RikuoAmero's Gaming Rig Benchmarks

[RikuoAmero]

Hey guys,

I got to thinking today, that my gaming rig is now a year old and yet I've never posted benchmarks here on Kametsu, despite the fact I received a TON of help from people with component and build tips (seriously, you guys just don't know how awesome you are. Love you!!). I do have a few videos up on Youtube, but no charts for easy viewing. So first off: in case you're reading this at a time when I've moved on to another computer, here's the rig.

Each slot is identical

I have two of these running in SLI. (Yes I know what the screenshot says, I changed that).

My OS is stored on a Corsair Force 120GB Solid State Drive, average boot time is about a minute, from when I press power button, speed type my password and open up a web browser. I've timed it: 1 min and 1 sec.

My Games and Programs are stored on a 2TB Western Digital Caviar Green (only noticed as I was writing this up its a Green drive, how could I have been so stupid?)

I will be doing benchmarks in two stages. The second stage will be in about two or three weeks, or whenever I get a delivery of a third graphics card. I've ordered a Geforce GT 630, a single slot card which will act as a PhysX card, so that the two 560 Ti's can act as render cards. I did notice some slowdown in Batman Arkham City before, so I'm hoping the 630 will take some of the strain off of the 560's.

Here's how the benchmarks will be run.

1) Shutdown any unnecessary programs. Anything that is taking up valuable clock cycles like Folding@Home.

2) Ensure monitor is set to 1920x1080p.

3) Run each benchmark in triplicate, then post the average. So for example, if I run Arkham City at 1080p, max settings three times and get an FPS of 40, 55, and 45, I'll post an FPS score of 46.6 FPS.

Edited December 21, 2012 by RikuoAmero

[RikuoAmero]

Okay, Batman Arkham Asylum Game of the Year Edition

BENCHMARK 1: 1080p, V-Sync, No AA, All other settings Max, High CPU PhysX

Average Frame Rate After Three Runs: 23 FPS

BENCHMARK 2: 1080p, V-Sync, 16XQ AA, All other settings Max, High CPU PhysX

Average Frame Rate After Three Runs: 22.66FPS (Judging by this, having PhysX on the CPU creates a bottleneck there).

BENCHMARK 2: 1080p, V-Sync, No AA, All other settings Max, No PhysX

Average Frame Rate After Three Runs: 59FPS

BENCHMARK 3: 1080p, V-Sync, 16XQ AA, All other settings Max, No PhysX

Average Frame Rate After Three Runs: 59FPS

BENCHMARK 4: 1080p, V-Sync, 16XQ AA, All other Settings Max, High PhysX on Second GPU (GPU doing Rendering and PhysX)

Average Frame Rate after Three Runs: 59FPS

BENCHMARK 5: 1080p, V-Sync, No AA, All other Settings Max, High PhysX on Second GPU (GPU doing Rendering and PhysX)

Average Frame Rate After Three Runs: 59FPS

BENCHMARK 6: 1080p, V-Sync, 16XQ AA, All other Settings Max, High PhysX on Second GPU (which is dedicated to PhysX)

Average Frame Rate After Three Runs: 58FPS

BENCHMARK 7: 1080p, V-Sync, No AA, All other settings Max, High PhysX on Second GPU (which is dedicated to PhysX)

Average Frame Rate After Three Runs: 59FPS

CONCLUSION: One GTX 560 Ti is enough to handle Arkham Asylum, but running PhysX on the CPU cuts my FPS by two thirds, to barely playable speeds. There is no difference in frame rate whether or not I have Anti-aliasing on. I think it's because Arkham Asylum is DirectX 9, and the cards don't break a sweat running that.

Tomorrow, I'll be running Arkham City benchmarks. I fully expect to see lower FPS than for Asylum.

