Crysis and GTX 460 - Am I doing something wrong?

[Aethren]

Greetings all.

Just installed Crysis on my new setup, and the results were... unimpressive.

Here's the rig:

Core i5 760 OC'd to 3.1 Ghz (Stock cooler, V6 hasn't arrived yet)
Asus Maximus III Formula
4 GB Corsair Dominator GT 1600Mhz (2x2 GB)
2x ENGTX 460 DirectCU 1GB OC'd to 800 / 1600 / 2000 (core, shader, mem on Afterburner) (260.99)
Caviar Black 1TB Sata 600 7200rpm HDD
HD audio
Windows 7 Ultimate 64

Results:

First, Crysis refuses to run on 32-bit, and to even get it to run at all i had to get a no-cd crack for the 64 version.

Settings:

1920x1080
DX10
Very High settings
2xAA
16x AF


With a single GPU enabled, I get

Min: 14 fps Avg: 18 fps Max: 21 fps

In SLI I get

Min: 40 fps Avg: 44 fps Max: 63 fps




Very playable, but I hear people manage 40 fps in Very High and 4xAA on a SINGLE GTX 460... I can't help but wonder, what gives?

 

[nforce4max]

Well it scaled which means that it is working but I can't say that this isn't perfect but what is there to expect on a 1156 build knowing the pci-e bottleneck.

 

[Aethren]

Well, this is the only game quite so disappointing atm. New Vegas runs like a charm, and scales wonderfully.
The 8x lanes aren't saturated even with twin 480s, and the performance loss is around 4%, so not really a bottleneck.

Weird thing is, SLI usually scales aroung 95% on the 460s... in my case, that was a 200% increase, which is weird in and of itself.

Unigine Heaven gives me a solid 66fps on all settings max with 16xQ SLI AA mode and Tessellation. It drops to 33fps with a single card, so I'd say the scaling is pretty spot-on.

What really bugs me is the fact that it runs so borderline... even turning the AA from 2x to 8x makes me lose like 28fps, making it sluggish.

On a side note: I just saw that it uses 1022MB of V memory, which is almost exactly what I got... could it be a frame buffer size bottleneck issue?

 

[Aethren]

Also, even 4x lanes aren't saturated:


"The Bottom Line
The results are actually a bit shocking to us to be honest. We weren’t so surprised that in the previous evaluation x8/x8 did not cause any differences at 2560x1600 but did at 5760x1200. However, we thought certainly at x4/x4 PCIe 2.0 mode there would be some kind of a bottleneck at 2560x1600, but the results have proven otherwise. Even with all the data that GTX 480 SLI is pushing across the PCIe bus, x4/x4 is NOT a bottleneck in a single display setup at 2560x1600 with AA enabled. The only game to show us any difference was AvP, but it did not affect the gameplay experience. Therefore, if you are on an aging PCIe 1.X system at x8/x8 mode (equivalent to PCIe 2.0 x4/x4) on a single display fear not, you are not holding back the performance of GTX 480 SLI or we guess with any CrossFireX or SLI configuration."
http://www.hardocp.com/article/201 [...] 16_vs_x4x4



I'd say I have some headroom, don't you agree?

 

[nforce4max]

That is not 200% but half that. If it is actually that low you have a bottleneck but that aside you are doing fairly well for your build.

 

[Aethren]

Like I said:

With a single GPU enabled, I get

Min: 14 fps Avg: 18 fps Max: 21 fps

In SLI I get

Min: 40 fps Avg: 44 fps Max: 63 fps

63 is exactly 3x 21... hence, if 21 = 100%, 63 = 300%, so the increase is 200%

That's what I meant.

The Unigine test scaled around 96%, which is expected.

My complaint is that, using a single 460 card, some people claim to get 40 fps on 1920x1080 at Very High with 4x AA... either this is false, or my system is acting up.

PS: Some comments point to a 12 to 15 fps increase when running in 32-bit mode, as opposed to 64-bit... unfortunately it refuses to run on 32 bit, for some reason.

