[Sweclocker] Anno 2225 benchmarked

[Silverforce11]

@psolord
There's reports from AMD users that forcing Tessellation on x16 nearly doubled their min fps and boosted avg fps by around 50%. No loss in image quality.

It seems like the buildings and terrain are using tessellation. Can you give it a test?

 

[tential]

@psolord
There's reports from AMD users that forcing Tessellation on x16 nearly doubled their min fps and boosted avg fps by around 50%. No loss in image quality.

It seems like the buildings and terrain are using tessellation. Can you give it a test?

So similar to the Witcher 3 amd users can boot performance with a tessellation forcing.

Why doesn't nvidia allow the same since it help their users?

Is there some kind of monopoly on having certain settings be available under driver setting? It seems wild amd and nvidia wouldn't copycat each other in that section and both be pretty much 100% equal. I can't force triple buffering or adaptive vsync in amd drivers and I can't force tessellation in nvidia.

Just seems odd they wouldn't fix deficiencies in their drivers that the other team has first.

 

[RussianSensation]

Graphics are good but going from 4xAA to 8xAA drops my performance from the high 60s/low 70s to the low 50s and 40s on Ultra High settings @ 1440p.

TPU's results are far worse. Even at 1080P with 4XMSAA, the performance is in the 40s on a 980Ti.

33 fps on last gen's $700 flagship 780Ti and only 33.9 fps on a 970. This is 1080P. If the 980TI was to be overclocked 20%, 49.6 x 1.20 = 59.5 fps or 76% faster than Fury X. That's insane.

I get that this game isn't an FPS so 29-33 fps is far more acceptable. Also, 980Ti performs exceptionally well against everything else besides the Titan X.

TPU confirms that the game has a memory leak:

"I'd like to mention one more thing: the current version of Anno 2205 suffers from a system RAM memory leak—you will see Anno use up all of your available RAM if you play for several hours on end without restarting the game, which can also cause GPU slowdowns shortly before the system runs out of memory, so make sure to restart the game from time to time."

I'd like to see the performance in 1-2 months after game patches and 2-3 versions of AMD/NV drivers.

Is this the most demanding new game of 2015?

 

[Silverforce11]

@tential
Witcher 3 devs added the tessellation slider no? They also reduce the default x64 to x32 in a patch, improving everyone's performance.

As to why not allow NV users to choose.. guess they don't feel the need to since their uarch runs x32 and x64 heaps better than AMD. If NV GPU take a 20% perf hit, its okay as long as AMD takes a 40% perf hit.

Except when AMD users force to x16 and barely take any hit like I've been doing in Witcher 3 from the start.

 

[Dygaza]

@psolord
There's reports from AMD users that forcing Tessellation on x16 nearly doubled their min fps and boosted avg fps by around 50%. No loss in image quality.

It seems like the buildings and terrain are using tessellation. Can you give it a test?

I get no real difference in fps on Fury X with default vs 16x tesselation. 35 fps on dense buildings, 80fps on water at both settings. 1080p, all max but 4x msaa.

 

[Azix]

So similar to the Witcher 3 amd users can boot performance with a tessellation forcing.

Why doesn't nvidia allow the same since it help their users?

Is there some kind of monopoly on having certain settings be available under driver setting? It seems wild amd and nvidia wouldn't copycat each other in that section and both be pretty much 100% equal. I can't force triple buffering or adaptive vsync in amd drivers and I can't force tessellation in nvidia.

Just seems odd they wouldn't fix deficiencies in their drivers that the other team has first.

There is always the possibility of them not doing it because it would be too costly on performance or too difficult to do.

 

[Techhog]

There is always the possibility of them not doing it because it would be too costly on performance or too difficult to do.

... What? I think there's a disconnect between what you read and what you think that you read. What do you think he was saying?

 

[AtenRa]

Anno 2205 is currently in the 65th place on Steam's Most Played Games.

11 days from now, nobody here will even remember this game.

 

[Erenhardt]

so we have some information on whats going on. On GCN its killing a single core while spreading it out on the other architectures. GCN is different from what the 5850 has. Scheduler was moved to the hardware iirc and nvidias GPUs are more similar to the 5850 in that regard?

But that is clearly not the case, otherwise you suggest that 4.0GHz Vishera core is almost the same as Haswell @ 4.4Ghz core and actual IPC of Vishera is higher than Haswell:

 

[sontin]

33 fps on last gen's $700 flagship 780Ti and only 33.9 fps on a 970. This is 1080P. If the 980TI was to be overclocked 20%, 49.6 x 1.20 = 59.5 fps or 76% faster than Fury X. That's insane.

That is not insane, this is PC gaming. :\

Anno 2205 is one of the best looking PC games right now. It even has a fantastic working MSAA implementation and uses all of the new fancy grahics effeks like PBS, Tessellation, new gpu based physic systems etc.

If your PC cant handle the ultra settings why dont you use the graphic options? Reduce the "Shader Quality" setting from ultra to very high and the performance will increase. On nvidia hardware up to 80%:
http://www.computerbase.de/2015-11/anno-2205-benchmarks/2/#abschnitt_benchmarks_mit_27_grafikkarten

Otherwise there wouldn't be any fun in your blame game, huh?

 

[Silverforce11]

Anno 2205 is currently in the 65th place on Steam's Most Played Games.

11 days from now, nobody here will even remember this game.

Same as Project Cars, but it was still used in benchmarks because it is tough on GPUs..

