TWC Total War Attila benchmark thread - last update: 26. December 2016

[alQamar]

Nice nice! You should upgrade to Win10, some people reported better FPS which is explainable due a better driver stack (WDDM 2.0)

[A Barbarian]

Nice nice! You should upgrade to Win10, some people reported better FPS which is explainable due a better driver stack (WDDM 2.0)

Yes, I am planning to but I am not in a hurry.

Here is actually a nice comparison between games on Win7/8/10:
http://pclab.pl/art63999-68.html

[Sharpe]

I always lag with like 10k plus men. Any tips lads?

[A Barbarian]

I always lag with like 10k plus men. Any tips lads?

It's pretty difficult not to get any lag in Attila with 10k + units. But what's your configuration ?

[Sharpe]

I5-4570
GTX 670 2gb
8GB ram

[A Barbarian]

@Sharpe
Ok, nothing to overclock in that hardware configuration. So it's only possible to reduce some settings. CPU heavy settings are 'shadows' and 'unit detail' (besides unit size of course). Reduce those 2 and see how performance does. Also, posting your settings screen would be helpful as well.

But due to the lacking optimization in Attila, even a super high-end PC will lag somewhat in Attila with 10k+ units.

[A Barbarian]

@alQamar
I noticed something interesting when comparing my SLI to the non-SLI results. In CPU limited benchmark areas, the SLI performance can be up to 10% lower compared to a single GPU. This sucks, lol.

Most extreme example is the 'Quality' preset which is completely CPU limited in SLI and single GPU as well.

SLI vs single 970
avg 69.7 min 57 max 94 (SLI)
avg 75.1 min 61 max 96 (single 970)

Test environment: 4770K@4.5Ghz, RAM 2400Mhz@10-12-12-32, Win 10, NumThreads=8 optimization in Attila configuration file (this boosts benchmark results by up to 10% with my 4 core + HT CPU)

Spoiler Alert, click show to read: 

SLI



Single GPU:

[alQamar]

wow... speechless. I've lost hope to see any change in Warhammer mate.
would you mind to post a full result set as others did? So posting single GPU and SLI with all the presets + maxed out settings in 1080p? I really would like to add them.

[A Barbarian]

Here u are

1920x1080 with Intel Core i7 4770K@4.5GHz, 32GB RAM 2400MHz CL 10-12-12-32, 2xASUS STRIXX GTX 970@1542@3900, Windows 10 pro 64bit, Resolution 1920x1080p,
Other: 8 'Number of threads' setting in config

single GPU:

extreme quality: (almost only GPU limited)
max FPS: 40
min FPS: 24
avg FPS: 34.5

maximum quality: (mostly GPU limited)
max FPS: 66
min FPS: 44
avg FPS: 54.2

quality: (more CPU limited)
max FPS: 96
min FPS: 61
avg FPS: 75.1

performance: (more CPU limited)
max FPS: 124
min FPS: 83
avg FPS: 103.3

maximum performance: (more CPU limited)
max FPS: 199
min FPS: 137
avg FPS: 175.8


SLI

extreme quality: (almost only GPU limited)
max FPS: 67
min FPS: 45
avg FPS: 53.7

maximum quality: (CPU/GPU limitation about 50/50)
max FPS: 79
min FPS: 47
avg FPS: 58.7

quality: (more CPU limited)
max FPS: 94
min FPS: 58
avg FPS: 69.7

performance: (more CPU limited)
max FPS: 115
min FPS: 77
avg FPS: 96

maximum performance: (mostly CPU limited)
max FPS: 251
min FPS: 148
avg FPS: 179

All benchmark screenshots in a zip file:
https://i9o5ta.bl3301.livefilestore....ownload&psid=1

[Druout]

Not trying to double post, but I can't edit my post and I had accidentally slammed my 880M results inside my 290X results, If someone could either delete my previous post or tell me how to do it I would appreciate it.