 

[Aethren]

Some other info:

Fallout: New Vegas

All settings maxed
All view distances maxed
Higher Rez texture mods enabled
8x Supersample AA transparency
16xQ SLI AA
15x AF
1920x1080
HDR


Vsync Forced Off (otherwise fps capped at 60)


Min fps: 66 Avg: 92 Max: 109



How can I run New Vegas at 109, and hit a wall at 63 on Crysis?...

 

[notty22]

gtx 460 sli using Crysis benchmarking tool
benchmark gpu
Run #1- DX10 1920x1080 AA=8x, 64 bit test, Quality: VeryHigh ~~ Overall Average FPS: 54.595
Run #1- DX10 1920x1080 AA=No AA, 64 bit test, Quality: High ~~ Overall Average FPS: 73.775

benchmark cpu

Run #1- DX10 1920x1080 AA=8x, 64 bit test, Quality: VeryHigh ~~ Overall Average FPS: 54.995
Run #1- DX10 1920x1080 AA=No AA, 64 bit test, Quality: High ~~ Overall Average FPS: 80.55
Run #1- DX10 1920x1080 AA=8x, 64 bit test, Quality: High ~~ Overall Average FPS: 75.145

Big fps hit from high to very high

 

[Doom3klr]

Funny because my 9800gtx+ would run crysis fairly well on high and my Ati 5770 also runs it on high both playable but not buttery smooth. But not bad for a 3 core amd system that cost me like 5 bills to make

 

[bystander]

Your SLI performance is pretty much what I'd expect, but it is odd how low the single GPU setup was.

 

[Aethren]

@ Dipa

No stuttering at all, runs like it's on wheels. Although i do use vsync and clamp the fps at 60, otherwise I get minor tearing. And I did notice that at 32x SLI AA, whenever i stare at round objects (really round, like circular wall vents and exhausts) I get a little lag spike (drops to 49fps for like 2 seconds). Doesn't do that on 16xQ though, which is why I settled for that. Also, I can't really tell a difference in AA beyond 8x


For Crysis:

I measured the fps in actual gameplay, while shooting people at the beach looking at the ocean.

I I move inland, I get about 10 more fps, just for not having water on the background.

The game was released in 2007, so I expected that with a 2010 SLI setup I could dare hope for like 80fps at max with 16xQ AA, but I guess this little monster hasn't seen the last of its hardware-grubbing days yet...

The only people I hear using 32x AA at absolute max with 60+ fps are the ones with triple or quad 480s SLI... I mean, DAMN!

 

[Tamz_msc]

Try to update Crysis to a new version.

 

[kimyeang88]

lol the game release in 2007 and you think you will get 80 FPS? the developer who design Crysis max out the graphic when they design it. Most graphics at that time barely run at MAX setting unless you have tri SLI or quad SLI . My old 8800 GTX got like 15 to 28 FPS at medium setting(sometime even hit 5 fps) , and AA OFF 1680 x 1050 resolution.

impossible to get 80 FPS with 1080p resulution and max setting because ATI 5870 and GTX 480 can not even pass 60 FPS at 1080P and AA off( yours has AA on)

check this graphics result for all nvidia card and ati card.

ATI= http://www.tomshardware.com/reviews/radeon-hd-5970,2474-8.html
NVIDIA= http://www.tomshardware.com/reviews/geforce-gtx-480-3-way-sli-crossfire,2622-6.html

ATI 5870 1080p resolution with 4x AA got only 31 FPS. how about GTX 460? you got x16 AA. probably at 15 FPS( the fps u were getting). 5870> GTX 460. If 5870 did use x16 AA, probably hit 23 to 25 FPS.

 

[Aethren]

Like, wow.

This game's a hardware hoe!

And I guess those ppl on youtube claiming 1080p 4xAA at 60fps with a single 460 are pretty fake, then....

 

[nforce4max]

Like, wow.

This game's a hardware hoe!

And I guess those ppl on youtube claiming 1080p 4xAA at 60fps with a single 460 are pretty fake, then....

It is not all that hard to cheat compared to most rigs out there.