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

[A Barbarian]

Finally, AMD released new drivers and they include a Attila CF profile which actually works

Performance wise, the 15.3 beta drivers are exactly the same as the previous 14.5

Here are the results:
min avg max
34 - 42 - 49 extreme
40 - 50 - 66 max quality
47 - 60 - 80 quality

Configuration:
Win7-64Bit
2x7970 @ 1050Mhz GPU, 1400Mhz VRAM
4770K @ 4.2Ghz
32GB RAM @ 1866Mhz

[alQamar]

CF = crossfire for anyone.

i haven't noticed a full release but only a beta release of the new AMD drivers. Is that correct? Can you please include your CPU-Z log Barbarian so I can take your results in? Also a run without CF would be a great addition.


max quality
21 / 34 / 26.5 - patch 1, 349.90
21 / 34 / 26.2 - RTM, 347.52

quality
36 / 61 / 44.8 - patch 1, 349.90
36 / 61 / 44.9 RTM, 347.52

performance
53 / 73 / 63.8 - patch 1, 349.90
53 / 73 / 63.5 - RTM, 347.52

max performance
63 /117 / 98.6 - patch 1, 349.90
63 /117 / 99.1 - RTM, 347.52

I've tested the new nvidia driver and in combination with patch 1 of Attila. All differences are so minor that we can assume measure tolerances instead of real performance impacts on the worse / better on my rig.

[SPARTAN VI]

Just tested nvidia's 347.88. No difference here either even though the release notes state improved SLI profile for Attila:

extreme quality:
max 42 / min 7 / avg 33,9 (driver 347.52)
max 42 / min 7 / avg 32,3 (driver 347.88)

maximum quality:
max 54 / min 31 / avg 40,6 (driver 347.52)
max 54 / min 32 / avg 40,4 (driver 347.88)

quality:
max 67 / min 40 / avg 49,4 (driver 347.52)
max 67 / min 40 / avg 48,4 (driver 347.88)

performance:
max 80 / min 56 / avg 66,0 (driver 347.52)
max 77 / min 56 / avg 62,6 (driver 347.88)

[LestaT]

i3-2100 @ 3.1Ghz
Win7-64Bit
4GB RAM @ 1333Mhz
Sapphire R7 260X 2GB
Catalyst Version 15.3 Beta

Max Performance

Spoiler Alert, click show to read: 

29 / 50.6 / 63

Spoiler Alert, click show to read: 

Performance

Spoiler Alert, click show to read: 

26 / 32.8 / 38

Spoiler Alert, click show to read: 

Quality

Spoiler Alert, click show to read: 

14 / 23.5 / 29

Spoiler Alert, click show to read: 

Max Quality

Spoiler Alert, click show to read: 

11 / 17.2 / 21

Spoiler Alert, click show to read: 

Extreme Quality

Spoiler Alert, click show to read: 

4 / 9.2 / 12

Spoiler Alert, click show to read: 

Running with 1900 x 1080. When I was using my 20' 1600 x 900 earlier when the game was released (before my daughter cracked my monitor screen) I think the fps was slightly higher. However I see no different when running 1600 x 900 resolution with this current 21.5" monitor.

p/s: from the advance setting can someone told me which is control by the cpu and which is by gpu. I know that unit size is by cpu so when I run the game on Max Quality but with Small unit size I can run the game much faster but since I can't play with anything lower than Ultra unit size I'm trying to find the best combination.

[Beorg]

Intel Core i7 2600K @ 4.31 GHz
8 GB 1333 MHz CL9 DDR3
XFX Radeon 7970 @ 1050/1500 MHz
Win 8.1 Pro x64
Catalyst 15.3 Beta
Total War: Attila 1.1 (pre-today's patch

All benchmarks done at 2560x1440. Imgur album with the results:

http://imgur.com/a/dfLh4



Extreme Quality:

Avg: 13.6, Min: 9, Max: 16


Maximum Quality:

Avg: 21.9, Min: 16, Max: 29


Quality:

Avg: 34.4, Min: 28, Max: 48


Performance:

Avg: 48.1. Min: 38, Max: 56


High Performance:

Avg: 64.2, Min: 41, Max: 88

[SPARTAN VI]

I noticed something interesting in some tests I ran yesterday: the first time I run the benchmark for a session, the performance is about 40% lower. It's as if it's not utilizing the second GPU in SLI. So in order to get "accurate" results, I have to start up a benchmark and immediately ESC out of it, then start it again.

Does anyone else notice this? Or is it just me?

[alQamar]

dont have an SLI but I could try to repro that. But wasn't it you that also had a similar issue back in time, I think you already said your results were better in a second run. This is still strange Sure no energy saving mode is enabled / or opposite have you set performance mode on in the Windows energy settings as suggested for the benchmark?