Attilla Performance Nvidia Driver 353.62
5930K@4.5Ghz /32GB 2400Mhz RAM with Matrix 780Tis CPUz : http://valid.x86.fr/cxqxb1
GPUz : http://www.techpowerup.com/gpuz/details.php?id=p7hh
Single 780Ti @1080p:

extreme quality: http://s1068.photobucket.com/user/lo...t3mku.jpg.html
max FPS: 44
min FPS: 28
avg FPS: 38.6

maximum quality: http://s1068.photobucket.com/user/lo...7kgmu.jpg.html
max FPS: 71
min FPS: 43
avg FPS: 57.6

quality: http://s1068.photobucket.com/user/lo...bfvqw.jpg.html
max FPS: 101
min FPS: 54
avg FPS: 72.6

performance: http://s1068.photobucket.com/user/lo...4qb3l.jpg.html
max FPS: 116
min FPS: 51
avg FPS: 86.7

maximum performance: http://s1068.photobucket.com/user/lo...5uai2.jpg.html
max FPS: 250
min FPS: 131
avg FPS: 173.8


SLI @1080p

extreme quality: http://s1068.photobucket.com/user/lo...nvsrb.jpg.html
max FPS: 72
min FPS: 03
avg FPS: 50.9 (34.1 on 347.88)

maximum quality: http://s1068.photobucket.com/user/lo...h0jgw.jpg.html
max FPS: 75
min FPS: 03
avg FPS: 52

quality: http://s1068.photobucket.com/user/lo...k1ail.jpg.html
max FPS: 90
min FPS: 03
avg FPS: 62.4

performance: http://s1068.photobucket.com/user/lo...woaue.jpg.html
max FPS: 110
min FPS: 03
avg FPS: 81.7

maximum performance: http://s1068.photobucket.com/user/lo...tml?sort=3&o=2
max FPS: 194
min FPS: 08
avg FPS: 142.0

single 780Ti @1440p:

extreme quality: http://s1068.photobucket.com/user/lo...zphsm.jpg.html
max FPS: 44
min FPS: 28
avg FPS: 24.6 (22.8 with 347.88)

maximum quality: http://s1068.photobucket.com/user/lo...z0ilk.jpg.html
max FPS: 51
min FPS: 29
avg FPS: 40 (34.7 with 347.88)

quality: http://s1068.photobucket.com/user/lo...rpoe0.jpg.html
max FPS: 81
min FPS: 48
avg FPS: 63.7

performance: http://s1068.photobucket.com/user/lo...vnajn.jpg.html
max FPS: 97
min FPS: 49
avg FPS: 83.8

maximum performance: http://s1068.photobucket.com/user/lo...3iuqv.jpg.html
max FPS: 189
min FPS: 91
avg FPS: 145.3

780 Ti SLI @1440p

extreme quality: http://s1068.photobucket.com/user/lo...yhvj8.jpg.html
max FPS: 51
min FPS: 01
avg FPS: 42 (23.3 on 347.88)

maximum quality: http://s1068.photobucket.com/user/lo...zbyy8.jpg.html
max FPS: 72
min FPS: 03
avg FPS: 51.6

quality: http://s1068.photobucket.com/user/lo...spwry.jpg.html
max FPS: 87
min FPS: 03
avg FPS: 52.9

performance: http://s1068.photobucket.com/user/lo...7qrv2.jpg.html
max FPS: 112
min FPS: 03
avg FPS: 81.9

maximum performance: http://s1068.photobucket.com/user/lo...dt33n.jpg.html
max FPS: 204
min FPS: 6
avg FPS: 147.1

4770K @3.89Ghz/16GB 1600Mhz RAM with Matrix 290X CF CPUz link http://valid.x86.fr/wr0vjp
GPUZ link: http://www.techpowerup.com/gpuz/details.php?id=wm488

Single 290X @1080p:

extreme quality: http://s1068.photobucket.com/user/lo...kvmne.jpg.html
max FPS: 36
min FPS: 23
avg FPS: 31.3

maximum quality: http://s1068.photobucket.com/user/lo...hunhr.jpg.html
max FPS: 56
min FPS: 33
avg FPS: 43.3

quality: http://s1068.photobucket.com/user/lo...cpgxi.jpg.html
max FPS: 69
min FPS: 43
avg FPS: 52.3

performance: http://s1068.photobucket.com/user/lo...owgzl.jpg.html
max FPS: 89
min FPS: 56
avg FPS: 70

maximum performance: http://s1068.photobucket.com/user/lo...rucwa.jpg.html
max FPS: 197
min FPS: 81
avg FPS: 129.8

