The TWC Total War Benchmark thread - compare your real TW performance

[alQamar]

19.08.2013: updated a lot of results to comply with the blood patch issue. Thanks anyone for your time and participation.
19.08.2013 added prooved results for CPU: AMD FX-4100
19.08.2013 added prooved results for GPU: AMD Radeon 6870, AMD Radeon 7970 Crossfire

[alQamar]

Hello everyone, I asked a moderator to clean up the thread. As the AMD "faction" refuse to deliver facts in form of own benchmarks on their rigs and submit their results instead of hot talks we will clean up the thread soon.

Obsolete stuff aswell as tech talk and discussion may cropped according to the thread rules.

I do not allow tech discussion in this thread according the thread rules to keep this thread informative and clean.

Aslong I get confronted by "contra" posts without a technical base and figures incl. logfiles I will not edit my tech myths and well objective weighted conclusions for that. I stopped counting people try telling me what everyworld rig can run TW on ultra* settings and everytime, yet, I prooved them they were wrong. And most of all they could accept they were wrong. That's life.



The figures we have are good enough to say AMD is not useable in TW in comparison to Intel. I / we did never say TW is not running on AMD but it is running comparingly SLOW (like hell) in certain scenarios.
If you cannot believe the numbers in the results, that is not my / our problem.
Causa finita.

Thank you for your understanding. If you still feel the need to fill the forums with presumptions certain things are wrong, or I may not use some particular words to face (non tech) people the truth please open a separate thread.

I am sorry if my words may sound harsh but all of you should be able to read the OP and so should notify the rules in this thread.
If you just pick up the discussion without facing facts like rules and results I cannot help you either.

remarks:
*ultra settings: are in the benchmark preparation files attached if you have more than 30 fps (the border when an ordinary human eye sees a fluid picture) you can count I would not deny you are allowed to say "i can run it on ultra) If you are below 30 in ANY of both test you cannot claim it runs ultra but it is stuttering. Of course the resolution is important, it would be ridicoulous outdated to fulfill the above conditions on a 1280x1024 resolution, which may be eligble for this person but does not represent the most people on Steam, that use 1080p (1920x1080)

I am yet thankful for anyone with:
- an Intel Dual Core (like E8600)
- an Intel Quad Core (like Q9550)
- an Intel i3 Core
- any laptop configuration especially with gaming GPU (760M or similar)
- and any AMD processors or AMD GPUs

to submit their results. Dont feel blamed if you are not top of the hill, your results still make a lot of sense for this thread and are really ESSENTIAL.

We need you and this thread lives only with your participation.

I planned to make a video for benchmarking, for those that may not understand my writings or are just to lazy. TL;DR don't help.

Should I start this video recording?

[gealany]

I do not allow tech discussion in this thread according the thread rules to keep this thread informative and clean.

The name of the Thread is "The TWC Total War Benchmark thread - compare your real TW performance" and not "The TWC Total War Benchmark thread - compare your ultra, hyper, mega settings TW performance".

Everybody who says, that AMD CPU's not unusable for shogun2, is lying.
When you say, AMD CPU's are not good for ultra settings... maybe then you are right.

Myth: I cant play Shogun 2 or Rome 2 with my AMD CPU!
Answer: You cant play Shogun 2 and Rome 2 on ultra settings by full hd solution. But on lower details with a smaller unit size you can play very well...

BTW: I'm not a AMD-Fanboy. I'm playing on a Q6600...

BTW 2: Weil mein englisch nicht so gut ist, hier nochmal 2 Sätze auf deutsch. Du nimmst dir hier raus repräsentativ zu sein. Das fängt bei den Systemeinstellungen an, die der größte Teil der Spielerschaft ohnehin nicht benutzt und hört bei den Tests von Usern auf die du sammelst. Du hast keinen Schimmer, was auf den einzelnen Rechnern noch im Hintergrund läuft und wie sauber die Systeme sind. Ich finde die Grundidee hinter dem Thread gut, aber pauschal zu sagen, dass AMD nicht funktioniert ist Blödsinn.

[alQamar]

The name of the Thread is "The TWC Total War Benchmark thread - compare your real TW performance" and not "The TWC Total War Benchmark thread - compare your ultra, hyper, mega settings TW performance".

