Total War SHOGUN 2 DX11 with built-in benchmark: even more demanding than Metro 2033!

[sniper_sung]

It has successfully owned my 2 x 6950 2GB Crossfire :banghead:

The latest 2GB patch from Steam now enables a small built-in benchmark, through the beginning scene of the demo battle.



The 3rd option is not hardware demanding at all. With the following steps it is possible to stress test your system:

a) Select "Play Total War: SHOGUN 2", and change everything to max in game options:



b) Exit game, and re-open the Steam startup menu and select "Benchmark Game Settings".

Here are results of 2 x 6950 2GB Crossfire:



and for single 6950 2GB:



and here is a plot of the output log file using Excel:



Crossfire suffers a lag spike in the 4th second or so. Not sure if others have the same problem.

I'd say such fps is not playable for multiplayer game sessions, or 3rd party MODs to enable larger size of troops.

[Mad1723]

Screenshots, any difference with DX9?

[Ket]

Astonishes me how a game comes out that has low FPS and suddenly its "demanding" and the thought never seems to occur to anyone the issue is actually piss poor optimisation in the game code.

[schoolslave]

Astonishes me how a game comes out that has low FPS and suddenly its "demanding" and the thought never seems to occur to anyone the issue is actually piss poor optimisation in the game code.

Hahahahahaha
Pretty visuals come at a price....duh.

[sniper_sung]

Astonishes me how a game comes out that has low FPS and suddenly its "demanding" and the thought never seems to occur to anyone the issue is actually piss poor optimisation in the game code.

This game fully uses at least 3 physical cores of my 980X@4GHz and I'd say the optimization is quite good (at least much better than other games). You can see that the GPU usage is very high, while the crossfire efficiency is almost 200%.

[cold2010]

This You are here ?

Shogun 2 updates DX11 patch, built to run sub-program, 6950CF killed, 580SLI dead or alive

[sniper_sung]

This You are here ?

Shogun 2 updates DX11 patch, built to run sub-program, 6950CF killed, 580SLI dead or alive

That's not one of the original post. The earliest was on chiphell.com I think.

Wait, what? You wanna use 1.2GB vram to mess up with a game specified by Mr. Vram? No no no, you won't succeed xD

[Origin_Unknown]

Astonishes me how a game comes out that has low FPS and suddenly its "demanding" and the thought never seems to occur to anyone the issue is actually piss poor optimisation in the game code.

please tell me how a game that can have approx 10,000 individual people fighting on a battle map isn't demanding while rendering the scenery etc etc

[BrightCandle]

I am seeing significant microstutter (>20%) as well on a 5970. Although it scales well a lot of the extra framerate is wasted. 2 way crossfire on any of ATI's cards does not appear to be enough to run this game on maximum graphics.

[Demo]

None of us can really determine if the game is piss in optimization or just runs like a dog either way the game is great!

[zalbard]

This game fully uses at least 3 physical cores of my 980X@4GHz and I'd say the optimization is quite good (at least much better than other games). You can see that the GPU usage is very high, while the crossfire efficiency is almost 200%.

It really doesn't mean much... Crysis 2, for example, loads both of my GPUs to the max, but FPS is still meh because it is using old DX9 shaders and rendering baths that take longer to execute compared to their DX11 counterparts.
The code algorithm could just be not very efficient, you should know this as a coder.
Doesn't mean that this is the case for Shogun 2. All the DX9 games with DX11 bolted on suck in performance department, though...

please tell me how a game that can have approx 10,000 individual people fighting on a battle map isn't demanding while rendering the scenery etc etc

2D or low polygon 3D models.

[killerxp100]

These are my settings. With HD 6870 crossfire I'm getting 40-50fps in a large battle. 1920x1200 res.

[ZenEffect]

Astonishes me how a game comes out that has low FPS and suddenly its "demanding" and the thought never seems to occur to anyone the issue is actually piss poor optimisation in the game code.

and what games do you have a credit to?

[NapalmV5]

2 x gtx580 (SLI) @ 950/2300
1920x1080
extreme settings/dx11
sgssaax2 (custom)
texture filtering lod bias @ -3
ambient occlusion @ high quality (custom)

40 pix: http://img694.imageshack.us/g/41262956.png/

[NapalmV5]

while those ^ are the settings i play every game at.. heres per OPs msaax8 ^^

40 pix: http://img855.imageshack.us/g/shogun...313565406.png/

[sniper_sung]

while those ^ are the settings i play every game at.. heres per OPs msaax8 ^^

40 pix: http://img855.imageshack.us/g/shogun...313565406.png/

That's a lower resolution. Could you run the benchmark with max settings at either 1920x1080 or 1920x1200 and upload the log file? And specify how many GPUs did you use? (2 or 3)

By the way: nice temperatures you've got there!

[NapalmV5]

That's a lower resolution. Could you run the benchmark with max settings at either 1920x1080 or 1920x1200 and upload the log file? And specify how many GPUs did you use? (2 or 3)

By the way: nice temperatures you've got there!

thanks lol that last line saved you from getting flamed about res

those of you at 1200p and higher how you guys expect me to run higher res than my hdtv's 1080p??? tell me please!! and i dont have anything else higher res

now you guys can certainly run at the same 1080p for comparisons sake

i thought sli and the on-screen info would be enough.. its 2 x 580 and if your wondering about tri-sli no its not tri-sli but heres a pretty accurate estimate: divide any of the fps in my screens by half and multiply by 3.. 99% accurate

you dont trust my screens ^ why would you believe a log file..

