you would think a gtx570 could max any game at 1280...

[lavaheadache]

SSAO has a big impact in Just Cause 2 and Amnesia

 

[toyota]

ran the Just Cause 2 desert sunrise bench and yeah SSAO is much more demanding than in some games but still not a killer. you can choose off, low, med, or high but I just tested with off and high. all other settings were maxed and 4x AA was used.

SSAO off
98.97 fps


image upload


SSAO high
83.99 fps


hosting images


as for Amnesia, running SSAO in and of itself is not super demanding but if you choose one of the higher samples it kills performance while looking no different. why they even offer anything above 32 makes no sense anyway.

 

[lavaheadache]

ran the Just Cause 2 desert sunrise bench and yeah SSAO is much more demanding than in some games but still not a killer. you can choose off, low, med, or high but I just tested with off and high. all other settings were maxed and 4x AA was used.

SSAO off
98.97 fps

SSAO high
83.99 fps


as for Amnesia, running SSAO in and of itself is not super demanding but if you choose one of the higher samples it kills performance while looking no different. why they even offer anything above 32 makes no sense anyway.

was that @ 1280x960? I would say that is pretty intense, 15fps difference

 

[toyota]

was that @ 1280x960? I would say that is pretty intense, 15fps difference

well 20% going from off to high SSAO is fairly huge but if you consider I am still above 80 fps too. and my context here is based on this comment which was really exaggerated.

If you don't turn off SSAO you'll pretty much kill performance on *any* card at *any* resolution. Turn it off and all of the sudden you have a fast system again.

 

[lavaheadache]

well 20% going from off to high SSAO is fairly huge but if you consider I am still above 80 fps too. and my context here is based on this comment which was really exaggerated.

I hear ya. I do want to note however that those #'s you posted are merely averages. The real picture that needs to be shown is that actual scenes where there is heavy SSAO and what the frame rate differences in said scenes are. Those numbers will surely be quite a bit different than the averages for the benchmark run.


I also dare mention that 80 fps @ 1.23 mega pixel with a near flagship GPU is unflattering. I'm not knocking you or your rig, just saying that JC2 can be very demanding


I ran the same benchmark but @ 1080p

SSAO on high and off.

desert sunrise
68.68
78.10

I'm sure you can figure out which one is which.

I will run some 2560 numbers for you if you want but I will have to do that tomorrow. My 8 month old is sleeping in the room where that monitor is.

 

[RussianSensation]

games are just dubbed automaticly as un optomized cause current gen cards don't play them well. That is what happend with crysis and is what is happening with Metro.

I'll have to disagree with you for this particular game. I think it's either a case of poorly optimized console port or a wasteful decision to use DX11 code path to achieve slightly better graphics at the cost of a huge performance hit.

Xbox 360 GPU ~ X1900XT = 7900GTX

GTX480 is 5-10x faster than 7900GTX, depending on the game (Benches: http://techreport.com/articles.x/18682/6)

So GTX570/580 should run this game at 60 FPS maxed out at all times OR the graphics should be WAY better than they are on the Xbox360, right

Yet, neither of these is true:

EDIT: Original sized PC vs. Xbox360 images can be seen here:
http://translate.google.de/translat...che-Unterschiede-festgestellt/PC/&sl=de&tl=en

Does it look better on a PC than on the Xbox360? Yes, but most of that is related to resolution/higher textures on the PC.

Given that this game struggles to run smoothly at 1280x1024 on a GTX570 given the graphics, it's one of the worst optimized games I've seen in a long time. Perhaps, it's not the developers fault though. I think the real culprit is the DX11 codepath in this game that just wastes GPU resources.

Well in all honesty, some games are also badly optimized for use with the latest hardware, or any hardware.

Exactly. Just because current hardware can do Tessellation and run DX11 code, doesn't mean this should be used if it incurs a 30-50% performance hit with minimal graphical benefit (like Tessellation in Metro 2033).

The use of Tessellation in this game is a joke. Honestly, in DX11 this game incurs a huge performance penalty over DX9/DX10 path with almost no graphical benefits. It may be true that this is a function of weak DX11 1st generation GPUs we have now, but that's not an excuse. That just highlights how useless DX11 has been in the last 2 years. Outside of Civ5 and BattleForge, DX11 has been a huge failure - bringing massive performance penalty and frankly immaterial graphical benefits. Imo, Crysis 2 is the only game where DX11 has shown its true worth.

Metro 2033 falls into the same category as Dirt 2, Alien vs. Predator, Lost Planet 2, where the performance hit of DX11 is simply not worth the trade-off in minor graphical improvements.

