Multiple core CPU utilisation

[Shaggy1973]

I think people have misunderstood bottleneck.

It is true that your FPS won't go down due to CPU bottleneck. However, the game will start to stutter. That is to say, that units will begin to jump across the land rather than smoothly walk.

The whole time this is happening, your FPS will be pretty steady. Even if you zoom in so that you can only see one unit, the unit will move past the camera in short jumps. The problem is that as you flood the battlefield with more and more soldiers, and they start fighting, this all takes CPU calculations. When the CPU is maxed, the actual game calculations of where soldiers are and what they are doing reached max, and things jump.

This thread was never created to talk about FPS.

[alQamar]

refer post #125

Vastator
In this way the CPU benchmark can loose its meaning, since it's meant to test the FPS in a condition of low GPU usage (so to test the "CPU FPS").
This is why all graphic settings of the CPU benchmark are set on low, except the unit details which are set on ULTRA (unit details are CPU intensive) and the number of men on the battle.

Thats even not correct by changing the resolution I ensure that the CPU is not lazy and I get more load, the change of this parameter has slowed it a lot (linear by the size of increased resolution), naturally, so I was able to see minor changes in a more dramatic way if you know what I mean.

This thread was never created to talk about FPS.

correct! At least not the FPS generated by the graphics card. The point why I ran the graphics benchmark was poor technical curiosity. According to my tests some years ago there we had a complete different situation. And there the goal was FPS and loading times. Now we talk about quad core utilisation and hyperthreading etc.

[Emperor Domitianus]

The use of a multicore CPU is now a must considering the high level of details avaible. I hope at least a dual/quad core.

[boxleitnerb]

76 fps is kind of high. Then I understand, since any CPU intensive action like animations and combat calculations let my fps drop like a stone into the 20s and 30s. So basically as long as there are no engagements whatsoever and the unit number is small, the CPU load is negligible?

On a different note:
I'm still curious how it is possible to have animation stuttering and high fps. Just start a battle with like 8 armies and look at just one unit walking directly at the beginning. No engagements yet, just a lot of units moving off-screen. At least that happened at the time I conducted my tests. fps could be in their 40s and 50s and yet the animations were terribly skippy.

[=Vastator=]

refer post #125


Thats even not correct by changing the resolution I ensure that the CPU is not lazy and I get more load, the change of this parameter has slowed it a lot (linear by the size of increased resolution), naturally, so I was able to see minor changes in a more dramatic way if you know what I mean.

Resolution is a GPU intensive parameter. It's called CPU benchmark, because all the CPU intensive settings (melee with many soldiers and unit details on ultra) are up, not because in this mode "all the stuff" is handled by the CPU.

[SamueleD]

Resolution is a GPU intensive parameter. It's called CPU benchmark, because all the CPU intensive settings (melee with many soldiers and unit details on ultra) are up, not because in this mode "all the stuff" is handled by the CPU.

Playing with ultra low details, even at 1080p, won't be intensive for the GPU, unless it's particularly crappy, but it will be intensive for the CPU, since you show more soldiers and have to calculate more on-screen fights and animations.

[talljoe]

76 fps is kind of high. Then I understand, since any CPU intensive action like animations and combat calculations let my fps drop like a stone into the 20s and 30s. So basically as long as there are no engagements whatsoever and the unit number is small, the CPU load is negligible?

On a different note:
I'm still curious how it is possible to have animation stuttering and high fps. Just start a battle with like 8 armies and look at just one unit walking directly at the beginning. No engagements yet, just a lot of units moving off-screen. At least that happened at the time I conducted my tests. fps could be in their 40s and 50s and yet the animations were terribly skippy.

I can confirm this. This does happen to me too- the stuttering units despite high FPS. It is strange. Everything including scrolling is still very smooth, but the units do start skipping and stuttering. It's pretty disorienting.

