Multiple core CPU utilisation

[Crazyeyesreaper]

that was already covered in the Basement
i tested sandy bridge and Ivy Bridge ivy was tested with and without Hyper Threading it made 0 difference along with a quick comparison with Bit-techs CPU benchmark comparison which shows most of todays CPUs,

http://www.twcenter.net/forums/showthread.php?t=542432

for proper performance it simply came down to clock speed and IPC (instructions per clock)

Intel win IPC wise and with overclocks performance continues to scale, memory bandwidth also shows improvements
essentially if the i3 CPUs from Intels Ivy Bridge lineup could be overclocked they would offer performance on par with the 3770k which is 3x the price, sadly overclocking the lower end chips doesnt work out so well thus forced buy higher end K series. but regardless,
from biggest performance boost to smallest
IPC > Clock Speed > Memory Bandwidth > Core count after 2 cores.

[Plan C]

I did some tests myself:
http://www.twcenter.net/forums/showt...5#post11696755

I would very much appreciate if some Shogun 2 owners would participate in the benchmarks, especially if you have a 4-core/8-thread CPU and a Radeon graphics card. Just follow the link above.

Gash... I own a P150M (2760QM, 8GB RAM, HD 6990M) and I cannot disable either specific cores, or hyperthreading, not on a mobile CPU (option unavailable in BIOS).

As mentioned in another thread, I can only do it through manually modifying BIOS image and re-flashing - S2TW is just not worth it for me sorry.

I will follow the results, though!

[boxleitnerb]

that was already covered in the Basement
i tested sandy bridge and Ivy Bridge ivy was tested with and without Hyper Threading it made 0 difference along with a quick comparison with Bit-techs CPU benchmark comparison which shows most of todays CPUs,

http://www.twcenter.net/forums/showthread.php?t=542432

for proper performance it simply came down to clock speed and IPC (instructions per clock)

Intel win IPC wise and with overclocks performance continues to scale, memory bandwidth also shows improvements
essentially if the i3 CPUs from Intels Ivy Bridge lineup could be overclocked they would offer performance on par with the 3770k which is 3x the price, sadly overclocking the lower end chips doesnt work out so well thus forced buy higher end K series. but regardless,
from biggest performance boost to smallest
IPC > Clock Speed > Memory Bandwidth > Core count after 2 cores.

The main reason why I'm posting this thread is because the weird behavior the Nvidia graphics driver can exhibit. Look at my results, I'm confused. 2C/2T=2C/4T but significantly slower than 4C/4T. Doesn't make sense. So I would like someone with a Radeon to try and reproduce. Would be interesting especially for people without quadcores.

[Crazyeyesreaper]

ive done the tests it makes 0 difference on my 7970 absolutely zero, CPU is the limitation for me, in melee battles it simple comes down to the GPU using a tiny fraction of its performance while 1 core gets bottlenecked in the CPU Bench, been running this test and comparing it with various reviewers on other hardware tech sites such as the motherboard reviewer at Techpowerup among others. he has multiple chips and various speeds and platforms and at least on intel, the 1155 sandy and ivybridge and 2011 perform about the same and hes tested single dual and triple GPU configurations with AMD graphics cards performance difference always comes down to IPC then clock speed followed by system memory bandwidth. and we all get the same thing, you can throw more cores at it but its worthless to do so, simply clock speed and IPC is all their is to it,

IPC means better performance vs older generations at the same speeds example Sandy Bridge vs Ivy Bridge depending on the situation Ivy is 5-15% faster at the same clock speed thats because of IPC improvements, now add overclocks the IPC improvements remain clock speed pushes perfromance up further,

you issue could very well be driver overhead due to SLI, but its doubtful, but when using 2x GPUs variation in performance is larger than with single GPU setups, that said with Radeon cards from the 6000 and 7000 series CPU bench is rock solid consistant with Radeon cards ive tested. can run the bench again and again and always be within 1.5 FPS of each other. through multiple runs,

Could also be the blood pack DLC, if you have that installed it further tanks CPU performance and since the blood splatters are never the same every time it can effect scores slightly altho not enough to show HT making that big a difference its a possibility.


With that said ive spent a ridiculous amount of time running Shogun 2 CPU bench, but regardless of core count thread count etc as long as im at 2 real cores or better performance doesnt change.

[boxleitnerb]

SLI was disabled for almost all tests.
I know that more cores are useless with the Warscape Engine. However, if you go below a certain numbers of cores/threads with an Nvidia GPU installed, performance drops unexpectedly. The question is - does this happen with a Radeon GPU as well?