 

[notty22]

You can't be serious if you think Metro 2033 is NOT a poorly optimized console port with slightly better graphics and a massive DX11 performance hit??

Xbox 360 GPU ~ X1900XT = 7900GTX

GTX480 is 5-10x faster than 7900GTX, depending on the game (Benches: http://techreport.com/articles.x/18682/6)

So GTX570/580 should run this game at 60 FPS maxed out at all times OR the graphics should be WAY better than they are on the Xbox360, right

Yet, neither of these is true:

Does it look better on a PC than on the Xbox360? Yes, but most of that is related to resolution/higher textures on the PC.

Given that this game struggles to run smoothly at 1280x1024 on a GTX570 given the graphics, it's one of the worst optimized games I've seen in a long time.

The use of Tessellation in this game is a joke. Honestly, in DX11 this game incurs a huge performance penalty over DX9/DX10 path with almost no graphical benefits. It may be true that this is a function of weak DX11 1st generation GPUs we have now, but that's not an excuse. That just highlights how useless DX11 has been in the last 2 years. Outside of Civ5 and BattleForge, DX11 has been a huge failure - bringing massive performance penalty and frankly immaterial graphical benefits. Perhaps Crysis 2 is the only game where DX11 has shown its true worth.

There are many that disagree with your opinion.
http://www.pcgameshardware.com/aid,...marks-with-DirectX-11-and-GPU-Physx/Practice/

The game is based on the 4A Engine, which delivers almost any technical feature possible today. With maxed out settings the PC version looks just stunning. This is cause by the superb lighting with high resolution textures, GPU Physx, Hardware Tessellation, Morphological Anti-Aliasing as well as Direct Compute calculated Depth of Field, Screen-space Ambient Occlusion, sub surface Scattering, Parallax Occlusion Mapping and Object Motion Blur. If you want to know more about the technological base of Metro 2033 we recommend our exclusive tech interview with 4A Games’ Chief Technical Officer Oles Shishkovstov.

Metro 2033: DirectX 9 to DirectX 11 and the detail incident
If you are running Metro 2033 on a DirectX 11 graphics card you get two additional video options: Tessellation and Depth of Field. The latter one is also active in DirectX 10 but the DX11 version is rendered by Direct Compute. In Metro 2033 not only NPCs but also weapons in your hands and certain objects like chairs are tessellated. But the developers did exaggerate a little bit: Like it has been the case in Stalker: Call of Pripyat, the tessellated objects look pumped up. But without a direct comparison you can’t really see the difference anyway. Except the two additional options the DirectX 11 mode offers the same graphics as the DirectX 10 mode and even the performance is on the same level.

http://www.pcgameshardware.com/aid,706182/Exclusive-tech-interview-on-Metro-2033/News/

PCGH: Besides DX11 and GPU-Physic are there any other visual as well as technical differences between the version developed for PC and the consoles? Which technical features can't be realized with the console version?

Oles Shishkovstov: When you have more performance on the table, you can either do nothing as you say (and as most direct console ports do) or you add the features. Because our platforms got equal attention, we took the second route. Naturally most of the features are graphics related but not all. The internal PhysX tick-rate was doubled on PC resulting in more precise collision detection and joint behavior. We "render” almost twice the number of sounds (all with wave-tracing) compared to consoles. That's just a few examples, so that you can see that not only graphics gets a boost. While on the graphics side (that's not a complete list):

• most of the textures are 2048^2 (consoles use 1024^2)
• the shadow-map resolution is up to 9.43 Mpix
• the shadow filtering is much, much better
• the parallax mapping is enabled on all surfaces, some with occlusion-mapping
• we've utilized a lot of "true” volumetric stuff, which is very important in dusty environments
• from DX10 upwards we use correct "local motion blur” (sometimes called "object blur&#8221
• the light-material response is nearly "physically-correct” on higher quality presets
• the ambient occlusion is greatly improved (especially on higher quality presets)
• sub-surface scattering makes a lot of difference on human faces, hands, etc.
• the geometric detail is somewhat better - because of different LOD selection (not even counting DX11 tessellation)​

 

[tviceman]

Are you running with the cpu or gpu overclocked? Would you be interested in comparing results on games that we both own? I'm running a gtx560ti @ 975mhz on the core and 2300mhz on memory with an i7 2600k @ 4.5ghz. Not at all interested to see who's setup is faster, just curious to see what performance differences between our two setups with your gpu being faster and cpu being ever so slightly slower.

