Friday, March 15th 2019

Three Kingdoms: Total War- System Specifications Revealed

The Creative Assembly has released the system specifications for Three Kingdoms: Total War. It seems those with less powerful systems will still be able to enjoy this title with relative ease considering the minimum specifications reference an Intel i7-8550U, UHD Graphics 620, and 6 GB of system memory. With a system of this caliber, users can expect to run the game with low settings at 1280x720 while getting 25-35 FPS. Minimum specifications change somewhat when looking at dedicated graphics cards. The CPU requirement drops from an Intel i7 to a 3.0 GHz Core 2 Duo, but only if the GPU is equal to or better than an NVIDIA GeForce GTX 650 Ti or AMD Radeon HD 7850.

In regards to pushing the game at high or ultra settings. Users will need a slightly beefier system. For High at 1080p you're looking at needing an Intel i5-6600 or Ryzen 5 2600X with 8 GB of memory and an Nvidia GeForce GTX 970 or AMD Radeon R9 Fury X. All these considered these requirements are quite reasonable. Moving on to Ultra settings is where the requirements take a large jump with the Creative Assembly suggesting the use of an Intel i7-8700K, 8 GB of memory and either an NVIDIA GeForce GTX 1080 or RTX 2060. Even with the large jump between high and ultra the fact remains these requirements are rather tame compared to some other AAA titles as of late.
The development team also touched on some graphical features they have been looking into. Specifically, they talked about the challenges of implementing SLI and Crossfire in an age of new anti-aliasing techniques. Years ago Multi-Sample Anti-Aliasing (MSAA) was the standard; however, it proved quite taxing. New techniques such as Temporal Anti-Aliasing (TAA) is according to the Creative Assembly a huge step forward for the Total War series. It offers good visual quality while being less performance intensive while also delivering better image quality than the likes of Fast Approximate Anti-Aliasing (FXAA). That said the use of TAA means the game is not compatible with currently used Alternate Frame Rendering (AFR) used by SLI and Crossfire and as such they won't support it. They did make not that with DX12 they do have the option of using Explicit Multi GPU as it does seem to be compatible with TAA, but due to the amount of effort necessary it is likely to remain on the back burner for the time being.

The game also supports multiple post-processing effects such as bokeh depth-of-field, two color palettes labeled as Romance which has a high color saturation for a vibrant image and the other being Records, which focuses on a lighter more realistic color saturation. To go with that, they also worked on their materials textures and added Screen Space Shadows for a higher level of detail in shadowed areas. With these changes, Three Kingdoms: Total War looks to be the best looking in the series by far and yet quite reasonable in regards to system requirements. The game will release on May 23, 2019, and is available for preorder on Steam.

PC Minimum Specifications (Integrated)
  • Expected around 25-35 FPS on campaign map and in a 1v1, 21 units vs 21 units battle, default graphics preset set to "Low", running at 1280×720
  • Operating System: Windows 7 64 Bit
  • Processor: i7-8550U 1.80GHz
  • RAM: 6GB
  • Video Card: Intel UHD Graphics 620
PC Minimum Specifications (Discrete)
  • Expected around 25-35 FPS on campaign map and in a 1v1, 21 units vs 21 units battle, default graphics preset set to "Low", running at 1280×720
  • Operating System: Windows 7 64 Bit
  • Processor: Intel Core 2 Duo 3.00Ghz
  • RAM: 4GB
  • Video Card: GTX 650 Ti | HD 7850 1GB VRAM
PC Recommended Specifications
  • Expected around 45-55 FPS on campaign map and in a 1v1, 21 units vs 21 units battle, default graphics preset set to "High", running at 1920×1080
  • Operating System: Windows 10 64 Bit
  • Processor: Intel i5-6600 | Ryzen 5 2600X
  • RAM: 8GB
  • Video Card: GTX 970 | R9 Fury X 4GB VRAM
PC 60 FPS+ Specifications
  • Expect 60 FPS+ on campaign map and in a 1v1, 21 units vs 21 units battle, default graphics preset set to "Ultra", running at 1920×1080
  • Operating System: Windows 10 64 Bit
  • Processor: Intel Core Intel i7-8700K
  • RAM: 8GB
  • Video Card: NVIDIA GeForce GTX 1080 | NVIDIA GeForce RTX 2060
Source: Total War
Add your own comment

30 Comments on Three Kingdoms: Total War- System Specifications Revealed

#1
lynx29
eh, not expecting much from this game
Posted on Reply
#2
dicktracy
8700k only? This is why you choose a gaming CPU instead of a server CPU ;)
Posted on Reply
#3
yakk
I guess CA's engine still isn't multi-threaded worth anything. This one is a hard pass for me.
Posted on Reply
#4
TheinsanegamerN
After CA's behavior the last few years, and them still using a garbage single thread engine in a game where multi threading should be easy, I'm gonna pass.