290X CF@1080p:

extreme quality: http://s1068.photobucket.com/user/lo...olkkn.jpg.html
max FPS: 62
min FPS: 11
avg FPS: 44.7

maximum quality: http://s1068.photobucket.com/user/lo...ppvjh.jpg.html
max FPS: 57
min FPS: 11
avg FPS: 44.2

quality: http://s1068.photobucket.com/user/lo...ycze6.jpg.html
max FPS: 66
min FPS: 13
avg FPS: 54.6

performance: http://s1068.photobucket.com/user/lo...i6k0h.jpg.html
max FPS: 86
min FPS: 13
avg FPS: 69.4

maximum performance: http://s1068.photobucket.com/user/lo...pcmkd.jpg.html
max FPS: 170
min FPS: 33
avg FPS: 115.3

Single 290X @1440p:

extreme quality: http://s1068.photobucket.com/user/lo...vkmy7.jpg.html
max FPS: 24
min FPS: 13
avg FPS: 20.1

maximum quality: http://s1068.photobucket.com/user/lo...eoqpu.jpg.html
max FPS: 41
min FPS: 24
avg FPS: 32.6

quality: http://s1068.photobucket.com/user/lo...s6a9b.jpg.html
max FPS: 64
min FPS: 41
avg FPS: 52.2

performance: http://s1068.photobucket.com/user/lo...a5yun.jpg.html
max FPS: 84
min FPS: 51
avg FPS: 70

maximum performance: http://s1068.photobucket.com/user/loki1944/media/1440p%20Maximum%20Performance_zpsyxpuxlwc.jpg.html
max FPS: 161
min FPS: 79
avg FPS: 116.6

290X CF@1440p:
extreme quality: http://s1068.photobucket.com/user/lo...mtka2.jpg.html
max FPS: 47
min FPS: 06
avg FPS: 40

maximum quality: http://s1068.photobucket.com/user/lo...3bogq.jpg.html
max FPS: 56
min FPS: 13
avg FPS: 45.1

quality: http://s1068.photobucket.com/user/lo...adunf.jpg.html
max FPS: 66
min FPS: 13
avg FPS: 54.2

performance: http://s1068.photobucket.com/user/lo...bbek7.jpg.html
max FPS: 84
min FPS: 13
avg FPS: 69.2

maximum performance: http://s1068.photobucket.com/user/lo...dnwnf.jpg.html
max FPS: 172
min FPS: 23
avg FPS: 115.7

i7 4810MQ @3.6Ghz/GTX 880M/16GB 1600Mhz RAM
CPUz http://valid.x86.fr/kcmwwg
GPUz http://www.techpowerup.com/gpuz/details.php?id=mgzdp

880M@1080p:

extreme quality: http://s1068.photobucket.com/user/lo...7bomp.jpg.html
max FPS: 21
min FPS: 11
avg FPS: 16.8

maximum quality: http://s1068.photobucket.com/user/lo...pm54y.jpg.html
max FPS: 36
min FPS: 19
avg FPS: 25.9

quality: http://s1068.photobucket.com/user/lo...acswf.jpg.html
max FPS: 49
min FPS: 31
avg FPS: 42.7

performance: http://s1068.photobucket.com/user/lo...xv4go.jpg.html
max FPS: 68
min FPS: 38
avg FPS: 57.3

maximum performance: http://s1068.photobucket.com/user/lo...md5gm.jpg.html
max FPS: 110
min FPS: 70
avg FPS: 95.6


i7 4700MQ @2.4Ghz/GTX 765M 2GB/8GB 1600Mhz RAM
CPUZ: http://valid.x86.fr/gzkbs0
GPUZ : http://www.techpowerup.com/gpuz/details.php?id=3sn9h

GTX 765M@1080p:

extreme quality: http://s1068.photobucket.com/user/lo...iupox.jpg.html
max FPS: 09
min FPS: 0
avg FPS: 7.3

maximum quality: http://s1068.photobucket.com/user/lo...yvskj.jpg.html
max FPS: 12
min FPS: 0
avg FPS: 8.8

quality: http://s1068.photobucket.com/user/lo...o7t7o.jpg.html
max FPS: 15
min FPS: 0
avg FPS: 10.3

performance: http://s1068.photobucket.com/user/lo...1a7ns.jpg.html
max FPS: 21
min FPS: 0
avg FPS: 16.7

maximum performance: http://s1068.photobucket.com/user/lo...8xa84.jpg.html
max FPS: 46
min FPS: 0
avg FPS: 31.2

i7 U640 @1.2Ghz/GT 335M 1GB/8GB RAM
M11X CPUZ http://valid.x86.fr/e985td
M11X GPUZ http://www.techpowerup.com/gpuz/details.php?id=nbwxu

GT335M@1024x768

maximum performance: http://s1068.photobucket.com/user/lo...vwpx2.jpg.html
max FPS: 22
min FPS: 0
avg FPS: 16

[A Barbarian]

@Druout
Wow, that's a lot of testing. Have some reputation

Your results also show that in the CPU limited scenarios, SLI performs noticeable worse than single a GPU. For AMD/CF this seems to be less the case.