[eskil]

I can confirm this. This does happen to me too- the stuttering units despite high FPS. It is strange. Everything including scrolling is still very smooth, but the units do start skipping and stuttering. It's pretty disorienting.

i have this problem only when huge bettale of 25vs 25 or more happen with maximum speed...if i use normal speed even at 40 vs 40 there is not shuttering and i'm playng with a laptop.
i think this happen due to the animation the cpu must calculate for all those soldier...it simply cannot keep up because it not optimized...for example i aways use 1 core to the fullet and the other one to 20%...it could be grat if i could use both to 80%...the performance would greatly increase XD

[alQamar]

i have this problem only when huge bettale of 25vs 25 or more happen with maximum speed...if i use normal speed even at 40 vs 40 there is not shuttering and i'm playng with a laptop.
i think this happen due to the animation the cpu must calculate for all those soldier...it simply cannot keep up because it not optimized...for example i aways use 1 core to the fullet and the other one to 20%...it could be grat if i could use both to 80%...the performance would greatly increase XD

Please review the PDF file guys, I used FXAA instead of normal antialiasing, it will boost your framefrates. In the PDF you can review how to setup this for NVIDIA at least.
At least I can third the problem. Good ol' warscape. I wonder when they or if they intend to release benchmarks comparisons between Rome 2 and Shogun2 so we finally get something of proove for their promised optimisations and let the pig out of its poke. What do you think? Only few days left...

[alQamar]

Hello everyone, I've not edited the previous post to bump. There was still a significant bug in the PDF file on my previous posting. I've attached a new one. Sorry for that inconvinience.

http://www.twcenter.net/forums/showt...1#post13057026

[Raimeken]

This is the reason why I always want CA to retire the Warscape. Sure its cheaper to make the games but in the end it will make the game much better. For example, many games these days look good, but they only utilize 1 or 2 cores because a vast majority are built on engines like Unreal 3, idTech 3, etc. However, take the example of Frostbite 2 that powers BF3. Frostbite 2 is known to be very good looking while being very optimized, and thats because DICE took the time to retire their older Frostbite engines and optimize it for more that 1 or 2 cores. They were so good at doing it that they even managed to make Hyper Threading, yes you heard me right, Hyper Threading to be useful in 64 player matches. Now if optimizing CPU load can do that to an FPS, what can it do to a Total War game. It'll at least make it less of a laggy mess.

[=Vastator=]

Playing with ultra low details, even at 1080p, won't be intensive for the GPU, unless it's particularly crappy, but it will be intensive for the CPU, since you show more soldiers and have to calculate more on-screen fights and animations.

Well, this makes more sense .

@alQamar
I don't understand the stats on PDF though, are you saying that a core2 is able to reach 31 avg FPS on CPU bench? It seems not right to me (but i think i understood wrong )

[alQamar]

You're completely right but when they keep the pace they have releasing "new" games I am really concerned about their capacity for a new engine. CA is fighting on several frontlines and lots of other projects like console games , TW arena or this fancy Shogun battles comic game. I cannot understand why but I think even they are still expanding I fear they don't have man power and funds for a new engine and giving quick releases a break. Their disadvantage is that they cannot buy an engine for TW and build a game around because TW is something unique in gaming. Till they don't pick u up this challenge we have skating units fps drops by 50% even as reported on a gtx 780 ridiculously because the battle engine bottleneck.

@alQamar
I don't understand the stats on PDF though, are you saying that a core2 is able to reach 31 avg FPS on CPU bench? It seems not right to me (but i think i understood wrong )

most likely. As I said the cpu bench did 31 FPS on a dual core (hardware limiting a i7-2600k @4.5 GHz)
I hoped the PDF release 1.2 is more understandable if you read all remarks within the document.

[=Vastator=]

most likely. As I said the cpu bench did 31 FPS on a dual core (hardware limiting a i7-2600k @4.5 GHz)
I hoped the PDF release 1.2 is more understandable if you read all remarks within the document.

About this you are wrong. You can't hw limit an i7 to 2 cores and say that it's equal to a Core 2 duo ! It doesn't work this way fortunately.
The Core 2, and the intel ix (first, second, third and fourth gen) have completely different architecture and performance per core, or single thread performance.
Here's a what a Core 2 duo at 3 GHz is "able" to achieve.

[alQamar]

OMG how did I knew that would happen. With all respect read the posting carefully and think before you post. Thank you very much Sir. Did you come here to start a fuss? I have done a remark on this in bold letters! There are also remarks in this PDF file I attached. You are mixing apples and carrots. The things you point out is a complete different approach of benchmarking. If you still have questions please look up Wikipedia what "normalisation" means in technical matters.

Your figures would be helpful if all the cpus shown would have clocked to a same speed.