Edited December 21, 2012 by RikuoAmero

[Marc05]

You shouldn't turn on V-Sync when doing benchmarks. This limits the FPS to a number relative to the refresh rate of your monitor. While I'm here, I'd like to show an awesome benchmark website that any gamer should have bookmarked: http://benchmark3d.com/

Edit: Also... f*cking finally have my 15th post. Now I can finally get what I came here for however many months ago I joined. Damn google kept bringing back here lol. I hope it's worth it.

Edit2: Well, it looks like it actually took me more than a year... but I didn't want to make posts just for the sake of posts; and I hardly ever post... anywhere.

Edited December 22, 2012 by Marc05

[RikuoAmero]

You shouldn't turn on V-Sync when doing benchmarks. This limits the FPS to a number relative to the refresh rate of your monitor. While I'm here, I'd like to show an awesome benchmark website that any gamer should have bookmarked: http://benchmark3d.com/

Edit: Also... f*cking finally have my 15th post. Now I can finally get what I came here for however many months ago I joined. Damn google kept bringing back here lol. I hope it's worth it.

Edit2: Well, it looks like it actually took me more than a year... but I didn't want to make posts just for the sake of posts; and I hardly ever post... anywhere.

I tried a benchmark with V-sync off, and I still got an average of 59FPS. I'll take your advice though and leave it off for the Arkham City benchmarks.

EDIT: I've run another two benchmarks in Asylum, I did V-Sync off, 16XQ AA/No AA and no PhysX. With both cards doing pure rendering, I still got a maximum of 59FPS in both, even though there's less strain on them with no PhysX. I think Asylum is hard-coded to only go that high.

EDIT 2: I went into the Nvidia Control Panel and found out that the drivers forced V-Sync for Arkham Asylum, even if I turn it off in the game's option menu. I made sure V-Sync was turned off, ran another benchmark (No AA, No PhysX) and got a whopping score of 275. I'll re-run the benchmarks for Asylum, this time with V-Sync off every time. Thanks Marco05!

Edited December 22, 2012 by RikuoAmero

[RikuoAmero]

Okay, redid all the Arkham Asylum Benchmarks, this time with V-Sync turned off, both in game and in the Nvidia Control Panel.

BENCHMARK 1: 1080p, No AA, All other settings Max, High CPU PhysX

Average Frame Rate After Three Runs: 22FPS

BENCHMARK 2: 1080p, 16XQ AA, All other settings Max, High CPU PhysX

Average Frame Rate After Three Runs: 22FPS (Judging by this, having PhysX on the CPU creates a bottleneck there).

BENCHMARK 2: 1080p, No AA, All other settings Max, No PhysX

Average Frame Rate After Three Runs: 278FPS

BENCHMARK 3: 1080p, 16XQ AA, All other settings Max, No PhysX

Average Frame Rate After Three Runs: 197.33FPS

BENCHMARK 4: 1080p, 16XQ AA, All other Settings Max, High PhysX on Second GPU (GPU doing Rendering and PhysX)

Average Frame Rate after Three Runs: 112FPS

BENCHMARK 5: 1080p, No AA, All other Settings Max, High PhysX on Second GPU (GPU doing Rendering and PhysX)

Average Frame Rate After Three Runs: 122.33FPS

BENCHMARK 6: 1080p, 16XQ AA, All other Settings Max, High PhysX on Second GPU (which is dedicated to PhysX)

Average Frame Rate After Three Runs: 87.33FPS

BENCHMARK 7: 1080p, No AA, All other settings Max, High PhysX on Second GPU (which is dedicated to PhysX)

Average Frame Rate After Three Runs: 126.66FPS

Edited December 22, 2012 by RikuoAmero

[RikuoAmero]

Okay, tonight I ran Crysis 2 benchmarks. I used this tool http://adrenaline.uo...hmark-tool.html and ran quite a few tests. Every run was at Ultra settings, 1920x1080p, had Laplace Edge Detection (a type of Edge Anti-Aliasing, I haven't got a clue what it means) and Hi-Res Textures On. With each benchmark, there were three maps, each of which was itself run three times, to get an average FPS. A quick glance through the results tells me that the LOWEST average I got was amazingly enough 59.9FPS. Oh yeah, the second GTX 560 Ti was doing both Rendering and PhysX.