[alQamar]

@Druout
Wow, that's a lot of testing. Have some reputation

Your results also show that in the CPU limited scenarios, SLI performs noticeable worse than single a GPU. For AMD/CF this seems to be less the case.

yes, that's definitely worth a rep man. I would be very pleased if you could do something similar for your rig mate .

[A Barbarian]

yes, that's definitely worth a rep man. I would be very pleased if you could do something similar for your rig mate .

My single GPU + SLI results are 2 posts above, maybe u overlooked them

[Druout]

I just got my 980Ti Strix cards today, so I'll be able to test those single and SLI and I shifted my 780Tis over to my 4770K rig so I can see how that does as well. I was pretty disappointed by how the 780Tis did in SLI, in most games they seem to do very well at 1440p, but I'm guessing Maxwell has an edge or something, similar to how things go with those and Witcher 3 maybe? Not sure, but in other benchmarks like Tomb Raider/Hitman Absolution/Thief/Shadow of Mordor they handily beat stock 980 SLI.

[A Barbarian]

I just got my 980Ti Strix cards today, so I'll be able to test those single and SLI

Nice. Note that 'Quality' and 'Performance' are essentially CPU limited. Another interesting bit about the 780TI: It was faster than a GTX970 in presets with GPU limit but slower in presets with more CPU limits. The CPU power was with 4.5Ghz and a 4770K vs a 5930K about the same (Attila barely profits from the additional cores of the 5930K).

The big problem of TW games compared to most other games are that they are often CPU limited in graphics heavy situations. Especially the animation- and shadow calculations/drawing seems to be responsible for that. It's also possible that this problem is at least partially caused by the 'single threadedness of DX11.

You could try whether changing the "number_of_threads" setting from Attila configuration file in "%AppData%\Roaming\The Creative Assembly\Attila\scripts" helps the performance (at least in the benchmark).
Setting it to 8 for my 4770K raises the benchmark score up to 10% in CPU limited parts. You could try 8 or 12 (because the 5930K has 6 cores).

Have fun

Looking forward to the results.

[Druout]

Ok finally got the 980Ti benchmarks wrapped up; Single and 2 Way SLI for 1080p and 1440p and 2 way SLI for 7680x1440p. For 1920x1080 and 2560x1440 I tried to hyperlink the extreme benchmarks since uploading and linking 20+photos is a bit much for me. I ran these benchmarks @980Ti STRIX Asus clocks, no overclocking. I did try the extreme 1080p benchmark with +150Mhz Core and +70Mhz Memory, but the difference was not very discernible, couple of fps at most. For the most part SLI doesn't seem to help much and actually costs a few frames as well at 1080p/1440p; the major exception was extreme @1440p from what I saw. With SLI I often had a very low frame right as the benchmark started; like 1 or 3 as well.

i7 5930K@4.5Ghz/32GB 2400Mhz RAM/SLI 980Ti
CPUZ: http://valid.x86.fr/ulffzz
GPUZ : http://www.techpowerup.com/gpuz/details.php?id=bh5e
HW Monitor/CPUID/RAM/GPUz screenshot : http://s1068.photobucket.com/user/loki1944/media/proof%20pic_zpsmxgclyei.png.htmlhttp://s1068.photobucket.com/user/loki1944/media/proof%20pic_zpsmxgclyei.png.html
Single 980Ti @1080p:
extreme quality:
max FPS: 61
min FPS: 38
avg FPS: 53.3