[Fresco]

Result of benchmark after this patch are lower than basic game:
Extreme Quality:

Avg: 16.8, Min:11, Max: 19 this patch: AVG:15,9 MIN:11 MAX 19


Maximum Quality:

Avg: 23.8, Min: 18, Max: 31 after this patch: Avg:23,1 Min:18 Max:29


Quality:

Avg: 40.3, Min: 31, Max: 54 after this patch: Avg:34,2 min:26 Max:49


Performance:

Avg: 57,2. Min: 48, Max: 68 after this patch: avg:48 Min:29 Max:61


High Performance:

Avg: 89.7, Min: 56, Max: 107 after this patch: avg:82.7 Min:51 Max:102


MY System: CPU i5-2500K OC@4,5 Ghz
GPU Zotac GTX 760 2GB AMP Edition!
RAM 8GB DDR3 1333Mhz
Mobo Asus Z-77A

[alQamar]

This can actually happen Fresco. Have a look at the Rome 2 extensive comparisons I've made and got banned back in time for the controversy and moderation issues it caused.

http://www.twcenter.net/forums/showt...ted-29-01-2015

Now that I was able to compare the game 100% objective using the same Windows state and driver it is revealed how much the performance gained on Rome 2 to in regards of using the benchmark at least.

I will do my next tests tomorrow but they are not very accurate as I have seen various changes like Upgraded my Windows 10 from b9926 to b10040, upgraded my graphics driver to 349.90
So cannot directly make the patch responsible if anything changed to the better or worse.

[alQamar]

I noticed something interesting in some tests I ran yesterday: the first time I run the benchmark for a session, the performance is about 40% lower. It's as if it's not utilizing the second GPU in SLI. So in order to get "accurate" results, I have to start up a benchmark and immediately ESC out of it, then start it again.

Does anyone else notice this? Or is it just me?

I have tried to run a benchmark with same settings several times in a row, sometimes there was a very minor difference if any. Also quitting the benchmark immediately and restarting it did not affect my results overall, sorry, I have no idea what's going on there for you.


max quality
21 / 36 / 27.9 - patch 2, 349.90, Windows 10 CTP b10040
21 / 34 / 26.5 - patch 1, 349.90, Windows 10 CTP b9926
21 / 34 / 26.2 - RTM, 347.52 , Windows 10 CTP b9926

quality
36 / 59 / 44.9 - patch 2, 349.90, Windows 10 CTP b10040
36 / 61 / 44.8 - patch 1, 349.90, Windows 10 CTP b9926
36 / 61 / 44.9 - RTM, 347.52, Windows 10 CTP b9926

performance
53 / 73 / 64.0- patch 2, 349.90, Windows 10 CTP b10040
53 / 73 / 63.8 - patch 1, 349.90, Windows 10 CTP b9926
53 / 73 / 63.5 - RTM, 347.52, Windows 10 CTP b9926

max performance
63 /117 / 99.2 - patch 2, 349.90, Windows 10 CTP b10040
63 /117 / 98.6 - patch 1, 349.90, Windows 10 CTP b9926
63 /117 / 99.1 - RTM, 347.52, Windows 10 CTP b9926


resumee of Attila Patch 2 on my computer:

slight increase with patch 2 / Attila 1.2 on max quality settings. max performance returned to the RTM, initial performance. Everything else remained quite unchanged (measure tolerance)

[SPARTAN VI]

I have tried to run a benchmark with same settings several times in a row, sometimes there was a very minor difference if any. Also quitting the benchmark immediately and restarting it did not affect my results overall, sorry, I have no idea what's going on there for you.


max quality
21 / 36 / 27.9 - patch 2, 349.90, Windows 10 CTP b10040
21 / 34 / 26.5 - patch 1, 349.90, Windows 10 CTP b9926
21 / 34 / 26.2 - RTM, 347.52 , Windows 10 CTP b9926

quality
36 / 59 / 44.9 - patch 2, 349.90, Windows 10 CTP b10040
36 / 61 / 44.8 - patch 1, 349.90, Windows 10 CTP b9926
36 / 61 / 44.9 - RTM, 347.52, Windows 10 CTP b9926

performance
53 / 73 / 64.0- patch 2, 349.90, Windows 10 CTP b10040
53 / 73 / 63.8 - patch 1, 349.90, Windows 10 CTP b9926
53 / 73 / 63.5 - RTM, 347.52, Windows 10 CTP b9926

max performance
63 /117 / 99.2 - patch 2, 349.90, Windows 10 CTP b10040
63 /117 / 98.6 - patch 1, 349.90, Windows 10 CTP b9926
63 /117 / 99.1 - RTM, 347.52, Windows 10 CTP b9926


resumee of Attila Patch 2 on my computer:

slight increase with patch 2 / Attila 1.2 on max quality settings. max performance returned to the RTM, initial performance. Everything else remained quite unchanged (measure tolerance)

Yeah it's bizarre. Basically I have to run a benchmark, ESC out, then every test I run thereafter is fine. Almost positive I'm not running power saving features.