Oddly, while the benchmarks were running, the action on screen was running like someone had pressed the fast forward button on a video. Basically, if I had actually been playing, there would have been no way for me to progress, because I wouldn't have been able to react fast enough to what was being shown on screen. Odd that, its only ever happened with Crysis 2. Anyone have any ideas as to why its acting that way? Here's a video that shows quite close to what's going on (the guy in the video is using just the one 560 Ti, but still, the game is running too fast). https://www.youtube.com/watch?feature=player_detailpage&v=GGFFkkh6SQ4#t=304s

BENCHMARK 1: Ultra, 1080p, 0x AA, Times Square Map, DirectX 9

Average Frame Rate After Three Runs: 94.5FPS

BENCHMARK 2: Ultra, 1080p, 0x AA, Downtown Map, DirectX 9

Average Frame Rate After Three Runs: 89.1FPS

BENCHMARK 3: Ultra, 1080p, 0x AA, Central Park Map, DirectX 9

Average Frame Rate After Three Runs: 101.8FPS

BENCHMARK 4: Ultra, 1080p, 2x AA, Central Park Map, DirectX 9

Average Frame Rate After Three Runs: 93.4FPS

BENCHMARK 5: Ultra, 1080p, 2x AA, Times Square Map, DirectX 9

Average Frame Rate After Three Runs: 87.7FPS

BENCHMARK 6: Ultra, 1080p, 2x AA, Downtown Map, DirectX 9

Average Frame Rate After Three Runs: 83.6FPS

BENCHMARK 7: Ultra, 1080p, 4x AA, Downtown Map, DirectX 9

Average Frame Rate After Three Runs: 84.3FPS

BENCHMARK 8: Ultra, 1080p, 4x AA, Times Square Map, DirectX 9

Average Frame Rate After Three Runs: 87.8FPS

BENCHMARK 9: Ultra, 1080p, 4x AA, Central Park Map, DirectX 9

Average Frame Rate After Three Runs: 93.2FPS

BENCHMARK 10: Ultra, 1080p, 0x AA, Times Square Map, DirectX 11

Average Frame Rate After Three Runs: 78.3FPS

BENCHMARK 11: Ultra, 1080p, 0x AA, Downtown Map, DirectX 11

Average Frame Rate After Three Runs: 69.9FPS

BENCHMARK 12: Ultra, 1080p, 0x AA, Central Park Map, DirectX 11

Average Frame Rate After Three Runs: 70.5FPS

BENCHMARK 13: Ultra, 1080p, 2x AA, Central Park Map, DirectX 11

Average Frame Rate After Three Runs: 67.2FPS

BENCHMARK 14: Ultra, 1080p, 2x AA, Downtown Map, DirectX 11

Average Frame Rate After Three Runs: 66.4FPS

BENCHMARK 15: Ultra, 1080p, 2x AA, Times Square Map, DirectX 11

Average Frame Rate After Three Runs: 67.2FPS

BENCHMARK 16: Ultra, 1080p, 4x AA, Times Square Map, DirectX 11

Average Frame Rate After Three Runs: 59.9FPS

BENCHMARK 17: Ultra, 1080p, 4x AA, Downtown Map, DirectX 11

Average Frame Rate After Three Runs: 66.0FPS

BENCHMARK 18: Ultra, 1080p, 4x AA, Central Park Map, DirectX 11

Average Frame Rate After Three Runs: 67.0FPS

Edited December 23, 2012 by RikuoAmero

[dark4yoyo]

Will you be posting actual benchmarks (not games)? like ungine heaven 3 for example, so it can be compared.

Edited December 23, 2012 by dark4yoyo

[RikuoAmero]

Will you be posting actual benchmarks (not games)? like ungine heaven 3 for example, so it can be compared.