I still have Medieval II, which IMO is either the peak or near the peak of Total War games.
Posted on Reply
#5
Vayra86
Yeah I'll stick with Warhammer 2 for now. Thanks :)
Posted on Reply
#6
crazyeyesreaper
Chief Broken Rig
THe worst part for me is I really enjoy this era of history and greatly enjoyed the Romance of the three kingdoms games. So this was a dream come true for me but I am still a bit apprehensive.

As for CPU usage depending on situation the game actually scales okay at least on older processors. Going from a 4c/8t to 6c/12t with more memory bandwidth can see minimum fps boosted considerably. However the game is very much still single thread dependent but that will always be the case due to the need to match animations per soldier in a melee battle. If the Creative assembly went back to the method they used in Rome / Medieval 2 its highly likely performance would be better however battles would be less visually appealing in screen shots and during closeups.

That said I don't know anyone that plays these games for the close up action. It gets played so people can watch tactical battles unfold with thousands of troops.

Eitherway going from 2 cores to 4 cores to 6 cores typically brought a 10-20% boost at the same clock frequencies. Total War is also one of the few games where typically higher memory frequencies scaled nearly linearly in CPU only work loads. For example each step 2400 / 2800 / 3200 / typically resulted in a 1-2 fps increase across the board and to minimum framerates. Back during the age of DDR3 going from 1066, 1333, 1600, 1866, 2133 meant from the bottom to the top tier you would gain 4 FPS on the minimum frame rates back in Shogun 2. When the CPUs at the time only delivered 12-20 FPS gaining 4 fps was quite the achievement back then. CPU clocks also contributed greatly but again that stems from the animation system they have been using since Empire Total War. The current game engine is better suited to gunpowder than melee and it shows.

Okay / end rant lol.

I am cautiously excited. Rome II and Atilla are good games. and Warhammer performs exceptionally well. I just dont care for the gameplay as much. Thats my issue with Three Kingdoms I am on the fence with the gameplay so I want to see more about it before I make up my mind.
Posted on Reply
#7
cryohellinc
Don't buy this - game is filled with bugs, broken AI, same old trash engine and it is further dumbed down.
Posted on Reply
#8
bug
Was that type of trebuchet really used in China in that period?
Posted on Reply
#9
VashCZ
crazyeyesreaper said:
However the game is very much still single thread dependent but that will always be the case due to the need to match animations per soldier in a melee battle. If the Creative assembly went back to the method they used in Rome / Medieval 2 its highly likely performance would be better however battles would be less visually appealing in screen shots and during closeups.

///taking note of this
CPU clocks also contributed greatly but again that stems from the animation system they have been using since Empire Total War. The current game engine is better suited to gunpowder than melee and it shows.
Please elaborate.
Why wouldn´t some Vulkan rewritten engine be much better even for future usage?
Posted on Reply
#10
Nuckles56
Is anyone else surprised that the R9 Fury is up against a 970? I'm now wondering if the ultra recommendation for AMD is Radeon VII in crossfire
Posted on Reply
#11
crazyeyesreaper
Chief Broken Rig
VashCZ said:
Please elaborate.
Why wouldn´t some Vulkan rewritten engine be much better even for future usage?
Its because every soldier has a sort of AI attached the game engine then pairs those soldiers together in some cases its 1 on 1 or 2 on 1 or even 3 on 1. The engine then syncs these animations along with physics to every single soldier. Then add in kill moves where soldiers must be at set distances to each other. Now imagine syncing that across 5000-10000 soldiers all at once. Meanwhile theres such aspects as momentum and force that get applied to attacks which also then influence animations. By making the animations multi-threaded that would require syncing across all workloads and all cores. Meaning much higher latencies. So while in theory you could improve performance and have more soldiers on screen the higher latencies would make more mismatches and problematic results. There is also the fact that animations are strung together while not entirely random there is still random elements thus adding visual flair. Making the animation system between soldiers somewhat dynamic. Also keep in mind sound effects are also synced to animations played in front of the player which adds yet more data to sync with the initial animations.

Just want a video showing up close melee fighting in the older games vs new.