 

[toyota]

Are you running with the cpu or gpu overclocked? Would you be interested in comparing results on games that we both own? I'm running a gtx560ti @ 975mhz on the core and 2300mhz on memory with an i7 2600k @ 4.5ghz. Not at all interested to see who's setup is faster, just curious to see what performance differences between our two setups with your gpu being faster and cpu being ever so slightly slower.

I am running everything 100% stock at the moment. just let me know what you want to run and what settings so we can directly compare. 1280x960 is the highest res I can go for now though. if you want to run the Just Cause desert sunrise bench I will go back and post screens of the exact settings for it.

 

[RussianSensation]

There are many that disagree with your opinion.

I am not saying the game does not look better in DX11 with everything turned on (such as DOF, AAA). But even based on the links you provided, esp. with regard to Tessellation, there is not much difference. The AAA does provide a much sharper aliasing though.

However, like I said GTX480/580 is 5-10x (and closer to 7-10x) faster than the GPU found in the Xbox360. The graphics on the PC are not even 2x better to me.

Case and point Toyota's own comments regarding running this game in DX10:

I played Metro 2033 just fine at 1920x1080 on DX10 high settings with a gtx260. in fact I put it on very high for the last half of the game and it was not too bad. it was certainly not sluggish enough for me to switch back to high settings.

It's either the DX11 features are way too demanding for this generation of cards or the code behind them is completely unoptimized for latest GPUs or most likely the combination of the 2. Either way, the performance hit vs. the graphical benefits over DX10 is out of this world for this game in my eyes. The graphics are good, but not THAT good that you need a GTX580 SLI setup to max this game out.

This game doesn't look as good as Witcher 2 or Crysis 1/2 to me but brings performance into the 20s. Just Cause 2 is a great looking game and runs without problems on a GTX570 or comparable GPU.

It'll be interesting to see if Metro Last Light runs better. We'll know the answer when it launches

 

[WMD]

I'll have to disagree with you for this particular game. I think it's either a case of poorly optimized console port or a wasteful decision to use DX11 code path to achieve slightly better graphics at the cost of a huge performance hit.

Xbox 360 GPU ~ X1900XT = 7900GTX

GTX480 is 5-10x faster than 7900GTX, depending on the game (Benches: http://techreport.com/articles.x/18682/6)

So GTX570/580 should run this game at 60 FPS maxed out at all times OR the graphics should be WAY better than they are on the Xbox360, right

Yet, neither of these is true:

Does it look better on a PC than on the Xbox360? Yes, but most of that is related to resolution/higher textures on the PC.

Given that this game struggles to run smoothly at 1280x1024 on a GTX570 given the graphics, it's one of the worst optimized games I've seen in a long time. Perhaps, it's not the developers fault though. I think the real culprit is the DX11 codepath in this game that just wastes GPU resources.



Exactly. Just because current hardware can do Tessellation and run DX11 code, doesn't mean this should be used if it incurs a 30-50% performance hit with minimal graphical benefit (like Tessellation in Metro 2033).

The use of Tessellation in this game is a joke. Honestly, in DX11 this game incurs a huge performance penalty over DX9/DX10 path with almost no graphical benefits. It may be true that this is a function of weak DX11 1st generation GPUs we have now, but that's not an excuse. That just highlights how useless DX11 has been in the last 2 years. Outside of Civ5 and BattleForge, DX11 has been a huge failure - bringing massive performance penalty and frankly immaterial graphical benefits. Imo, Crysis 2 is the only game where DX11 has shown its true worth.

Metro 2033 falls into the same category as Dirt 2, Alien vs. Predator, Lost Planet 2, where the performance hit of DX11 is simply not worth the trade-off in minor graphical improvements.

Like i said the poor performance is mostly due to the (overly) high precision calculation of volumetric lighting and particle effect and advanced DoF. Turn DoF off and switch to high and the game performs very well with minimal loss in IQ. Just because there exists a max setting that places insane workload on current hardware doesn't mean you must use it to get good visuals.

 

[RussianSensation]

Like i said the poor performance is mostly due to the (overly) high precision calculation of volumetric lighting and particle effect and advanced DoF. Turn DoF off and switch to high and the game performs very well with minimal loss in IQ. Just because there exists a max setting that places insane workload on current hardware doesn't mean you must use it to get good visuals.

:thumbsup:

In this game, the DX11 performance hit is enormous.

http://gamegpu.ru/Action-/-FPS-/-TPS/Metro-2033-Direct-X-11-test-GPU.html

Very HQ 1280x1024 AAA/16x AF in DX10: HD5870 = 48 fps

vs.