Seems to be the case, make a game that runs like crap, more likely to be used in benchmarks than a very well optimized game that runs fast.

 

[AtenRa]

Same as Project Cars, but it was still used in benchmarks because it is tough on GPUs..

Seems to be the case, make a game that runs like crap, more likely to be used in benchmarks than a very well optimized game that runs fast.

I highly doubt sites will use Anno 2205 in reviews. More interesting games(DX-12) are coming very soon.

 

[psolord]

@psolord
There's reports from AMD users that forcing Tessellation on x16 nearly doubled their min fps and boosted avg fps by around 50%. No loss in image quality.

It seems like the buildings and terrain are using tessellation. Can you give it a test?

I did a quick test and I got some mixed results on the 7950.

When the camera is top down I didn't see any difference. Framerate is around 30ish.

The I used a 45 degree angle and the framerate dropped to around 25ish while it was 19 before.

I will have to do a complete run of my benchmark to get a valid benchmark number, but today is Black Ops III day!

 

[RussianSensation]

That is not insane, this is PC gaming. :\

Anno 2205 is one of the best looking PC games right now. It even has a fantastic working MSAA implementation and uses all of the new fancy grahics effeks like PBS, Tessellation, new gpu based physic systems etc.

If your PC cant handle the ultra settings why dont you use the graphic options? Reduce the "Shader Quality" setting from ultra to very high and the performance will increase. On nvidia hardware up to 80%:
http://www.computerbase.de/2015-11/anno-2205-benchmarks/2/#abschnitt_benchmarks_mit_27_grafikkarten

Otherwise there wouldn't be any fun in your blame game, huh?

I don't think you got the point of my post or we don't see eye-to-eye on this game. I don't think this game looks spectacular enough to warrant 33-34 fps at 1080P on barely a 2-year-old $700 780Ti or a 1-year-old $330 970. 49.6 fps at 1080P on the $600 980TI is ridiculous to me. If you think this game warrants this level of perfomrance, I won't argue with you since it's just opinions and you are free to disagree with me. My point was 980Ti OC outperforming Fury X by 75%+ is insane. It's not logical to have such a delta in a brand agnostic, well optimized title. We are suddenly discussing the type of a generational/node performance difference which makes no sense in a well-optimized game, sorry.

I'll revisit this title's performance standing in 3-6 months after patches and newer drivers.

 

[Dygaza]

This game generally only have issues on temperature map type. For me, my fps is about 35 on dense buildings, and 80 on water areas, but on both arctic and lunar levels I'm hitting over 150fps on my Fury X. Visually it doesn't make any sense why one map type is so demanding to others.

 

[Erenhardt]

Reviving this thread to dispel some BS people were coming up with.

Here are my tries on benchmarking Anno 2205, for whoever may be interested.
I mean the 5850+q9550 system seems to spread the load as the 970+2500k system does. Also it has the same driver, same OS and same game version as the other AMD system. AMD you are confusing me.

In short "its a driver issue" not the big bad wolf Nvidia.

Huge AMD Driver overhead killing modern intel big core is a lie:
AMD gpu scaling with CPU: from 2,5Ghz to 4,5Ghz is close to none:

not much of CPU bottleneck here. Otherwise 4.4 Ghz (grey) would be much faster than 2,5 Ghz (green)

On the other hand, nv:

15% difference between those two CPUs.

So, to me it seems like there is something else holding amd back. Any bets?

 

[Goatsecks]

So, to me it seems like there is something else holding amd back. Any bets?

Tessellation?

 

[psolord]

Reviving this thread to dispel some BS people were coming up with.






Huge AMD Driver overhead killing modern intel big core is a lie:
AMD gpu scaling with CPU: from 2,5Ghz to 4,5Ghz is close to none:

not much of CPU bottleneck here. Otherwise 4.4 Ghz (grey) would be much faster than 2,5 Ghz (green)

On the other hand, nv:

15% difference between those two CPUs.

So, to me it seems like there is something else holding amd back. Any bets?

I never said my AMD system was presenting a bottleneck due to AMD's driver. I said that the AMD system, the one with the 7950 showed extreme usage on a single core.

Read again
http://forums.anandtech.com/showpost.php?p=37814173&postcount=95

If you had bothered to look, the gpu usage on the video of the 7950 is maxed out.



Please don't be a fanboy and if you need to be, address someone else. I only study performance on all hardware I got.

 

[gamervivek]

AMD are most probably front-end bottlenecked here, computerbase.de's cpu sclaing charts were posted here before as well.

As for cpu overhead, AMD having a hardware scheduler vs. nvidia's software scheduler means that it's more likely that nvidia have higher cpu overhead, but due to their driver being multi-threaded, it's less likely to overload the main core.

So it can be that AMD's driver uses 70% of one core, while nvidia's uses 50% of the same but also 25% of two other cores, but AMD appears more cpu bottlenecked.

 

[Erenhardt]

AMD are most probably front-end bottlenecked here, computerbase.de's cpu sclaing charts were posted here before as well.

As for cpu overhead, AMD having a hardware scheduler vs. nvidia's software scheduler means that it's more likely that nvidia have higher cpu overhead, but due to their driver being multi-threaded, it's less likely to overload the main core.

So it can be that AMD's driver uses 70% of one core, while nvidia's uses 50% of the same but also 25% of two other cores, but AMD appears more cpu bottlenecked.

Then increasing CPU frequency from 2.5 to 4.4 shoudl give a greater boost to amd than nvidia, but it is not the case.