<div class="youtube-embed" data-id="CCuQjAuu3ZY:29"><img src="https://i.ytimg.com/vi/CCuQjAuu3ZY:29/hqdefault.jpg" /><div class="youtube-play"></div><a href="https://www.youtube.com/watch?v=CCuQjAuu3ZY:29" target="_blank" class="youtube-title"></a></div>
Shogun 2 Melee notice that often 1 on 1 fighting occurs here animations are selected per soldier and play out with their counter part animations. Now apply this across again 5000-10000 soldiers while keeping physics / animations / kill moves / momentum (charge and force) all synced. By running on multiple cores it sees extreme latency. Now factor in something like AMD's Ryzen with the CCX latency and you have a game that would run worse than it does already. Warhammer has even more varied animation and troop times which adds more complexity.

Now look at Medieval II
<div class="youtube-embed" data-id="bqgBeO8Bn60:42"><img src="https://i.ytimg.com/vi/bqgBeO8Bn60:42/hqdefault.jpg" /><div class="youtube-play"></div><a href="https://www.youtube.com/watch?v=bqgBeO8Bn60:42" target="_blank" class="youtube-title"></a></div>

Notice the animations are not synced they don't land they just swing at the air. This is far simpler to deal with computationally but less visually appealing. If CA went back to this type of method its like the game could have far more soldiers on screen in massive battles. Modern IPC + simpler animation calculations = better performance.

So essentially you asking for a rewrite of the engine. The current game engine was written for Empire with gunpowder in mind. That was a simple animation system. Melee was never its strong suit. Also to switch now to a new engine would require years of development. The fact is alot of the tech CA uses is proprietary and built for the current code base. Considering the current game engine was in the works before the end of Rome Total Wars DLC expansions with a shift to DX10 later. The code base is in fact quite old. However they have managed to improve multi-threading and API support with each release.

Keep in mind for Warhammer launch back during the Intel 4000-5000 series CPUs going from 2c/4t to 4c/8t to 6c/12 resulted in minimum FPS going up by around 10-20% in worst case situation. Thats actually better CPU scaling than most games get.
Posted on Reply
#12
VashCZ
Thanks for long answer.
Wow.

Well I know that there is far more dice rolling in new total war games and I totally understand it takes some performance.
But what you now mention is latency. Could you please direct me to some independent articles about this? I can not imagine the problem.

First game I got really into was Empire which I loved, problem like manual replenishment for huge armies was sick, but "fixed" in Napoleon.
Then Shogun 2. - with same amount of men I got less FPS caused by CPU
Then Rome II and Attila - even lower FPS cap caused by CPU

So according to my experience it got worse performance on same CPU for same amount of soldiers. No improvement. If I bought newer CPU I would only get back lost FPS and maybe some more but.. yea as you written, not thanks to optimization.

I admit I did not play DX12 Warhammers. Also did not try Brittania as articles said it is technically sucessor of Attila instead of Warhammer games. So maybe there is improvement thanks to new API and game optimization in Warhammers. Is this game in your opinian going to perform better than Warhammers?

Thanks,
VashCZ


EDIT: I also want to note that even though I do not hate simplier games, I would prefer something new by tactical-strategical games like TW series.
Diplomatics were not improved since like... I dunno, Rome? Are melee battles still so clunky in Warhammer? Can I even set my soldiers waypoints for more accurate pathing? Sometimes it is hard to avoid traps on field. I know where they are but by mouse order I can not micromanage and my squad usually dies :-D. Total fail of improvement. Town sieges are so sterile all the time. Any improvement, new scripts or anything in Warhammer?
Summing up, no inovations. Makes more restrictions for modders which have possibility to fix their games. How can they be worth my money. It is like paying for new Far cry games when I already played FC 3. Reskinned dullness.
Well but it still is not as bad as Arma series when we talk about optimization to be positive :-D. 2001 engine even this day.

Tell me that something changed since Attila. Force me to think it did. Otherwise I would be sad :-(.
Posted on Reply
#13
bug
VashCZ said:
Thanks for long answer.
Wow.

Well I know that there is far more dice rolling in new total war games and I totally understand it takes some performance.
But what you now mention is latency. Could you please direct me to some independent articles about this? I can not imagine the problem.
Imagine you run a factory. You go from 10 to 100 to 1,000 workers. Latency is the time it takes from when you issue an order till you get the end product. With 10 workers, eveybody's probably in the same shop so you can just shout your orders and they'll get done quickly. With 1,000 workers, the order have to follow a chain of command and even if the overall production works faster, 10 workers finishing their product at the same time can't all enter the warehouse at the same time to deposit them. The extra time needed to get your orders to the workers and for workers to hand over their finished products is the added latency.
Posted on Reply
#14
crazyeyesreaper
Chief Broken Rig
VashCZ said:
Thanks for long answer.
Wow.