[jjonestemp]

I think that's the height of affordable air technology... I'm running overclocked everything with ram at 2400 and cpu i 7 4820k at 4300 Ghz-3 nvidia cards (2x560Ti 2 Gb , and a 260 dedicated to PhysX) ... Thanks to my asus black rampage 4 everything is under 39° C, but the best I can pull 3 screen 3-d all high is 18 fps, with 34 the max... 3 purpose built total war rigs later, nothing is incredibly improved except the Europa-Barborarum direction, and optimized npc turn times (32 gb of ram might help, there)... I'll always play Total War, no matter how far they fall after the Avatar system- Shogun clans MP was the best game ever made: let's hope for a return to the clan/avatar system ASAP

[alQamar]

I think that's the height of affordable air technology... I'm running overclocked everything with ram at 2400 and cpu i 7 4820k at 4300 Ghz-3 nvidia cards (2x560Ti 2 Gb , and a 260 dedicated to PhysX) ... Thanks to my asus black rampage 4 everything is under 39° C, but the best I can pull 3 screen 3-d all high is 18 fps, with 34 the max... 3 purpose built total war rigs later, nothing is incredibly improved except the Europa-Barborarum direction, and optimized npc turn times (32 gb of ram might help, there)... I'll always play Total War, no matter how far they fall after the Avatar system- Shogun clans MP was the best game ever made: let's hope for a return to the clan/avatar system ASAP

And how does a low temperature help you to gain FPS? (I am running on a partly water-cooling too). I don't think we can blame the technology when a game doesn't run as intended on max setting and we are forced to overclock things, which does help to an extend only.

[A Barbarian]

@jjonestemp
I am not sure what your reasoning is because

(2x560Ti 2 Gb , and a 260 dedicated to PhysX

GTX560TIs are low end cards now. And Attila had problemsm with SLI until recently. Have u installed the latest drivers and checked that both cards are utilized?

and a 260 dedicated to PhysX

PhysX is not utilized in TW games

Thanks to my asus black rampage 4 everything is under 39° C

What does your motherboard have to do with the temperature of your GPUs and CPU ? I don't quite understand.

32 gb of ram might help, there

TW games are 32 bit and therefore limited to about 3GB RAM (but they actually only use about 2GB max). Therefore they don't profit from 32GB, not even from 16GB. 8GB RAM is easily enough. Even 4GB RAM can be enough if essentially nothing else is running in Windows

but the best I can pull 3 screen 3-d all high is 18 fps, with 34 the max

Why should that be a surprise with the GPUs?

[SPARTAN VI]

@jjonestemp
I am not sure what your reasoning is because


GTX560TIs are low end cards now. And Attila had problemsm with SLI until recently. Have u installed the latest drivers and checked that both cards are utilized?


PhysX is not utilized in TW games


What does your motherboard have to do with the temperature of your GPUs and CPU ? I don't quite understand.


TW games are 32 bit and therefore limited to about 3GB RAM (but they actually only use about 2GB max). Therefore they don't profit from 32GB, not even from 16GB. 8GB RAM is easily enough. Even 4GB RAM can be enough if essentially nothing else is running in Windows


Why should that be a surprise with the GPUs?

Yeah he was kinda all over the place. I couldn't tell if he's happy or upset with the performance, much less what was his point.

What does your motherboard have to do with the temperature of your GPUs and CPU ? I don't quite understand.

Some southbridge/northbridges can run relatively hot. My guess is that would increase ambient case temperature, therefore increasing temperature of everything else. I'd bet cable management and proper use of static pressure and airflow fans would be much more important than that though.

[revan.be]

Nvidia
Extreme
screen range 15-5
avg fps 8.7
eyeballed range 6-12
http://steamcommunity.com/sharedfile.../?id=417585310

Max quality
screen range 21-11
avg fps 14.0
eyeballed range 11-15
http://steamcommunity.com/sharedfile.../?id=417589658

Quality
screen range 33-18
avg fps 24.0
eyeballed range 18-28
http://steamcommunity.com/sharedfile.../?id=417585373

Performance
screen range 41-23
avg fps 33.6
eyeballed range 27-39
http://steamcommunity.com/sharedfile.../?id=417589714

Max Performance
screen range 61-31
avg fps 49.0
eyeballed range 31-61
http://steamcommunity.com/sharedfile.../?id=417589756

