[Sweclockers]Total War: Atilla benchmarks.

[Mondozei]

This game has five general performance settings.
The lowest is called "max performance", of course. The highest is called "extreme quality".

Sweclockers tried the middle setting, called just "quality" without AA.

1080p

1440p

Rendering times for AMD 290X

Rendering times for GTX 980

Sweclockers also tried the "extreme quality" setting. The GTX 980 got 36 fps average in lowly 1080p, no AA. The 970 got 30, the 290X got only 22. All these numbers are average. Because the results were so atrocious they didn't bother to test for the other cards. You can see the table in the story if you want to.

They used an overclocked i5 Haswell.

They then looked at CPU rendering times constrasted to the GPU:

What they basically found was that the GPUs didn't get to work that much, but neither did the CPU, meaning that there wasn't any real CPU bottleneck which in turn strongly implies very poor optimisation(the same old disease that fell Rome 2).

I think that CA's excuses are now finished, we just have to conclude that they have unskilled programmers who are bad at what they do.

 

[Head1985]

http://pclab.pl/art61577.html

 

[Cerb]

Their CPU graph is not nearly enough info to determine CPU usage efficacy. Windows likes to move threads around all the time, so that graph could just mean they don't use as many threads as the hardware has logical CPUs. You really need to compare GPU time and CPU time of the threads, to make a reasonably good determination.

Nonetheless, the results not good, and I would avoid it from the spikes in frame times alone, because that would drive me up the wall, being so common and huge.

 

[RussianSensation]

The game isn't out until Feb 17th, which means neither AMD nor NV have optimized drivers for it. I also expect patches to follow as is almost always the case with modern games.

 

[wand3r3r]

Nonetheless, the results not good, and I would avoid it from the spikes in frame times alone, because that would drive me up the wall, being so common and huge.

It also demonstrates how FPS are worthless beyond an initial indication of performance. The two best cards from both sides have horrible frametimes so the gameplay would be potentially a stutter fest.

 

[RussianSensation]

It also demonstrates how FPS are worthless beyond an initial indication of performance. The two best cards from both sides have horrible frametimes so the gameplay would be potentially a stutter fest.

Yup, both the 980 and 290x are spiking to 40-60ms and quite often. Awful performance --> Likely horrible optimization.

 

[Annisman*]

CA screwed me for the last time with Rome 2. I will buy Atilla when it is 5$ in the bargain bin.

 

[dangerman1337]

Why is 1080p and 2560x1440 only minor performance drop?

 

[RussianSensation]

This game is a GPU/CPU killer.

 

[parvadomus]

This game is a GPU/CPU killer.

This game is a CPU killer. Perfect game for DX12.

 

[StinkyPinky]

Gee, you guys do know this is a total war game right? Those games are so buggy that I wouldn't read anything into performance until at least the first major patch.

 

[Silverforce11]

CA screwed me for the last time with Rome 2. I will buy Atilla when it is 5$ in the bargain bin.

Heard lots of bad things about Rome 2 so I didn't get it, shame, because I liked Shogun 2 a lot and that game ran very well on all hardware.

 

[lehtv]

Why is 1080p and 2560x1440 only minor performance drop?

CPU bottleneck.

 

[StinkyPinky]

Heard lots of bad things about Rome 2 so I didn't get it, shame, because I liked Shogun 2 a lot and that game ran very well on all hardware.

Rome 2 was my biggest gaming disappointment in years. Nothing held up to the promise. Graphics were downgraded, the gameplay on the map was worse than the original, the battle AI so badly broken you could defeat armies 10x your size, and the sieges were 100% win rate because the AI just charged down one path and then lost morale. Also the AI on the map was so passive I never once got attacked. Not once.

 

[lehtv]

Rome 2 was my biggest gaming disappointment in years. Nothing held up to the promise. Graphics were downgraded, the gameplay on the map was worse than the original, the battle AI so badly broken you could defeat armies 10x your size, and the sieges were 100% win rate because the AI just charged down one path and then lost morale. Also the AI on the map was so passive I never once got attacked. Not once.

That sounds pretty disappointing. I'm glad I never picked it up... I had it on my Steam Wishlist for quite some time but then decided it would probably not be worth it after my lousy experience with Shogun 2

 

[boozzer]

rome 2's campaign ai got fixed. siege ai got somewhat fixed. but all the bad design decisions is still there. what I want the most from attila is the UI, wish the devs could patch the new UI into rome 2.

please don't pre order and watch user reviews on youtube. I read several of the reviews from major sites already, they are all @#$%.