And before I have to engage anything again. The reason of this benchmark was not who is fastest but if and how the TW engine supports quad core cpus and hyperthreading. According to my remarks there is nothing wrong not even the leveling I've made.

[=Vastator=]

OMG how did I knew that would happen. With all respect read the posting carefully and think before you post. Thank you very much Sir. Did you come here to start a fuss? I have done a remark on this in bold letters! There are also remarks in this PDF file I attached. You are mixing apples and carrots. The things you point out is a complete different approach of benchmarking. If you still have questions please look up Wikipedia what "normalisation" means in technical matters.

Your figures would be helpful if all the cpus shown would have clocked to a same speed.

Wow, no need to be offensive.
We are discussing things, if you don't want to, thinking you are absolutely right, then you should have said it.
I posted that only to show what the Core 2 performance is.
You are saying that a Core 2 is able to reach the marvellous 31 FPS (a bit different then 12..., even considering differents clock speed). All this based on your "approach to benckmarking" that to have a Core 2 DUO you have to take an i7 and made it working at 2 cores...
If this is right to you, an makes sense, then ok....

[alQamar]

Sorry was offensive because I never said or at least not intended to say a core 2 Duo will deliver 31 fps at 4.5 GHz theoretically. As this is if course wrong as you said. I said a dual core with 4.5Ghz as stated a 2600k generated it.

[Adreno]

This is the reason why I always want CA to retire the Warscape. Sure its cheaper to make the games but in the end it will make the game much better. For example, many games these days look good, but they only utilize 1 or 2 cores because a vast majority are built on engines like Unreal 3, idTech 3, etc. However, take the example of Frostbite 2 that powers BF3. Frostbite 2 is known to be very good looking while being very optimized, and thats because DICE took the time to retire their older Frostbite engines and optimize it for more that 1 or 2 cores. They were so good at doing it that they even managed to make Hyper Threading, yes you heard me right, Hyper Threading to be useful in 64 player matches. Now if optimizing CPU load can do that to an FPS, what can it do to a Total War game. It'll at least make it less of a laggy mess.

hmm this is not really fair.. a lot of software is almost impossible to run well over multiple cores, for some software its actually detrimental to run it over multiple cores and not every piece of software can be made to utilize a lot of cores.. in fact most games are lacking this and might not get helped at all by an engine that can utilize multiple cores, im sure that if game developers could do this they would do so because it would only increase their potential market, but it also creates more problems with bughunting and patching on top of that, which might not make it viable at all.. same thing for hyperthreading really, if they can use it great, but dont expect it to get used in every game because they cant

[Totalheadache]

hmm this is not really fair.. a lot of software is almost impossible to run well over multiple cores, for some software its actually detrimental to run it over multiple cores and not every piece of software can be made to utilize a lot of cores.. in fact most games are lacking this and might not get helped at all by an engine that can utilize multiple cores, im sure that if game developers could do this they would do so because it would only increase their potential market, but it also creates more problems with bughunting and patching on top of that, which might not make it viable at all.. same thing for hyperthreading really, if they can use it great, but dont expect it to get used in every game because they cant

this to me sounds like complete guff/twaddle!

where do you get this info from? i.e a lot of software is almost impossible to run over multiple cores..and not every piece of software can be made...

games and software can be made to use 4 cores if the company want to and code it into their engine. the tech is here but some companies dont do it because either it costs to much or they dont see the need.

the frostbite2 engine is a clear example of where the time+money has been spent. and you arguably get a better gaming experience for it.

so if you are somehow trying to say warscape doesnt use 4 cores because of something other than monetary reasons you are quite delusional. (they could do it but would mean a reworking of the code which would add x man hours and money, in their infinite wisdom they arent doing this...hence why we have threads such as these).

i.e nothing to do with being detrimental or creating more bugs etc. what absolute tosh. if you dont know what are talking about, dont post.

[mrartist]

I know this is a bit off topic but this thread is relative to my question. When I play Shogun 2 on dx 11 at battle loading screens in the game every now and then I get A white screen of death and have to reboot my comp. It's brand new and all the regular tests have been run to rule out ram and the processor etc. Was I supposed to be playing this in dx 9? My pc is a 4770k quad core liquid cooled 2x sli gtx 770's 240 gb ssd running the monitor at 2550 something x 1080. is it the game on dx 11 or do I really have some unfound problem. I would really appreciate the help of the guru's on this forum thread. Thanks