which CPU is best for AC?

[Tim Meuris]

I mean how many cores does it need to play this game best
and ARMA 3 as well (on 3 screens)

sorry bit of a noob in this area
so I hope i asked the right question...

 

[Saul Boucher]

AC recommends a quad core.

My recommendation for a gaming PC is usually an i5-4670k for a mainstream machine, or an i7-4770k for something a little more high end.

However, for graphics performance that typically falls on the GPU.

 

[Michael Hornbuckle]

My i5 2500k hasn't broke a sweat with any game I have played yet.....AI in AC will be the real test....I doubt you can go wrong with any current quad core i5 or i7 cpu's.

 

[nobrakes]

Some info from Stefano Casillo about the AI (still in development) and the needed CPU power at the moment:
https://twitter.com/KunosStefano/status/419793942473744384

 

[Tim Meuris]

Thnx for all the info guys!

hmmm, sounds like a demanding game CPU-wise as well.
Anyone any experience with ARMA3 in relation to CPUperformance by any coincidence?

 

[Bryan O'Connor]

Thnx for all the info guys!

hmmm, sounds like a demanding game CPU-wise as well.
Anyone any experience with ARMA3 in relation to CPUperformance by any coincidence?

Yes. And it's terrible. Uses 1 core most of the time and leaves you with horrible frame rates because of it.
The engine is so old and poorly optimised it's sheer pot luck rather than your system build as to whether you get a decent framerate.
There's all sorts of vajazzle you can do to try and make it run better like unparking cores and stuff, but what's the point? It should utilise multi-core out of the box like 99% of all other games made after 2006.
I can't be bothered with it atm. If i want to watch stop-start animation i will watch Wallace and Gromit.

I have an i5 3570K, 8gb ram and r9 270x. And it runs terrible. It's not the best rig out there i know, but still....

 

[Tim Meuris]

Yes. And it's terrible. Uses 1 core most of the time and leaves you with horrible frame rates because of it.
The engine is so old and poorly optimised it's sheer pot luck rather than your system build as to whether you get a decent framerate.
There's all sorts of vajazzle you can do to try and make it run better like unparking cores and stuff, but what's the point? It should utilise multi-core out of the box like 99% of all other games made after 2006.
I can't be bothered with it atm. If i want to watch stop-start animation i will watch Wallace and Gromit.

I have an i5 3570K, 8gb ram and r9 270x. And it runs terrible. It's not the best rig out there i know, but still....

I read this about that coreusage before
shame cause for such a big world it must have awesome graphics
ACengine uses more cores then?

 

[Simon Roberts]

Im on a 3570K at 4.8Ghz, hoping that will be enough!

I run Arma3 also with no issues what soever with w GTX780.

 

[Bryan O'Connor]

I read this about that coreusage before
shame cause for such a big world it must have awesome graphics
ACengine uses more cores then?

To be honest i don't know. But i imagine it will. Especially when AI is on track.
AC is a new engine and i am sure Kunos know what they are doing. It's not like in Arma's case where they are re-hashing a decade old game engine each time and adding some new graphical fancyness.
Most home PC's have minimum of Dual Core's now, so most developers know this and build with multi-core in mind.
The original Arma engine was built before the mass proliferation of Dual and Quad cores. And rather than start again, they just keep adding more and more on top. It's probably not their fault and down to budget constraints, but it still doesn't make for a great experience, like it should.

 

[Dinca Andrei]

Since i will not use AI so much,only if its out before Multiplayer,i am not so worried...I have a
AMD Athlon II X4 640

 

[Raig Strehlow]

i would use a intel xeon e3-1230v3 (when i build a new pc in the next year), got the power of an i7 + hyper threading and is a little bit cheaper. I dont overclock, so it would be perfect for me. Maybe they use mantle and we got nice usage of all cores.

 

[Juris]

Just wanted to flag others on RD to a post on the AC forums regards using the program ProcessLasso to help limit CPU warnings which some are saying is improving their fps, likely due to PL creating a smoother cpu usage to prevent bottlenecks.

I had used ProcessLasso for a good while on a few laptops to help battery life but didn't use it on my main gaming iMac thinking it could slow performance. Think I may have to revise that assumption and test. Posting a link to it as anyone who hasn't stolen the game will be able to log in.

mod-edit: you can post it here at RaceDepartment as well as Kunos is reading this forum

 

[Bokonon]

I read this about that coreusage before
shame cause for such a big world it must have awesome graphics
ACengine uses more cores then?

I use an I5 2500K, in the BMW E30 quick race all 4 cores are loaded between 50 and 85% and fairly evenly too. I would imagine that anything less than 4 cores, or older than the intel 920/930 era will suffer some performance issues with AC.

 

[Tim Meuris]

I use an I5 2500K, in the BMW E30 quick race all 4 cores are loaded between 50 and 85% and fairly evenly too. I would imagine that anything less than 4 cores, or older than the intel 920/930 era will suffer some performance issues with AC.

just what I hoped for, thnx!

 

[Pawel Ziolkowski]

I use an I5 2500K, in the BMW E30 quick race all 4 cores are loaded between 50 and 85% and fairly evenly too. I would imagine that anything less than 4 cores, or older than the intel 920/930 era will suffer some performance issues with AC.

Just for the record, the way windows show CPU usage is usually not accurate. A game might use single core only and it will still sow roughly even load across the board.