Everybody who says, that AMD CPU's not unusable for shogun2, is lying.
When you say, AMD CPU's are not good for ultra settings... maybe then you are right.

Myth: I cant play Shogun 2 or Rome 2 with my AMD CPU!
Answer: You cant play Shogun 2 and Rome 2 on ultra settings by full hd solution. But on lower details with a smaller unit size you can play very well...

BTW: I'm not a AMD-Fanboy. I'm playing on a Q6600...

BTW 2: Weil mein englisch nicht so gut ist, hier nochmal 2 Sätze auf deutsch. Du nimmst dir hier raus repräsentativ zu sein. Das fängt bei den Systemeinstellungen an, die der größte Teil der Spielerschaft ohnehin nicht benutzt und hört bei den Tests von Usern auf die du sammelst. Du hast keinen Schimmer, was auf den einzelnen Rechnern noch im Hintergrund läuft und wie sauber die Systeme sind. Ich finde die Grundidee hinter dem Thread gut, aber pauschal zu sagen, dass AMD nicht funktioniert ist Blödsinn.

Blödsinn is you do not understand how the bechmarks work. The cpu bench is independent from your gpu and the ultra settings only apply to graphics.

I explained myself why we need to use ultra settings in the GPU bechmark

Instead of endless discussions no one of those complaint submit only one test while all AMD CPU results are from members that do not complain about facts we have collected by them. Without these brave members we would not have any compareable figure about AMD yet.

The x4 465 system is one I've installed my own and the results are similar to others so your presumptions of configuration issues disqualify themselves.


To your german part:

You won't like to tell me that any bad configured or background task on a system is causing a 50% performance gap randomly on all tested rigs. That's silly.

The cpu tests are not running on modified settings just the resolution is 1080p and blood pack disabled (for good).

Anyway it would be cool if can just run the cpu bechmarks aswell for your system.

[A Barbarian]

I found a problem with the crossfire profile in the AMD 13.8 beta (version 1 & 2) drivers (possibly affects also older versions). When zooming in on the troops in a battle, framerate is only about half of what it should be and GPU utilization is only about 50% (the closer the worse). For some reason this seems not to affect the benchmark (maybe because it's launched by the benchmark batch file).

The solution is to use the Civilization DX11 crossfire profile in the "3D Application settings" in the CCC. For AMD CrossFire Mode first select 'Use AMD pre-defined profile' and then select 'CivilizationV_Dx11.exe'. With that profile, GPU utilization will be close to 100% again (if not CPU limited) and performance will almost double.

[Keithzor.]

I did a benchmark test on high and got 28.6125. Works fine for me.

[alQamar]

°°° not only results stating please submit logfiles according to the thourough "how to" Thank you

[Keithzor.]

Is that what you wanted? I can run it on Ultra now.

[A Barbarian]

@Baldos
There is actually almost no performance difference between the 3570K/3770K and their successors, the 4670K/4770K. The big difference here comes from a benching mistake. the 3570K result was done with blood pack enabled and the others without. That accounts for about 30% performance difference. The remaining difference stems almost only from the RAM speed (1333Mhz versus 1866Mhz). That is when the clock speed is the same. Also the 4670K actually has no hyperthreading. So, don't worry, there is absolutely no need to upgrade from Ivy Bridge to Haswell. Also, even a 2500K@4.6Ghz will be about as fast as a 4770K@4.2Ghz (with the same RAM speed).

[alQamar]

Hello everyone, I've updated some results and got a new conclusion in aswell:

The blood patch DLC is not only a nice feature (which is silly looking) but costs a lot of performance in big battles on the CPU. It does not affect your graphics card results. Thanks to babydoc and some others.

@Baldos
There is actually almost no performance difference between the 3570K/3770K and their successors, the 4670K/4770K. The big difference here comes from a benching mistake. the 3570K result was done with blood pack enabled and the others without. That accounts for about 30% performance difference. The remaining difference stems almost only from the RAM speed (1333Mhz versus 1866Mhz). That is when the clock speed is the same. Also the 4670K actually has no hyperthreading. So, don't worry, there is absolutely no need to upgrade from Ivy Bridge to Haswell. Also, even a 2500K@4.6Ghz will be about as fast as a 4770K@4.2Ghz (with the same RAM speed).

second that.

