once, and what you're asking for is a CPU with a high core/thread count and high clock speed, which basically means an i7-6800k overclocked.

[typographie]

16 minutes ago, RespawnRedemption said:

We all know there are no games that need a 20 core beast for gaming as most games do not support more than 4-8 cores, even than few on the higher core values. I was wondering though, if I was to dedicate cores to individual prosseses, would this improve anything? IE 4 cores running nothing but my game, 4 cores running nothing but my video software, 4 cores recording video, 4 cores running stream and the rest running everything else. Its been bugging me a while now. Would this effect communication between everything creating bottlenecks?

What you're suggesting is basically default Windows behavior, at least in theory. Your CPU has to divide its attention between all tasks running on your PC at once, and having more cores (or even virtual cores) can help with that, to a degree that depends on the tasks.

 

The problem with a 20-core Xeon in games isn't the core count, it's that those higher-end Xeons tend to run at very low clock speeds. Even in games that are really well-threaded, a ~2 GHz frequency is likely to cost more performance than the core count is likely to make up for.

 

If you're buying a PC primarily for gaming and you're interested in heavy multitasking, one of the X99 i7's, like the i7-6800K/6900K, are better choices. You'll have plenty of physical/virtual cores for gaming and streaming/recording/editing, they start at clock speeds that are already acceptable for games, and they can be overclocked beyond that.

[xAcid9]

If your priority is more into gaming, don't get many cores Xeon with low single core performance.

[RespawnRedemption]

I just had to ask because of a few games requireing me to remove 2 of my 6 cores affinity in order to run such as the aformentioned Spore. So wondered if anyone tried benchmarking with cores assigned lke this. It just seemed odd to me as you would think the games or OS would just not use the cores it couldent access. My next build is definatly going to be a X99. 

 

Maybe I should ask the question "Could setting affinities improve proformance on a benchmark?"

[Techno_Reverend]

30 minutes ago, RespawnRedemption said:

This questions has probably been answered somewhere and I appoligize preemtively if so. I scrolled for a while and seen lots of posts about them but not quite answering my question. 

 

We all know there are no games that need a 20 core beast for gaming as most games do not support more than 4-8 cores, even than few on the higher core values. I was wondering though, if I was to dedicate cores to individual prosseses, would this improve anything? IE 4 cores running nothing but my game, 4 cores running nothing but my video software, 4 cores recording video, 4 cores running stream and the rest running everything else. Its been bugging me a while now. Would this effect communication between everything creating bottlenecks? 

 

It might be annoying to have to organize the cores everytime you need to reset but the hardcore would likely not care about this if it give them even 10 more fps.

 

This has been bothering me since I had to disable cores on my old 6 core phenom 2 to let my nephew play Spore.

 

I'm saving for a new PC now and while I'm not likely to get a 20 core xeon, I might consider a 12 core prossesor to split 3 ways between game, video recording and everything else.

There has been MAJOR processor performance improvements by intel, a 6 core intel CPU will decimate the Phenom, even starting from Nehalem with the 980x i7.

 

I would go x99 and buy a 5820k, good price, HEDT X99 platform and upgradable.

 

 

[typographie]

12 minutes ago, RespawnRedemption said:

I just had to ask because of a few games requireing me to remove 2 of my 6 cores affinity in order to run such as the aformentioned Spore. So wondered if anyone tried benchmarking with cores assigned lke this. It just seemed odd to me as you would think the games or OS would just not use the cores it couldent access. My next build is definatly going to be a X99. 

 

Maybe I should ask the question "Could setting affinities improve proformance on a benchmark?"

What you're describing just sounds like Spore was designed without much consideration given to exotic hardware configurations, which isn't awfully surprising for a 2008 release with a 30 FPS lock. I wouldn't expect to have any issues like that with most games released over the past several years.

 

Every so often you see someone recommend setting CPU core affinity as a supposed solution to issues with particular games, claiming performance benefits. I've personally never seen it make a difference, though. I suppose its possible that some benchmark could show an improvement by doing that, but like I said above, that should already be happening by default. Modern versions of Windows don't normally need you to tell it what cores to use for what. It would probably have to be a very contrived benchmark.