I've read a comment on AC forums that AC utilizes 2-3 cores max. (I think physics isn't multi-threaded, but things like graphics and AI are probably running on separate threads.

So i would think a quick 4 core would be better than slower 8 core. I wouldn't be surprised if it run well on a very quick dual core either.

 

[Jorge Alves]

using the program ProcessLasso to help limit CPU warnings which some are saying is improving their fps, likely due to PL creating a smoother cpu usage to prevent bottlenecks.

Thanks a lot for posting Juris

I just tryed the ProcessLasso with my FX 6300 6 x 4.5Ghz and i got good improvements

I didnt get more FPS but they are much more stable now ... without ProcessLasso FPS could drop from 90 to 65 on starts but now only drops to 75 ... very good

The only configs i changed on ProcessLasso was disabling the CPU Parking (6 cores all working now during races) and enabled the Game/Multimedia Mode

I also let the ProBalance enabled and it does a great job eliminating micro-stuttering

Good software

 

[Bokonon]

Just for the record, the way windows show CPU usage is usually not accurate. A game might use single core only and it will still sow roughly even load across the board.

I've read a comment on AC forums that AC utilizes 2-3 cores max. (I think physics isn't multi-threaded, but things like graphics and AI are probably running on separate threads.

So i would think a quick 4 core would be better than slower 8 core. I wouldn't be surprised if it run well on a very quick dual core either.

This is a phenomenon I've not heard of before, my experience in games for instance where a process has been assigned to an idle core by a patch, it has been reflected in what has been displayed in the 'performance' tab both before and after. In fact in all my time moitoring CPU usage for various reasons, overclocking, benchmarking, optimizing performance in particular games/ applications I have never had the impression personally or heard from any source that the perfomance monitor is anything less than accurate. Citations please?

 

[Pawel Ziolkowski]

This is a phenomenon I've not heard of before, my experience in games for instance where a process has been assigned to an idle core by a patch, it has been reflected in what has been displayed in the 'performance' tab both before and after. In fact in all my time moitoring CPU usage for various reasons, overclocking, benchmarking, optimizing performance in particular games/ applications I have never had the impression personally or heard from any source that the perfomance monitor is anything less than accurate. Citations please?

Sigh i have hard time remembering how the special cpu core utilization method was called and my google fu is failing today so i cannot give a good citation right now .

Best source i found was here: http://blogs.technet.com/b/winserve...cpu-utilization-for-performance-analysis.aspx
(it's around middle of the blog)

The problem is that even single threaded application can be switched between cores constantly, so for example: quad core CPU at maximum possible single threaded load can show 25% utilization of each core. And while it would be true that each core is "working" maximum performance would be the same as if we got only one.
And of course it's possible for applications to override this and get specific core affinity (idk if AC does this, and if yes to what extent) so it would show only one core on 100% load. There is also CPU parking that puts some cores to sleep, to distorts things even more.
Different CPU types have can show different utilization for the same software due to differences in architecture, so it's really hard to get some useful information from resource monitor / task manager CPU usage.

 

[Bokonon]

However it is achieved, the situation is AC on an I5 2500k is pretty even across the cores, of course this achieved through software optimisation, hence all the complaining with titles such as Arma and World of Tanks that don't optimise properly for multicore.
What your article seems to be talking about, is the same software can be compared on two systems with different configs and the comparisons are rarely 'apples to apples' comparisons, and on that point I wouldn't argue. Thats not to say though that the performance tab in Task manager is inaccurate, it is most certainly not. It's just that me showing a graph from task manager in the same senario in AC as you for instance wouldn't necessarelly mean too much without detailed system specs to give some perspective.
For instance as a result of this discussion I ran 24 lotus's at Silverstone yesterday while my CPU was at stock, and for brief instances I got the 'High Utilization' message. I have now OC'd to 4.2Ghz and will report back with results.

 

[Pawel Ziolkowski]

However it is achieved, the situation is AC on an I5 2500k is pretty even across the cores, of course this achieved through software optimisation, hence all the complaining with titles such as Arma and World of Tanks that don't optimise properly for multicore.
What your article seems to be talking about, is the same software can be compared on two systems with different configs and the comparisons are rarely 'apples to apples' comparisons, and on that point I wouldn't argue. Thats not to say though that the performance tab in Task manager is inaccurate, it is most certainly not. It's just that me showing a graph from task manager in the same senario in AC as you for instance wouldn't necessarelly mean too much without detailed system specs to give some perspective.
For instance as a result of this discussion I ran 24 lotus's at Silverstone yesterday while my CPU was at stock, and for brief instances I got the 'High Utilization' message. I have now OC'd to 4.2Ghz and will report back with results.

Yeah that article wasn't as useful as it was interesting . I worded wrongly too, task manager info is accurate but i was trying to say that it is not very useful in understanding what kind of CPU AC needs. If the physics run on only one thread a well OC'd i5/i7 would be best, if it's properly multithreaded 8core cheap AMD could be best for the buck. It's just that we can't tell from task manager how many threads an application is running, as even one can be sent to all 4 cores and still be limited to single core performance.

One thing you can find useful in testing AC, is the render stats app (stock app). The bottom two numbers show CPU stats:
C : cpu time
O : cpu occupancy
Not sure what they actually mean but lower is better.
On a i5 4670K (no OC yet), a grid of 24 lotuses (both GTE and '49) at silverstone i get C: ~2 and O: from ~50 at start lights to ~60 when AI are at T1. (never saw the high utilization message)