Well I know that there is far more dice rolling in new total war games and I totally understand it takes some performance.
But what you now mention is latency. Could you please direct me to some independent articles about this? I can not imagine the problem.

First game I got really into was Empire which I loved, problem like manual replenishment for huge armies was sick, but "fixed" in Napoleon.
Then Shogun 2. - with same amount of men I got less FPS caused by CPU
Then Rome II and Attila - even lower FPS cap caused by CPU

So according to my experience it got worse performance on same CPU for same amount of soldiers. No improvement. If I bought newer CPU I would only get back lost FPS and maybe some more but.. yea as you written, not thanks to optimization.

I admit I did not play DX12 Warhammers. Also did not try Brittania as articles said it is technically sucessor of Attila instead of Warhammer games. So maybe there is improvement thanks to new API and game optimization in Warhammers. Is this game in your opinian going to perform better than Warhammers?

Thanks,
VashCZ


EDIT: I also want to note that even though I do not hate simplier games, I would prefer something new by tactical-strategical games like TW series.
Diplomatics were not improved since like... I dunno, Rome? Are melee battles still so clunky in Warhammer? Can I even set my soldiers waypoints for more accurate pathing? Sometimes it is hard to avoid traps on field. I know where they are but by mouse order I can not micromanage and my squad usually dies :-D. Total fail of improvement. Town sieges are so sterile all the time. Any improvement, new scripts or anything in Warhammer?
Summing up, no inovations. Makes more restrictions for modders which have possibility to fix their games. How can they be worth my money. It is like paying for new Far cry games when I already played FC 3. Reskinned dullness.
Well but it still is not as bad as Arma series when we talk about optimization to be positive :-D. 2001 engine even this day.

Tell me that something changed since Attila. Force me to think it did. Otherwise I would be sad :-(.
Its quite simple if you run threads on separate cores and instructions are in that specific cores cache sending that data to another core takes time. With AMD processors and the CCX layout when data needs to cross to a different core complex there is much higher latency than is typical with a monolithic (large die) CPU. so if latency was 20 ns between cores, that might slow things down but now imagine if it crosses to another CCX that adds 80 ns of latency. Once the data is sent work needs to be done then sent back and the data synced between the various cores. The more cores the more waiting the higher the latency. Now there are ways to deal with this.

CA could sync animations etc via unit rather than soldiers. Thus making battles between singular units run on a single core and divide out from there. But thats far more complex. Bottom line is companies want to make money. Making the switch makes no difference.

As for your FPS comment between games its bogus. Try matching the graphics settings and see where you FPS ends up. Shadows are still typically CPU based in many games except for features like SSAO etc. Depending on stuff like that performance can be immediately boosted upwards.

There is a reason Shogun 2 had a CPU only benchmark (CPU being the limiting factor) clock speed and memory speeds directly impacted performance.
That issue remains today but now add on more features, more complex animations, better physics, etc. Everyone likes to complain but you know what.

GO back and play Medieval II total war. Guess what same number of soldiers on Screen vs Warhammer total war FPS is higher in Warhammer. Thats with a 5.0 GHz 8700k and 3466 MHz memory. Even then Melee in the old title = 50-60 fps. In Warhammer I see an average of 80+

Looking at warhammer total war today AMD Ryzen performs awful nearly 30-40% slower than Intel due to internal latency caused by the CCX design. Even so the games run quite well by today's standard
Posted on Reply
#15
VashCZ
crazyeyesreaper said:

As for your FPS comment between games its bogus. Try matching the graphics settings and see where you FPS ends up. Shadows are still typically CPU based in many games except for features like SSAO etc. Depending on stuff like that performance can be immediately boosted upwards.
Specifically my GPU settings were all lowest but unit size(large, dunno why primarly in graphics setting when it is clearly mainly gameplay setting) Because when 2 vs 2 armies fight hard it is usually no honey for the system. But inevitably in late game there are 3vs3 or 4vs4 battles. And that is soooo baaaad and that why I don t play TW every now and then just like... Dota can be good example. Not mentioning tiny battlemaps :-/ etc..
DX 11 first core got overstuffed. That s it.