maximum quality:
max FPS: 86
min FPS: 53
avg FPS: 64.7

quality:
max FPS: 102
min FPS: 68
avg FPS: 79.5

performance:
max FPS: 132
min FPS: 81
avg FPS: 101.8

maximum performance:
max FPS: 254
min FPS: 131
avg FPS: 172.0

980Ti SLI @1080p
extreme quality:
max FPS: 94
min FPS: 03
avg FPS: 61.7

maximum quality:
max FPS:77
min FPS: 06
avg FPS: 60.4

quality:
max FPS: 95
min FPS: 03
avg FPS: 72.3

performance:
max FPS: 110
min FPS: 03
avg FPS: 90.7

maximum performance:
max FPS: 249
min FPS: 08
avg FPS: 169.5



Single 980Ti @1440p:
extreme quality:
max FPS: 39
min FPS: 23
avg FPS: 34

maximum quality:
max FPS: 73
min FPS: 43
avg FPS: 57.7

quality:
max FPS: 104
min FPS: 61
avg FPS: 78.5

performance:
max FPS: 124
min FPS: 66
avg FPS: 99.0

maximum performance:
max FPS: 250
min FPS: 124
avg FPS: 174.1

980Ti SLI @1440p
extreme quality:
max FPS: 62
min FPS: 03
avg FPS: 53.9

maximum quality:
max FPS: 75
min FPS: 01
avg FPS: 59.5


quality:
max FPS: 97
min FPS: 03
avg FPS: 73.4

performance:
max FPS: 118
min FPS: 03
avg FPS: 94.6

maximum performance:
max FPS: 244
min FPS: 8
avg FPS: 163.7

980Ti SLI @7680x1440 (Surround with 3xROG Swift Gsync off)

extreme quality:
max FPS: 42
min FPS: 01
avg FPS: 28.5


maximum quality:
max FPS: 51
min FPS: 01
avg FPS: 43.3

quality:
max FPS: 62
min FPS: 03
avg FPS: 50.6

performance:
max FPS: 77
min FPS: 04
avg FPS: 66.6

maximum performance:
max FPS: 122
min FPS: 14
avg FPS: 103.7

[A Barbarian]

For the most part SLI doesn't seem to help much and actually costs a few frames as well at 1080p/1440p; the major exception was extreme @1440p from what I saw.

Yes, that's because of the CPU limitation in most benchmarks. I guess that there is some additional overhead in the drivers and/or the TW engine for handling SLI and that's why SLI is actually slower in CPU limited situations.

Personally, I use the 970s in SLI for DSR@4K in order to reduce the atrocious aliasing. Switching off SSAO and 'Screen Space Reflections' also helps SLI performance a lot. In order to reduce CPU load, it's most effective to lower shadows, unit size and/or unit details.

With SLI I often had a very low frame right as the benchmark started; like 1 or 3 as well.

Yes me too. And sometimes also in the single GPU benchmarks. I actually ignored those values when stating the min framerate because for me that's more of a benchmark glitch.

[Druout]

Yes, that's because of the CPU limitation in most benchmarks. I guess that there is some additional overhead in the drivers and/or the TW engine for handling SLI and that's why SLI is actually slower in CPU limited situations.

Personally, I use the 970s in SLI for DSR@4K in order to reduce the atrocious aliasing. Switching off SSAO and 'Screen Space Reflections' also helps SLI performance a lot. In order to reduce CPU load, it's most effective to lower shadows, unit size and/or unit details.


Yes me too. And sometimes also in the single GPU benchmarks. I actually ignored those values when stating the min framerate because for me that's more of a benchmark glitch.

Well, still decent performance on single cards; though I had to bench some other games to make sure my new cards were actually working in SLI due to how close single and SLI were in Attila. I don't notice the AA issue as much with 1440p myself, usually just run max settings and turn on MLAA at that resolution.

[Fanest]

This might be interesting to u guys

https://www.reddit.com/r/pcmasterrac..._support_dx12/

TL;DR: Gefore 900 series doesn't support Async compute\shaders, nvidia cards will suck at dx12 games, especially if you wanna use VR. (also it does explain why did nvidia cards had worse FPS in dx12 than in dx11 in some cases in that RTS benchmark.)

[Druout]

I'm not too worried about it, everything coming out in the near future will be DX11 compatible, and Warhammer TW will be DX11 compliant I'm like 150% sure based on the steam survey; by the time I upgrade to windows 10 it will be time for new cards anyway, I guess I'll get DX12 on my xbox one though lol. Frankly with how bad AMD drivers have been for me on my 290X CF rig the past 8 months I wouldn't be surprised if Nvidia pulls out ahead anyway. AMD driver updates are about as speedy as a life sentence and their crossfire profile updates are even worse from what I've seen and I had more instances of stuttering in CF than in SLI; which is very bad for VR. Also on the VR thing, I have a DK2 and I'm not going to bother with any more VR; I find surround way better and also much less of a PITA to get running; plus no dizziness.