Didn't know there was a v3 of Unigine Heaven, thanks. I'm running the benchmarks now.

BENCHMARK 1: OpenGL, Extreme Tessellation, High Shaders, 16x Anisotropy, 8x AA, 1080p

Average Frame Rate After Three Runs: 39.46FPS

BENCHMARK 2: DirectX 9, High Shaders, 16x Anisotropy, 8x AA, 1080p

Average Frame Rate After Three Runs: 90.7

BENCHMARK 3: DirectX 10, High Shaders, 16x Anisotropy, 8x AA, 1080p

Average Frame Rate After Three Runs: 75.4

BENCHMARK 4: DirectX 11, High Shaders, 16x Anisotropy, 8x AA, 1080p

Average Frame Rate After Three Runs: 42.13FPS

I was initially against running Unigine because it's not really useful. Unigine isn't a game, and such, benchmarking on it says nothing on how my rig will handle Game XYZ.

[dark4yoyo]

I was initially against running Unigine because it's not really useful. Unigine isn't a game, and such, benchmarking on it says nothing on how my rig will handle Game XYZ.

Well Ungine is an easy way to compare differents rigs, I'll be posting my scores with my new HD 7950 once I get back home.

I'm waiting to see how my older CPU and newer graphic card compared against your newer cpu and older but dual graphic cards.

Edited December 27, 2012 by dark4yoyo

[RikuoAmero]

I was initially against running Unigine because it's not really useful. Unigine isn't a game, and such, benchmarking on it says nothing on how my rig will handle Game XYZ.

Well Ungine is an easy way to compare differents rigs, I'll be posting my scores with my new HD 7950 once I get back home.

I'm waiting to see how my older CPU and newer graphic card compared against your newer cpu and older but dual graphic cards.

Take a look at my DX11 score. I got 42FPS. Now take a look at the most extreme DX11 run in Crysis 2 (when I was also doing PhysX) and there I got 67FPS. Despite the fact that as a program, Unigine is much simpler and thus much less demanding (261MB install versus Crysis 2's 10GB), I got a worse score on that. Hence why its crap to use it as a baseline to compare different systems.

[dark4yoyo]

Take a look at my DX11 score. I got 42FPS. Now take a look at the most extreme DX11 run in Crysis 2 (when I was also doing PhysX) and there I got 67FPS. Despite the fact that as a program, Unigine is much simpler and thus much less demanding (261MB install versus Crysis 2's 10GB), I got a worse score on that. Hence why its crap to use it as a baseline to compare different systems.

LOL.

Is that how you compare how visually demanding a game is ?

Then a android game that's 1GB is more demanding than ungine Heaven 3.0 ?

Crysis is sevral hours long, wheras Heaven Ungine 3.0 is only a 5min loop.

You indeed have quite low scores in Heaven, which is probaly due to the 8XAA which creates a bottleneck on your 1GB vram graphic cards.

This is heaven is usually not tested at highest settings, because of memory bottleneck issues on cards with less than 2GB vram.

BTW, Crysis 2 is not compatible with PhysX, it doesn't change anything.

Edited December 27, 2012 by dark4yoyo

[RikuoAmero]

Take a look at my DX11 score. I got 42FPS. Now take a look at the most extreme DX11 run in Crysis 2 (when I was also doing PhysX) and there I got 67FPS. Despite the fact that as a program, Unigine is much simpler and thus much less demanding (261MB install versus Crysis 2's 10GB), I got a worse score on that. Hence why its crap to use it as a baseline to compare different systems.

LOL.

Is that how you compare how visually demanding a game is ?

Then a android game that's 1GB is more demanding than ungine Heaven 3.0 ?

Crysis is sevral hours long, wheras Heaven Ungine 3.0 is only a 5min loop.

You indeed have quite low scores in Heaven, which is probaly due to the 8XAA which creates a bottleneck on your 1GB vram graphic cards.