crazyeyesreaper said:
Its quite simple if you run threads on separate cores and instructions are in that specific cores cache sending that data to another core takes time. With AMD processors and the CCX layout when data needs to cross to a different core complex there is much higher latency than is typical with a monolithic (large die) CPU. so if latency was 20 ns between cores, that might slow things down but now imagine if it crosses to another CCX that adds 80 ns of latency. Once the data is sent work needs to be done then sent back and the data synced between the various cores. The more cores the more waiting the higher the latency. Now there are ways to deal with this.
https://hexus.net/media/uploaded/2017/3/1f34ebdb-fd8d-40de-9085-54cf055c02bf.jpg - but this thing is not monoblock?
Wow thought that Ryzen works more like Intel CPU than Bulldozer :-/
So when I buy 6 cores highly clocked Intel will DX 12 Total war use it all whole?

crazyeyesreaper said:

CA could sync animations etc via unit rather than soldiers. Thus making battles between singular units run on a single core and divide out from there. But thats far more complex. Bottom line is companies want to make money. Making the switch makes no difference.
I do not feel I can judge this. Also because I don t know what beauty would be taken off or not.
So you think it is not worth for them for further games(as they probably plan many) and they can optimize game more efficiently in different parts, good to know.

Allright so it sounds like they do what they can and can not do much better because of existing CPUs on the market. Right?
OK I am considering buying one of Warhammers when bundled with DLCs for some reasonable price. Which one is better?

Thanks.
Posted on Reply
#16
yakk
VashCZ said:
I admit I did not play DX12 Warhammers. Also did not try Brittania as articles said it is technically sucessor of Attila instead of Warhammer games. So maybe there is improvement thanks to new API and game optimization in Warhammers. Is this game in your opinian going to perform better than Warhammers?
DX12 for Warhammer is just a wrapper so it should be a bit slower.
Posted on Reply
#17
crazyeyesreaper
Chief Broken Rig
VashCZ said:
Specifically my GPU settings were all lowest but unit size(large, dunno why primarly in graphics setting when it is clearly mainly gameplay setting) Because when 2 vs 2 armies fight hard it is usually no honey for the system. But inevitably in late game there are 3vs3 or 4vs4 battles. And that is soooo baaaad and that why I don t play TW every now and then just like... Dota can be good example. Not mentioning tiny battlemaps :-/ etc..
DX 11 first core got overstuffed. That s it.


https://hexus.net/media/uploaded/2017/3/1f34ebdb-fd8d-40de-9085-54cf055c02bf.jpg - but this thing is not monoblock?
Wow thought that Ryzen works more like Intel CPU than Bulldozer :-/
So when I buy 6 cores highly clocked Intel will DX 12 Total war use it all whole?


I do not feel I can judge this. Also because I don t know what beauty would be taken off or not.
So you think it is not worth for them for further games(as they probably plan many) and they can optimize game more efficiently in different parts, good to know.

Allright so it sounds like they do what they can and can not do much better because of existing CPUs on the market. Right?
OK I am considering buying one of Warhammers when bundled with DLCs for some reasonable price. Which one is better?

Thanks.
It wont use all cores entirely but the fact is depending on the situation Total War does scale across more cores than people expect. The game is just very memory / cache sensitive. Has been since Empire Total War.

Shogun 2 was the best example with its ultra low CPU only benchmark. Again back then DDR3 going from 1066 -> 1333 -> 1600-> 1866 -> 2133 etc resulted in nominally a 1 FPS boost at each step. Back then when CPU bogged down due to melee battles. with most cpus being in the 20 fps range. going from say 20 to 25 FPS was nothing to sneeze at adding higher clock speeds further scaled it due to IPC. Also overclocking and higher memory frequencies also resulted in faster cache speeds. All this culminated in large gains. There were well known sweet spots back then for maximizing the CPU performance on that title because that CPU specific benchmark existed.

As for AMD, no they have an MCM design (multi-chip module) so depending on the CPU they have 2 4 or 6 etc MCMs connected with infinity fabric. If a workload moves from one CCX (core complex) to another through said infinity fabric their is a delay. Thus Ryzen CPUs still fall behind in Total War titles even now. However intel side performance still goes up with more cores. Additional content such as the blood pack DLCs also hit the CPU harder further resulting in performance drops.

Basically the engine is multi-threaded however the tech to make it work better just isn't available. Now CA has made serious strides in this regard but for most people yeah 4 v 4 stacks is just not going to work well.