[Baldos]

A. Barbarian & alQamar
THANKS I feel much better now . I was worried that I had made a big mistake by not waiting for/getting the 4th Gen CPU/Mobo considering the price is almost the same.

[Biggus Splenus]

Hello everyone, I've updated some results and got a new conclusion in aswell:

The blood patch DLC is not only a nice feature (which is silly looking) but costs a lot of performance in big battles on the CPU.

That's what happens when you have performance hungry effects that aren't hard-coded into the engine

Also, even a 2500K@4.6Ghz will be about as fast as a 4770K@4.2Ghz (with the same RAM speed).

I've seen that it's more like; 2500K@4.8 = 4670K@4.2

[alQamar]

hello everyone,

I've added some more tech myths regarding Total War

This time I am focusing on i7 / hyperthreading performance and the blood patch DLC.

[Aragorn-Elessar]

@alQamar: thank you for your tutorial (for improve intel i3 performance +16%) here you have an image of my shogun 2 specs with no lag:https://plus.google.com/u/0/101954244147761344321/posts
Do you think rome 2 will be the same?

[alQamar]

@alQamar: thank you for your tutorial (for improve intel i3 performance +16%) here you have an image of my shogun 2 specs with no lag:https://plus.google.com/u/0/101954244147761344321/posts
Do you think rome 2 will be the same?

They made more LOD (layer of details) so we can hope that low end rigs like yours can run even better than in Shogun 2. Anyway we do not have facts. They put a curtain of silence on that matter, and that is not a good sign.
Great idea with the petition thread. I signed it immediately.

[Aragorn-Elessar]

They made more LOD (layer of details) so we can hope that low end rigs like yours can run even better than in Shogun 2. Anyway we do not have facts. They put a curtain of silence on that matter, and that is not a good sign.
Great idea with the petition thread. I signed it immediately.

Thx alQamar

[Dominian]

The tech myth thread is really shaping up nicely! Good job alQamar.

This thread is also very helpful in regards to CPU and RAM testing. But I was wondering what's up with the GPU tests?

We have a GTX 770 clocked in at 84 fps with FXAA , a GTX 780 with 84fps with MLAA and a Titan with 104fps MLAA. How can we compare the FXAA and the MLAA? Should the 780 be as fast as the 770 with that difference and why is my GTX 780 with 108fps faster than the tested Titan? Is it the overclocking? And if so, shouldn't we add this? People can save a lot of money by buying the OC 780 instead of a normal Titan.

[Dominian]

The tech myth thread is really shaping up nicely! Good job alQamar.

This thread is also very helpful in regards to CPU and RAM testing. But I was wondering what's up with the GPU tests?

We have a GTX 770 clocked in at 84 fps with FXAA , a GTX 780 with 84fps with MLAA and a Titan with 104fps MLAA. How can we compare the FXAA and the MLAA? Should the 780 be as fast as the 770 with that difference and why is my GTX 780 with 108fps faster than the tested Titan? Is it the overclocking? And if so, shouldn't we add this? People can save a lot of money by buying the OC 780 instead of a normal Titan.

[alQamar]

Of course we cannot compare FXAA to MLAA these figures, some of my own, are just named to make present the boost you get with this mode.
I don't know why the Titan is comparingly slow to your 780. Anyway I've read a lot of articles recently that a 780 can even be faster working than a stock Titan. Of course OC of the vendors on the cards vary the results but they are just a here to give an overview.

So far the GPUs are not influenced by
- number of CPU cores
- the blood DLC
- system RAM speed.

So the only things influencing are
- the CPU speed
- hyperthreading (negatively on i7, positively on i3)
- the AA / MLAA / FXAA settings
- and the vendor / user OC.
- driver version
- OS Version

but I validate every result.

For some graphics cards (EVGA) there are also firmware updates available that can change / improve things.

[crzyrndm]

So far the GPUs are not influenced by
- the blood DLC

Possibly because the GPU bench is a pre-battle cutscene with no blood involved. Having the setting on or off should have minimal/no impact until there is actually some combat (and therefore blood).

Also a quick note for anyone with borderline performance and a multiple monitor setup, just having a secondary screen rendering the windows background was costing me 2 fps in the CPU bench.