This is heaven is usually not tested at highest settings, because of memory bottleneck issues on cards with less than 2GB vram.

Well for one thing, the size of the programs means that there is much less textures in Unigine versus Crysis 2, or less detailed textures. That's only the first thing. Crysis 2 has PhysX, which is a very intensive process for a Geforce GPU. Unigine doesn't. Since Unigine isn't running PhysX, that's means a lighter workload for the GPUs, which should have translated to somewhat of a better score. Not only that, but there's other things happening in the Crysis 2 Benchmark, like enemy and NPC AI. Unigine doesn't have that.

Long story short, Crysis 2 is a vastly more complex program, that demands more of the hardware, and thus, should have gotten lower frame-rates than Unigine. However, since that didn't happen...

[dark4yoyo]

Okay another approach then.

First off, Crysis 2 does NOT use PhysX.

Secondly, NPC AI whatever you call that is generated using the CPU's power.

Ungine Heaven 3.0 generates a much smaller world than Crysis 2, thus the difference in size.

Ungine Heaven 3.0 was made to stress the LATEST GPU's, particularly on tesslation, at which gtx 560 ti, are pretty bad.

And as I said before Heaven Ungine 3.0 at max settings uses over 1GB vram, thus there is a bottleneck on your 1GB 560 ti.

[RikuoAmero]

Okay another approach then.

First off, Crysis 2 does NOT use PhysX.

Secondly, NPC AI whatever you call that is generated using the CPU's power.

Ungine Heaven 3.0 generates a much smaller world than Crysis 2, thus the difference in size.

Ungine Heaven 3.0 was made to stress the LATEST GPU's, particularly on tesslation, at which gtx 560 ti, are pretty bad.

And as I said before Heaven Ungine 3.0 at max settings uses over 1GB vram, thus there is a bottleneck on your 1GB 560 ti.

Huh...I actually had to look it up and you're right there, Crysis 2 doesn't have PhysX support...which is odd considering the first game and Warhead did.

[dark4yoyo]

Huh...I actually had to look it up and you're right there, Crysis 2 doesn't have PhysX support...which is odd considering the first game and Warhead did.

They didnt support PhysX either, very few games support PhysX in fact, because it is only used by Nvidia GPU's, and because it's not really worth it.

[ZeroPenguins]

^And there has been a proven fact that AMD cards can run PhysX as well, also sometimes even better than nVidia cards.

[Marc05]

LOL.

Is that how you compare how visually demanding a game is ?

Then a android game that's 1GB is more demanding than ungine Heaven 3.0 ?

Crysis is sevral hours long, wheras Heaven Ungine 3.0 is only a 5min loop.

You indeed have quite low scores in Heaven, which is probaly due to the 8XAA which creates a bottleneck on your 1GB vram graphic cards.

This is heaven is usually not tested at highest settings, because of memory bottleneck issues on cards with less than 2GB vram.

BTW, Crysis 2 is not compatible with PhysX, it doesn't change anything.

I read this in an asshole-ish voice which is what it came off as to me. Maybe that wasn't your intent though. If it was, being a bit nicer would be good for everyone.

As for PhysX on AMD cards... that requires unofficial patches and such, so I wouldn't say it's supported (and probably never will be since they are competitors). And although PhysX is certainly not the main reason to choose a card over another, it's certainly a good feature and selling point.

Edited December 29, 2012 by Marc05

[dark4yoyo]

With my single HD 7950 @ 1240/1680 :

DirectX 11, High Shaders, 16x Anisotropy, 8x AA, 1080p: 52.6 fps.

The reason you get lower scores is probably because of your 1GB memory, I get ~1300MB memory usage with all settings max.

Try using the settings in this image: http://img38.imageshack.us/img38/8783/heaven2012121503092424.jpg with those settings I get ~900MB memory usage.

I got 81 fps with my HD 7950 @ 1200/1670 (max temp 63°), at stock HD 7950 clocks (800/1250), I get 56.4 fps (max temp 50°).