If each army is 2000 men thats 16000 soldiers that the CPU has to calculate once again: animations, pathfinding for animation and movement, physics, AI tactics, shadows, extra effects (blood pack DLC) etc.

AND whille DX12 is just a wrapper for Warhammer in the first game it made a good performance boost for my system and I have seen overall better and far more stable performance vs DX11 mode in that title. In fact in large melee battles the Warhammer games run far better than the older Atilla / Rome II.

Another way to look at it what other game series offers that number of soldiers on screen with that level of animation detail and with this level of graphical fidelity? None basically. Its far more difficult to optimize than people likely realize.

Granted the charts below are not indicitive of a very large melee. But as you can see performance gains were made and in general more cores still helps quite a bit. However eventually a bottleneck is reached no matter what and while thats a shame it is still something that just scream "optimize it better" isn't gonna fix. Until tech advances further you can only do so much. I would suggest paying more attention to the DX11 side and the minimum FPS numbers. i3 4360 + GTX 1080 = 23.8 FPS min vs i5 4690 + GTX 1080 = 29.7 FPS min while a 5930K with quad channel memory it also has 20 way cache vs 12 way cache on the other processors. You will notice even in a game where single thread is the bottleneck that massive gains are still visible. Again problem is when doing 4 v 4 thats 8 armies up to 16k troops and thats a whole lot of number crunching.

This is why you see many other games with (units) tend to have them act as a cohesive whole with very little individuality. It simplifies CPU load thereby better performance. Until CA writes an entirely new engine from scratch with Vulkan / DX12 AND works out a way to move animation / pathfinding etc to more cores. This problem will persist.

Posted on Reply
#18
Xuper
By looking at PC 60 FPS vs PC Recommended spec , Pretty sure They don't have Vega and haven't test it yet.
Posted on Reply
#19
Enterprise24
WH2 cannot take advantage from 8700K. Max utilize is around 33% (with one thread running at 90-100% 4-5 threads at 20% or so and the rest is "idle") even with 720p low settings. No matter in DX11 or DX12.
I also test 4C4T vs 4C8T vs 6C6T vs 6C12T (at 5Ghz) and ensure that GPU isn't bottleneck. The result is pretty much the same.
I doubt 3K will. Totalwar is always singlethread intensive not multithread.
CPU killer is always shadows + unit details on modern CPU while terrain and unit size can also hurt older CPU.
I am sure I will need to disable shadows + unit details again to prevent sub 30fps when zoom in large battle.
Posted on Reply
#20
droopyRO
I gave up on stable 60 fps in Total War at 1440p. Maybe at 1080p you can achieve that, but in big 2 v 2 battles or large siege ones, framerate will not be stable.
As for the game, after playing about 1000 hours of Warhammer 1 and 2, these historical Total War games look boring, same units with different colors fighting it out, no magic, no flying units. Maybe when the game is 10$ i will pick it up.
Posted on Reply
#21
VashCZ
crazyeyesreaper said:
It wont use all cores entirely but the fact is depending on the situation Total War does scale across more cores than people expect. The game is just very memory / cache sensitive. Has been since Empire Total War.

Shogun 2 was the best example with its ultra low CPU only benchmark. Again back then DDR3 going from 1066 -> 1333 -> 1600-> 1866 -> 2133 etc resulted in nominally a 1 FPS boost at each step. Back then when CPU bogged down due to melee battles. with most cpus being in the 20 fps range. going from say 20 to 25 FPS was nothing to sneeze at adding higher clock speeds further scaled it due to IPC. Also overclocking and higher memory frequencies also resulted in faster cache speeds. All this culminated in large gains. There were well known sweet spots back then for maximizing the CPU performance on that title because that CPU specific benchmark existed.

As for AMD, no they have an MCM design (multi-chip module) so depending on the CPU they have 2 4 or 6 etc MCMs connected with infinity fabric. If a workload moves from one CCX (core complex) to another through said infinity fabric their is a delay. Thus Ryzen CPUs still fall behind in Total War titles even now. However intel side performance still goes up with more cores. Additional content such as the blood pack DLCs also hit the CPU harder further resulting in performance drops.

Basically the engine is multi-threaded however the tech to make it work better just isn't available. Now CA has made serious strides in this regard but for most people yeah 4 v 4 stacks is just not going to work well.

If each army is 2000 men thats 16000 soldiers that the CPU has to calculate once again: animations, pathfinding for animation and movement, physics, AI tactics, shadows, extra effects (blood pack DLC) etc.