 

[DarkKnightDude]

What is up with the VRAM increase?

Rome 2 was like only 1.5 GBs for 1080p, Atilla doesn't look amazingly better for nearly 3GB.

 

[dangerman1337]

What is up with the VRAM increase?

Rome 2 was like only 1.5 GBs for 1080p, Atilla doesn't look amazingly better for nearly 3GB.

I think that's because Total War Attila has MSAA on which most benchmarks just carelessly leave on these days (I'm looking at you TechpowerUp) and give the impression that it is super intensive :\.

 

[cen1]

The game will be broken from start and it will take a year to be fully patched, just like Rome 2. The first and the last time I made a pre-order, never again. And I love CA and their older games, I must have spent hundreds of hours in TW series..

 

[DarkKnightDude]

I think that's because Total War Attila has MSAA on which most benchmarks just carelessly leave on these days (I'm looking at you TechpowerUp) and give the impression that it is super intensive :\.

The game doesn't even have MSAA from what I heard. It only uses that crap MLAA.

 

[dangerman1337]

Here's an offical note from CA about performance: http://forums.totalwar.com/showthread.php/146776-An-Official-note-on-ATTILA-amp-Performance

Hi guys!

There’s a few threads popping up with people worried about their kit running the game, we just wanted to clear up a couple of misconceptions that might help you out.

1. The “Recommended Spec” is not the spec that is required to run the game on full max quality settings at 1440p. It is a spec which we believe gives you a good experience of the game as we’d recommend it is played. If you have better hardware you will get better results.

There are 5 quality presents you can select in Total War: ATTILA (you can always go into ‘Advanced’ and tweak yourself of course).

• ‘Max Performance’ – integrated chipsets and older cards
• ‘Performance’ – low end cards
• ‘Quality’– mid range cards
• ‘Max Quality’ – high end cards
• ‘Extreme Quality’ – future cards – no card is set to this out of the box

The official “Recommended Spec” is pegged to the ‘Quality’ setting at 1080p ‘out of the box’.

‘Extreme Quality’ also activates 4XMSAA, which will have a particularly big performance impact. Please note, that the ‘Extreme Quality’ setting is meant for future graphics cards, not for current gen. This is why it's above Maximum Quality.

2. The Min and Rec specs are higher than ROME II, it is a new game that is going on two years more recent and is based on a newer iteration of our TW Engine. As a result it is more demanding, if you ran ROME II at a certain combination of settings, ATTILA will be different. If you have high end kit though, ATTILA will certainly look more impressive.

Note also that the ‘Extreme’ setting is not the same in each title.

3. There has been work done on performance and optimisation since the review build. How much that might effect your particular spec though we wouldn’t be able to say.

4. The way the game handles texture resolution has changed. In ROME II it was tied to the quality setting. In ATTILA it is based on available VRAM.

Finally, make sure you update all your drivers to the latest versions before you play on Tuesday, that is likely to make a huge difference.

Seems "Extreme" means future cards (GM200 and R300? or afterwards?).

 

[Headfoot]

Heard lots of bad things about Rome 2 so I didn't get it, shame, because I liked Shogun 2 a lot and that game ran very well on all hardware.

Rome II was a train wreck at release. It's 100x better now, very fun. Not quite as compelling as Shogun 2 but pretty damn close

 

[Annisman*]

Rome II was a train wreck at release. It's 100x better now, very fun. Not quite as compelling as Shogun 2 but pretty damn close

it still doesn't have proper AA support...

 

[Deders]

What is up with the VRAM increase?

Rome 2 was like only 1.5 GBs for 1080p, Atilla doesn't look amazingly better for nearly 3GB.

Rome 2 will use over 2.5GB in some scenes, if you have less Vram it will automatically turn off features like AA to save memory.

 

[Red Hawk]

This game is a CPU killer. Perfect game for DX12.

Yup. The Total War series has always been a CPU killer. Massive armies rendered and animated in real time. It really does stand to benefit from Mantle/DirectX 12 CPU optimization.

I never played Rome 2, the last one I played was Shogun 2. If it's true that Rome 2 and Attila don't have MSAA, that's a bit confusing, because Shogun 2 did have MSAA. I guess they moved from a forward renderer to a deferred renderer with Rome 2? That's the usual explanation for why a developer didn't implement MSAA. MSAA with deferred rendering is possible, but it's very performance heavy, hard to optimize, and the effect is more limited.