Edited December 31, 2012 by dark4yoyo

[RikuoAmero]

Okay guys, I'm gonna do Metro 2033 and Batman Arkham City all in one go tonight. Then I'll do them again tomorrow, because I'm picking up a Geforce GT 630, which will act as a PhysX card, leaving my GTX 560 Ti's to concentrate on rendering. Hopefully, I'll see at least some improvement (and in case anyone's curious, if I do get a decent FPS boost, I'll only ever leave the 630 in whenever I'm playing a PhysX game. No point in leaving it in, sucking up electricity otherwise).

METRO 2033

BENCHMARK 1: DirectX 9, 1080p, Very High Quality, PhysX On, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 61.94FPS

BENCHMARK 2: DirectX 9, 1080p, Very High Quality, PhysX Off, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 63.33FPS

BENCHMARK 3: DirectX 9, 1080p, Very High Quality, PhysX On, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 60.67FPS

BENCHMARK 4: DirectX 9, 1080p, Very High Quality, PhysX off, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 63.67FPS

BENCHMARK 5: DirectX 10, 1080p, Very High Quality, PhysX On, AAA, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 68FPS

BENCHMARK 6: DirectX 10, 1080p, Very High Quality, PhysX Off, AAA, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 71.67FPS

BENCHMARK 7: DirectX 10, 1080p, Very High Quality, PhysX On, AAA, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 67FPS

BENCHMARK 8: DirectX 10, 1080p, Very High Quality, PhysX Off, AAA, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 70.67FPS

BENCHMARK 9: DirectX 10, 1080p, Very High Quality, PhysX On, 4x MSAA, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 53.33FPS

BENCHMARK 10: DirectX 10, 1080p, Very High Quality, PhysX Off, 4x MSAA, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 54FPS

BENCHMARK 11: DirectX 10, 1080p, Very High Quality, PhysX On, 4x MSAA, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 51.37FPS

BENCHMARK 12: DirectX 10, 1080p, Very High Quality, PhysX Off, 4x MSAA, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 55.33FPS

BENCHMARK 13: DirectX 11, DOF On, 1080p, Very High Quality, AAA, PhysX On, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 43.67FPS

BENCHMARK 14: DirectX 11, DOF On, 1080p, Very High Quality, AAA, PhysX Off, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 46.67FPS

BENCHMARK 15: DirectX 11, DOF On, 1080p, Very High Quality, AAA, PhysX On, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 43.67FPS

BENCHMARK 16: DirectX 11, DOF On, 1080p, Very High Quality, AAA, PhysX Off, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 46.67FPS

BENCHMARK 17: DirectX 11, DOF On, 1080p, Very High Quality, 4x MSAA, PhysX On, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 29.33FPS

BENCHMARK 18: DirectX 11, DOF On, 1080p, Very High Quality, 4x MSAA, PhysX Off, 4x Anisotropic Filtering

Average Frame Rate After Three Runs: 31.67FPS

BENCHMARK 19: DirectX 11, DOF On, 1080p, Very High Quality, 4x MSAA, PhysX On, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 27.67FPS

BENCHMARK 20: DirectX 11, DOF On, 1080p, Very High Quality, 4x MSAA, PhysX Off, 16x Anisotropic Filtering

Average Frame Rate After Three Runs: 32.67FPS

Conclusion: PhysX doesn't alter the frame-rate all that much. The most difference is 5 frames. Well here's to hoping the 630 can give back those frames and then a few more.

[RikuoAmero]

Okay I got the 630 in and did a few runs on Metro 2033 and Arkham City. The only real improvement was when I repeated Metro Benchmark 19, and saw my FPS rise from 27 to 38. However, that came at a cost of louder cards and temperatures for the 560 Ti's in the 95+ C range. Way too hot for my tastes. So I took out the 630 (had to anyway, if I wanted to leave it in, I had to lose my front panel USB 2.0 ports). I'll stick with just the 560's, until I build a new computer.