There definitely is an issue with the Nvidia driver and the number of threads:
http://www.forum-3dcenter.org/vbulle...d.php?t=514213
This has also been confirmed by the German Hardware magazine PCGH.de

Could you please do 3 quick runs with the proposed settings and post in the thread? That would be very helpful

[SonOfCrusader76]

It's from your thread, the figures come from the cpu benchmark results.

How do you go about doing a CPU only benchmark? I'd be interested to see my results.

[Crazyeyesreaper]

uh click on Shogun 2 in your steam library click the big PLay button at the bottom it will say DX9 CPU benchmark

[DavidtheDuke]

ive done the tests it makes 0 difference on my 7970 absolutely zero, CPU is the limitation for me, in melee battles it simple comes down to the GPU using a tiny fraction of its performance while 1 core gets bottlenecked in the CPU Bench, been running this test and comparing it with various reviewers on other hardware tech sites such as the motherboard reviewer at Techpowerup among others. he has multiple chips and various speeds and platforms and at least on intel, the 1155 sandy and ivybridge and 2011 perform about the same and hes tested single dual and triple GPU configurations with AMD graphics cards performance difference always comes down to IPC then clock speed followed by system memory bandwidth. and we all get the same thing, you can throw more cores at it but its worthless to do so, simply clock speed and IPC is all their is to it,

What sucks is that using multiple threads to do the battle calculations would be a nightmare, and I'm not even a programmer, but I can imagine that being very difficult (my best friend is one and he said multiple threads is hard when many factors are at play). Hopefully they can optimize the battle animation calculations. I believe these are horribly unoptimized, because previous TW could run the same general numbers easily. As I said, I do believe at least the latest Intel processors are barely being used by Shogun effectively; especially since the game isn't even 64-bit.

[SonOfCrusader76]

uh click on Shogun 2 in your steam library click the big PLay button at the bottom it will say DX9 CPU benchmark

How the heck did I miss that? Cheers. It'll be interesting just how much the poor CPU optimisation is holding my game back.

[alQamar]

perhaps they will have a public beta... then is the time to raise things again and benchmark ;9 ... and they'd better done their homework in that topic lads.. otherwise i will really think over to buy it.

[boxleitnerb]

Okay finished the benchmarks. In a nutshell:

Warscape engine benefits from 3 cores, although not perfectly, but reasonably well. Hyperthreading is utterly useless in this game unless you only have a singlecore CPU. Clock speed is everything above all else. Perfect scaling there.

[Totalheadache]

At this point I know that unless multi-core support is implemented I will simply not buy the game.

There is no excuse for TWS2's poor performance.

yeah +1

yeah crazy's got it spot on I think +1 to him as well.

tbh if Rome 2 does the same as Shogun2 it will be a really bad joke.

also not only the lag but the loading times must be related? i know running a faster cpu+overclock my PC is snappier and I dont think hard drive is 2 blame for the loading times.

[The Useless Member]

The last time i checked Napoleon TW worked great on an i7

Yup, it works great for core i7, especially mine, when I'm playing TW games, the game gives me seven idle processors.

Works great!

[G-Spike]

I dont know if it will be possible to fix those problems with multiple cores on warscape ever. I just got the feeling that this is such a misconstruction in the core of the engine.
Also i really dont get why you would give up a tile based campaign map. In my opinion this had a negative impact on gameplay because you couldnt really tell where you can position your army to get this map for your battle.
For example in shogun II i placed my army at a bridge, anticipating a great bridge battle but when i was on the battle map there wasnt even a river. Thats really annoying.
Also i dont like building slots.

there are quiet a view things that suck with warscape without even speaking about modding.. im just hoping that its gonna be different in rome but yeah.. not gonna happen i guess :/

[Andreius Pretorianus]

what matters more for TW games?

The procesor or the video card?

[Greve Af Göteborg]

Processor

[Shaggy1973]

and with R2 being yet another evolution game, we seem to be in trouble here...

CA any word?

[General George S. Patton]

and with R2 being yet another evolution game, we seem to be in trouble here...

CA any word?

Its actually a cross between evolution and revolution, so we get the best of both worlds for rome 2

[windsupernova]

what matters more for TW games?

The procesor or the video card?

Both, but ... generally for games and for what people want more in games this days(awesome graphics at good resolutions) a video card is much more important. You can skimp a little on the CPU for a gaming computer.

[General George S. Patton]

Both, but ... generally for games and for what people want more in games this days(awesome graphics at good resolutions) a video card is much more important. You can skimp a little on the CPU for a gaming computer.

If you do this though, sometimes your running at like 40fps with lag.