AND whille DX12 is just a wrapper for Warhammer in the first game it made a good performance boost for my system and I have seen overall better and far more stable performance vs DX11 mode in that title. In fact in large melee battles the Warhammer games run far better than the older Atilla / Rome II.

Another way to look at it what other game series offers that number of soldiers on screen with that level of animation detail and with this level of graphical fidelity? None basically. Its far more difficult to optimize than people likely realize.

Granted the charts below are not indicitive of a very large melee. But as you can see performance gains were made and in general more cores still helps quite a bit. However eventually a bottleneck is reached no matter what and while thats a shame it is still something that just scream "optimize it better" isn't gonna fix. Until tech advances further you can only do so much. I would suggest paying more attention to the DX11 side and the minimum FPS numbers. i3 4360 + GTX 1080 = 23.8 FPS min vs i5 4690 + GTX 1080 = 29.7 FPS min while a 5930K with quad channel memory it also has 20 way cache vs 12 way cache on the other processors. You will notice even in a game where single thread is the bottleneck that massive gains are still visible. Again problem is when doing 4 v 4 thats 8 armies up to 16k troops and thats a whole lot of number crunching.

This is why you see many other games with (units) tend to have them act as a cohesive whole with very little individuality. It simplifies CPU load thereby better performance. Until CA writes an entirely new engine from scratch with Vulkan / DX12 AND works out a way to move animation / pathfinding etc to more cores. This problem will persist.


Thanks :-D.

Problem is, that I, as a player want to enjoy game. Late game is always the special part of biggest pleasures. But when late game becomes unplayable and it is KNOWN, is not a bug, but is a feature... what to do.
1. I remember unit sizing affected gameplay. All other but default "large" was wrong.
In Napoleon If I set very large, then I would not buy horses(low number to hunt behind enemy lines), because they are not many in one squad, oppose to infantrymen in squad.
But if I played small unit size then I would have 2/3 of army consisted of cavalry, as the ratio infantry/horsemen is much lower. Smaller units size - more cavalrymen against less infantrymen.
I am afraid these days to change it not to mess up gameplay.
2. It could be better to lower default size of units. Divided by 2 or like that so in big battles it would be no problem. Then add some micromanagment for each man.
I remember 4vs4 armies fights. Some of my first 2 active army men dies, my support arttilery riding from edge of map in the middle of map - somewhere behind enemy lines, being masacred by enemy horses. More squad dies, more coming from edge of map somewhere semirandomly on battlemap for death because no time to form some tactics.

I think this can be temporal solution until new engine is found out. Less actuall units at default. As you said it is mainly performance problem of each individual man.


droopyRO said:
I gave up on stable 60 fps in Total War at 1440p. Maybe at 1080p you can achieve that, but in big 2 v 2 battles or large siege ones, framerate will not be stable.
As for the game, after playing about 1000 hours of Warhammer 1 and 2, these historical Total War games look boring, same units with different colors fighting it out, no magic, no flying units. Maybe when the game is 10$ i will pick it up.
I have read before that there will be heroic units, which will have special abilities(not generals).
Posted on Reply
#22
bonehead123
Geeze, when are ya'll gonna realize that if ya can't play crysis at 2.46B FPS, nothin else matters, no matter what you do :respect: :laugh: :eek:
Posted on Reply
#23
dirtyferret
Doubtful the game needs anything more powerful then a ryzen 1600 or Intel 4790 for ultra settings.
Posted on Reply
#24
crazyeyesreaper
Chief Broken Rig
VashCZ said:
Thanks :-D.

Problem is, that I, as a player want to enjoy game. Late game is always the special part of biggest pleasures. But when late game becomes unplayable and it is KNOWN, is not a bug, but is a feature... what to do.
1. I remember unit sizing affected gameplay. All other but default "large" was wrong.
In Napoleon If I set very large, then I would not buy horses(low number to hunt behind enemy lines), because they are not many in one squad, oppose to infantrymen in squad.
But if I played small unit size then I would have 2/3 of army consisted of cavalry, as the ratio infantry/horsemen is much lower. Smaller units size - more cavalrymen against less infantrymen.
I am afraid these days to change it not to mess up gameplay.
2. It could be better to lower default size of units. Divided by 2 or like that so in big battles it would be no problem. Then add some micromanagment for each man.
I remember 4vs4 armies fights. Some of my first 2 active army men dies, my support arttilery riding from edge of map in the middle of map - somewhere behind enemy lines, being masacred by enemy horses. More squad dies, more coming from edge of map somewhere semirandomly on battlemap for death because no time to form some tactics.