Very HQ 1280x1024 AAA/16x AF in DX11: HD5870 = 31 fps

 

[notty22]

It's all opinion what games look better than others, and the it's just a console port accusation, debate tactic is getting old. BF3 is also coming out on consoles. There are going to be haters calling BF3 a console port.
Games like NFS Hot Pursuit are console ports, or Star Wars the Force Unleashed.
And there may be people that disagree with me on those ?

 

[tweakboy]

I max all games.. 1080p 8xAA 16xAF high quality. vsync on. single gpu performance.

All games run smooth its frame rate is stuck on 60fps 90 percent of the time,, sometimes 57fps.. This is Crysis 2 ,, so..

My cpu uses 60 percent average to run game. thx speedfan....

 

[tweakboy]

At 1280x yoru framerate will max out and never drop PERIOD.!

 

[tviceman]

I am running everything 100% stock at the moment. just let me know what you want to run and what settings so we can directly compare. 1280x960 is the highest res I can go for now though. if you want to run the Just Cause desert sunrise bench I will go back and post screens of the exact settings for it.

Stock is fine, overclock is fine, doesn't matter by me. Both would be good too! But it doesn't make any sense to compare benchmarks with your current resolution, so lets wait until you have a new screen and if you still have the time and want to do it then awesome.

 

[AtenRa]

Just for reference in AVP, my system with a single card (bench ver 1.3)

1280 x 960/Very High/High/16/SSAO Off/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 8546
Average Frame Time: 12.3ms
Average FPS: 81.5


1280 x 960/Very High/High/16/SSAO On/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 7915
Average Frame Time: 13.2ms
Average FPS: 75.5

Difference is ~8%

 

[toyota]

Just for reference in AVP, my system with a single card (bench ver 1.3)

1280 x 960/Very High/High/16/SSAO Off/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 8546
Average Frame Time: 12.3ms
Average FPS: 81.5


1280 x 960/Very High/High/16/SSAO On/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 7915
Average Frame Time: 13.2ms
Average FPS: 75.5

Difference is ~8%

I guess a stock gtx570 is measurably quicker than a 6950 even at just 1280x960. and your cpu at 4.0 would be faster than my stock 2500k too. so what is your 6950 clocked at?

 

[AtenRa]

default 810MHz , Cat 11.8

I had a lot of firefox tabs open when benched, ill re-bench again with firefox off to see if ill get more fps

 

[AtenRa]

1280 x 960/Very High/High/16/SSAO Off/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 9349
Average Frame Time: 11.2ms
Average FPS: 89.1


1280 x 960/Very High/High/16/SSAO On/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 8866
Average Frame Time: 11.8ms
Average FPS: 84.5

Difference is 5.44%

Well, it makes a bigger difference with Firefox on/off than SSAO lol

 

[Dribble]

Like i said the poor performance is mostly due to the (overly) high precision calculation of volumetric lighting and particle effect and advanced DoF. Turn DoF off and switch to high and the game performs very well with minimal loss in IQ. Just because there exists a max setting that places insane workload on current hardware doesn't mean you must use it to get good visuals.

+1

That said I play this game on high but in 3D. That looks significantly more realistic then maxed but not 3D. If you want this game to look really good you have to play it in 3D.

 

[WMD]

1280 x 960/Very High/High/16/SSAO Off/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 9349
Average Frame Time: 11.2ms
Average FPS: 89.1


1280 x 960/Very High/High/16/SSAO On/TS On/AS On/0X
Used AvP_D3D11_Benchmark.exe

Number of frames: 8866
Average Frame Time: 11.8ms
Average FPS: 84.5

Difference is 5.44%

Well, it makes a bigger difference with Firefox on/off than SSAO lol

6950 with unlocked shaders? I am getting 80.3 on the same settings and clocks.

 

[AtenRa]

No unlocks but its on 810MHz by default not 800MHz

cat 11.8

 

[WMD]

No unlocks but its on 810MHz by default not 800MHz

cat 11.8

I am running 810mhz. I guess the dual tessellator really helps on the 6950.

 

[dpodblood]

Hmm not sure why you are having such an issue in GTA IV. I have the same set up as you (2500k, GTX570) and did not run into any issues with that game. I was able to run it at max w/ 4xAA 16xAF and shadows on medium with no noticeable slowdown. As for metro I had to choose between DX11 and medium detail, or DX9 with high detail in order to get decent performance. I didn't notice any real IQ loss either way. It's one of those games like Crysis that wont be able to max out until later generations of hardware.