[sniper_sung]

thanks lol that last line saved you from getting flamed about res

those of you at 1200p and higher how you guys expect me to run higher res than my hdtv's 1080p??? tell me please!! and i dont have anything else higher res

now you guys can certainly run at the same 1080p for comparisons sake

i thought sli and the on-screen info would be enough.. its 2 x 580 and if your wondering about tri-sli no its not tri-sli but heres a pretty accurate estimate: divide any of the fps in my screens by half and multiply by 3.. 99% accurate

you dont trust my screens ^ why would you believe a log file..

It's not that I don't trust your screenshots, but it's hard for me to use my cards to reproduce the scenes you used, that's why a benchmark is necessary to make a (fair) comparison. I'm very curious to see the vram usage of the game under 8AA (which I've already seen thanks to you ), and curious to see the benchmark result of a dual 580 SLI setup (which I still have no luck to see any)... The only relevant result I've seen before was just a GTX590 getting 34 fps under 1200p at default clock frequency.. If you could run the benchmark at 1080p that would be great - windiff on the log file can tell if we used the same settings or not

[ZenEffect]

It's not that I don't trust your screenshots, but it's hard for me to use my cards to reproduce the scenes you used, that's why a benchmark is necessary to make a (fair) comparison. I'm very curious to see the vram usage of the game under 8AA (which I've already seen thanks to you ), and curious to see the benchmark result of a dual 580 SLI setup (which I still have no luck to see any)... The only relevant result I've seen before was just a GTX590 getting 34 fps under 1200p at default clock frequency.. If you could run the benchmark at 1080p that would be great - windiff on the log file can tell if we used the same settings or not

windiff will throw a time and date error. best to use hexcmp or similar program. any differences in the text would be highlighted in red

[sniper_sung]

windiff will throw a time and date error. best to use hexcmp or similar program. any differences in the text would be highlighted in red

Windiff worked perfectly when I was comparing my 4AA log file of 6950CF with that of 570SLI though

[sniper_sung]

Never mind, I have finally met someone else to upload the log file of 580 SLI. Now I can finally make the comparison:



Rig 1: i7 2600K [4C8T] @ 5GHz, 2 x 580 SLI @ 797/1594/2025, Driver 270.61
Rig 2: i7 980X [6C12T] @ 4.5GHz, 2 x 6950 [1536SP] CF @ 950/1375, Catalyst 11.5 WHQL + 11.5CAP1

Rig 1 got 46.6 fps average, and Rig 2 got 51.1 fps average.

It is said that the upcoming new driver from nVidia can further improve the performance in Shogun 2.

[cold2010]

Testing at maximum quality settings in the benchmark 1920x1080 MSAA 8X DX 11

With the activation of DirectX 11 benchmark were only graphics cards with memory capacity of 1536 and 2048 megabytes. Acceptable level of performance will show the graphic level Radeon HD 6970 or GeForce GTX 580. Comfortable same performance could only show their two-chip solutions.

,,,,,,,,,,,,,,,,,,,,,,

Testing at maximum quality settings in the game 1920x1080 MSAA 8X DX 11

The picture in this case is almost similar to the gaming benchmarks, but the position of AMD cards have gone down a little lower. As in the past the event playing without any problems can only holders of two-chip solutions.

,,,,,,,,,,,,,,,,,,,,,,

Testing in the gaming benchmark 1080P DX 11

Total War: Shogun 2 has the same benchmark and with their attitudes, conducted in 1920x1080 mode Full HD DirectX 11 and is available through the Start menu Steam. In this case, more or less acceptable performance will be compatible with DirectX 11 graphics card-level Radeon HD 6850 or GeForce GTX 460.

If we start from the purchase card just for this game, with a balanced quality setting, the currently most preferred look Radeon HD 6870 or GeForce GTX 560 Ti. But as you can see from the tests conducted above, this "extremals" to satisfy his "appetite" will require two-chip solutions.

.......................

Testing CPU-in game benchmark in DirectX 9 mode

As we can see none of our tested processors could not show a decent result in this game. From the results of testing of CPU-to Total War: Shogun 2, we can conclude that the game requires a minimum of four core processor with a very high clock frequency.

http://gamegpu.ru/RTS-/-Strategii/To...-test-GPU.html

[shoota]

Astonishes me how a game comes out that has low FPS and suddenly its "demanding" and the thought never seems to occur to anyone the issue is actually piss poor optimisation in the game code.

I'm currently reading this thread zoomed in on my phone and as I read this post I thought "I bet this is Ket", scrolled over and sure enough it was. Broken record much?

[kaktus1907]

wow it seems amd's gaming evolved really working well in benchmark.. but not enough to hold the crown in game

i3 faster than q9550.. time to change my old one

[sniper_sung]

Testing at maximum quality settings in the benchmark 1920x1080 MSAA 8X DX 11

No need the *complexity and uncertainty* introduced by testing it in game. There is a more reliable way:

For those with less than 1.5GB video memory but still wish to run the benchmark at max settings (i.e. 8AA), the following trick would do it:

Go to %AppData%\The Creative Assembly\Shogun2\scripts, open preferences.script.txt, locate the following line

gfx_video_memory 0; # gfx_video_memory <int>, Override available video memory (bytes) #

and change it into

gfx_video_memory 2147483648; # gfx_video_memory <int>, Override available video memory (bytes) #

That should do the trick.

I'm curious to see how good a 1GB card can get, cuz according to the OP on pcinlife his GTX 560 Ti 1GB ran the benchmark and got 10.2 fps average at 1920x1200 8AA, and 3 fps average at 2560x1600 8AA. The GTX 560 Ti 1GB in the review got twice fps there at 1920x1080 8AA, either because 1920x1200 just passes a hard breakpoint of vram, or because the reviewer has a freaky computer there :p

Also, a newer version of nvidia driver is incoming soon, improving the fps in the benchmark.