I think this can be temporal solution until new engine is found out. Less actuall units at default. As you said it is mainly performance problem of each individual man.



I have read before that there will be heroic units, which will have special abilities(not generals).
Unit size affects CPU load as does shadows. However until the single core handling the animation and pathfinding gets bogged down frame rates will remain high. My advice at this point if you don't like lower frame rates with massive numbers of troops. Don't play the game or reduce unit sizes. Drop shadow settings a setting or two and if your running Blood DLC then disable it since it can impact frame rates by up to 20%.

I have played Empire Total War : 70 hrs / Medieval 2 Total War: 1000 hrs / Rome Total War: 1500 hrs / Rome II Total War: 488 Hrs / Atilla Total War: 170 Hrs / Warhammer Total War: 70 hrs / Total War Shogun II: 375 hrs

In all games at the highest difficult I seldom ever see more than 1-2 enemy armies within range of each other. IF they are they seldom are full stacks. This will differer per person playing obviously but suffice to say I don't experience big enough slow downs to impact my ability to play the game or my enjoyment. (Total War has been quite sporadic depending on system configuration).

That said for massive 4v4 fights CA would have been better off limiting engagements to 2v2 thus solving the issue. The problem is there just isn't a way to dsync animations / physics etc in a way that would work well in a multi-threaded environment as of yet. Need GPU compute to become more common with new tech that can leverage it. Right now well that requires a new engine. New game engines are expensive. Just ask Amazon about Lumber Yard or Crytek about Cryengine. Even Unreal needs constant cash influxes for R&D.

That said, when your a studio that makes primarily PC only games for a niche community. The cash return on spending millions developing state of the art game engines just isn't there. The cost to achieve this is high enough that no other game developers even attempt to compete in the little corner CA has entrenched themselves in. The last one to try was Neocore games with the King Arthure Role Playing War Game series and a few others. Even then they just recycled for multiple releases back to back to get their investment back.
Posted on Reply
#25
VashCZ
I understand.

I play shooters, RPGs and strategies... all good games.
But I don t buy games with stack and loads and tons of DLCs. Unless it is in package.
And since Arma 3 and TW Rome II I don´t buy unoptimized games. You can call it bad engine games. And DLC dependent games. These teached me well.

I think games like TW or Arma must have good gameplay mechanisms and play smoothly, both of them operate in big worlds and heavy load is inevitably part of late game.
I am pretty sure writing new engine would come back cash easy. It would be also very much futureproof. Rather than pushing to limits old days engine which served well for Empire/Napoleon and Shogun II loosing time by impossible optimization, they could simplify some sophisticated gameplay mechanisms. Less dice rolling. Because when I see those crowds on battlefield I don t undestand all stuff going on. And if lightweighted it would work 2x faster at 90% current gameplay feeling. They could add new mechanics and get new players. Not only reskin whole games that I don t even buy because they are incomplete and barely playable.

crazyeyesreaper said:

In all games at the highest difficult I seldom ever see more than 1-2 enemy armies within range of each other. IF they are they seldom are full stacks. This will differer per person playing obviously but suffice to say I don't experience big enough slow downs to impact my ability to play the game or my enjoyment. (Total War has been quite sporadic depending on system configuration).

That said for massive 4v4 fights CA would have been better off limiting engagements to 2v2 thus solving the issue.
It is probably mod that finally made more chalenging AIs. Defeating 2 armies with my 3 or 4 stack was reallly boring. I was not limited, AI was.
I remember enemy 5 stacks. I think I had only 2 so I autofight for less suffering x). But it is something... that makes me want to be better at playing. (Diplomatics usually fucks this up eventually. Many enemies declare me war. I can t pay them off even when I play economically strength; or friendlies I got since beginning declare war to some village size state which is puppet of supersized :-D; So I quit game cause too hard cause diplomacy. So easy or hard, not normal difficulty.)

You must realize, that limiting to exclusive 2vs2 armies fight is high limitation of strategy. Myself I winning battles with cheap upkeep/strength ratio soldiers. 2 or 3 my armies vs 1 max 2 of them. Partially early and mid game mostly.
When only 2vs2 is the limit, then there will be only elite troops hiring. And without mod to make longer way to elite troops technology tree it would be boring. Only elite, no diversion... why would there be cheap ones? Then delete war tech tree? ... not a good idea
Posted on Reply
Add your own comment