Intel Integrated Graphics 4600

http://steamcommunity.com/sharedfile.../?id=418143242
used unlimited video memory

Extreme
Screen range lots of 2´s - two 1´s
Avg FPS 2.1
Eyeballed range 2-1
http://steamcommunity.com/sharedfile.../?id=418143579

Max Quality
Screen range two 4´s - two 2´s
Avg FPS 3.5
eyeballed range 2-4
http://steamcommunity.com/sharedfile.../?id=418143987

Quality
Screen range two 6´s - 3
Avg FPS 5.1
Eyeballed range 6-3

http://steamcommunity.com/sharedfile.../?id=418144224

Performance
Screen range 9-6
Avg FPS 7.9
Eyeballed range 6 - 9
http://steamcommunity.com/sharedfile.../?id=418153797

Max Performance
Screen range 18-9
Avg FPS 14.0
Eyeballed range 9-16
http://steamcommunity.com/sharedfile.../?id=418154165


Intel(R) Core(TM) i5-4210H CPU @ 2.90GHz
Intel(R) HD Graphics 4600 2GB + MSI Nvidia GeForce 850M
The MSI Nvidia GeForce 850M has 1GB according to windows,
though attila says the MSI Nvidia GeForce 850M has 1972mb vram.
The Intel(R) HD Graphics 4600 has 1Gb according to Attila.


Eyeballed range means i looked at the top right of the screen and noted the highest and lowest number manually,
even if the highest or lowest only appeared for a split second

Screen range means wysiwyg of the results screen , the range at which the benchmark is noted,
thusly also what is the min and max fps according to attila,
which is not always the same as the eyeballed range ,
i even took photos of the results screen and posted to steam.

[alQamar]

Hi guys,
we could actually need some more advertising for this thread and more results, especially on AMD CPU / GPUs and mid-class and entry computers aswell as on laptops. Would be cool if you can help to propagate this thread and include it in your signature - thanks in advance.

[revan.be]

well , i think one of my mates 2 pcs is an AMD , and another friend might also be convinced later...

[alQamar]

Thanks man, please explain your figures, have you run the benchmark? what it is the meaning of those ranges?!

It would be very helpful if you can run the benchmarks with the preset as needed. If you have time run it with Nvidia once and another time with the Intel card (assignable in the nvidia driver). Just for Intel please enable unlimited vram.

[revan.be]

Nvidia
Extreme
screen range 15-5
avg fps 8.7
eyeballed range 6-12
http://steamcommunity.com/sharedfile.../?id=417585310

Max quality
screen range 21-11
avg fps 14.0
eyeballed range 11-15
http://steamcommunity.com/sharedfile.../?id=417589658

Quality
screen range 33-18
avg fps 24.0
eyeballed range 18-28
http://steamcommunity.com/sharedfile.../?id=417585373

Performance
screen range 41-23
avg fps 33.6
eyeballed range 27-39
http://steamcommunity.com/sharedfile.../?id=417589714

Max Performance
screen range 61-31
avg fps 49.0
eyeballed range 31-61
http://steamcommunity.com/sharedfile.../?id=417589756

Intel Integrated Graphics 4600

http://steamcommunity.com/sharedfile.../?id=418143242
used unlimited video memory

Extreme
Screen range lots of 2´s - two 1´s
Avg FPS 2.1
Eyeballed range 2-1
http://steamcommunity.com/sharedfile.../?id=418143579

Max Quality
Screen range two 4´s - two 2´s
Avg FPS 3.5
eyeballed range 2-4
http://steamcommunity.com/sharedfile.../?id=418143987

Quality
Screen range two 6´s - 3
Avg FPS 5.1
Eyeballed range 6-3

http://steamcommunity.com/sharedfile.../?id=418144224

Performance
Screen range 9-6
Avg FPS 7.9
Eyeballed range 6 - 9
http://steamcommunity.com/sharedfile.../?id=418153797

Max Performance
Screen range 18-9
Avg FPS 14.0
Eyeballed range 9-16
http://steamcommunity.com/sharedfile.../?id=418154165


Intel(R) Core(TM) i5-4210H CPU @ 2.90GHz
Intel(R) HD Graphics 4600 2GB + MSI Nvidia GeForce 850M
The MSI Nvidia GeForce 850M has 1GB according to windows,
though attila says the MSI Nvidia GeForce 850M has 1972mb vram.
The Intel(R) HD Graphics 4600 has 1Gb according to Attila.


Eyeballed range means i looked at the top right of the screen and noted the highest and lowest number manually,
even if the highest or lowest only appeared for a split second

Screen range means wysiwyg of the results screen , the range at which the benchmark is noted,
thusly also what is the min and max fps according to attila,
which is not always the same as the eyeballed range ,
i even took photos of the results screen and posted to steam.