• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Crysis

L

llmercll

[H]ard|Gawd
Joined
Feb 17, 2010
Messages
1,369
Hello everyone,

I recently upgraded to an i5 2500k and "sidegraded" my 5850 to a gtx 560 (gave the 5850 to my brother who needed a good card). Almost all my games are playing smoother with the exception of one, crysis.

Any ideas why that is? Does crysis just run better on ati cards? googling just brings up crysis 2, which seems to run better on nvidia cards. Maybe I'm missing a patch or driver? This is the original crysis btw, not warhead.

the 560gtx is OC'd to 940/1880 and if I'm not mistaken should be approaching ti performance.

Any input would be appreciated, thanks!
 
The answer is that Crysis is still a challenge even after all of these years and your graphics card is inadequate. In fact, my GTX590 is sort of inadequate.

However there is one other possibility and this is not widely known. If you are using a certain type of display and you have VSYNC enabled, Crysis may be misreading refresh rate data from the screen and capping your frame rate at 24fps. The solution is turn off VSYNC, period. This bug exists in Crysis, Warhead, and Crysis 2. It UTTERLY sucks.
 
What settings are you running it on? Also something to consider for Crysis is maybe try running it in DX9 mode.

I just recent played and about midway through the game I would get random crashes and freezes. While in DX10 mode.

I can performance was great for me, I was using ati cards 2 5870s in CF. But there were areas where the framerate dropped to like 20-25fps.

I did have that bug where it would ran at 24fps on my setup, the workaround is swap desktops to make sure the main display was designated 1 in CCC. Keep in this ATI, don't know if this affects nvidia cards the same way.
 
Any chance you're playing on a 1080p monitor? Crysis (and Warhead) love to run in 24P mode by default. That's a problem for all cards, but is more dependent on your monitor. TV's pretty much all do this, FYI.
A pseudo fix is to start the game and ALT+TAB 5-6 times quickly. Sometimes you even have to play for a second and then re-load the save.
 
Crysis is actually just THAT fuckin demanding (also very poorly optimized)...I actually returned a gtx 590 because it was dropping to 50 even 40 fps....if I can't play an almost 5 year old game on a $750.00 card then something is not right
 
you guys must have some issues. I just played it for about 15 minutes on stock 2500k and stock gtx570 and averaged around 45 fps on DX10 very settings at 1920x1080. I never went below 35 fps except on the auto save places. during actual gameplay the framerate was plenty high and this is a game that feels smooth even at 35 fps so 45-50 fps is more than enough. if I never looked at the framerate then I would assume 60 fps in this game.

even ran a couple of benchmarks and again the drops below 35 were where the game auto saves.

Frames, Time (ms), Min, Max, Avg
5888, 126158, 31, 64, 46.672

Frames, Time (ms), Min, Max, Avg
8486, 177950, 29, 65, 47.688

EDIT: now it did run like crap at those settings when I was using an E8500 and gtx470. for some reason that setup had lot of hitches and stutters in Crysis.
 
Last edited:
cannondale06 said:
you guys must have some issues. I just played it for about 15 minutes on stock 2500k and stock gtx570 and averaged around 45 fps on DX10 very settings at 1920x1080. I never went below 35 fps except on the auto save places. during actual gameplay the framerate was plenty high and this is a game that feels smooth even at 35 fps so 45-50 fps is more than enough.

even ran a couple of benchmarks and again the drops below 35 were where the game auto saves.

Frames, Time (ms), Min, Max, Avg
5888, 126158, 31, 64, 46.672

Frames, Time (ms), Min, Max, Avg
8486, 177950, 29, 65, 47.688

EDIT: now it did run like crap at those settings when I was using an E8500 and gtx470. for some reason that setup had lot of hitches and stutters in Crysis.
Click to expand...



Dropping below 60 isint acceptable to some people.
 
RaZoRbLaDeD said:
Dropping below 60 isint acceptable to some people.
Click to expand...
then those people need to forget about playing modern games. nearly every single newer game will drop below 60 fps at at times no matter what setup you have. and honestly that's a silly attitude to have because unless you stare at the framerate counter you will never know. framerates are only part of the equation as some games feel smooth even at low framerate while others can feel sluggish even over 60 fps.
 
I am not seeing any stutter so far on my current setup. it was quite bad with E8500 and gtx470 though.
 
ballpoint said:
Never let Crysis determine your PC rig's worthiness. Crysis is very poorly optimized.
Click to expand...

I'm sure you are an authoritative source. You've obviously seen the code and are a game dev who would know what optomized code is.
 
Skakruk said:
It's only in a select few areas. In fact I'm pretty certain you could play the whole game through without ever experiencing it because the maps are so large and there are so many different ways through them. Here is a vid of the stutter on a GTX285: http://www.youtube.com/watch?v=q9s3jhqY6pY

It's still there with my GTX580. It's obviously a driver bug as it's completely absent on ATI hardware.

You had a GTX 470 paired with Wolfdale? Must have been one hella OC on that chip...
Click to expand...
I have spent many many hours playing every inch of that game and never had stutter anything like that in Crysis on any setup I have used. there was a slight hitch at times when using the gtx470 with the E8500 which was annoying but nothing like that video.

yeah the E8500 was 3.8. I got the gtx470 with intentions to upgrade the cpu to an i5 750 at that time. the gtx470 was so noisy and hot that I sent it back and waited for the gtx570 and 2500k to come out.
 
I'm playing on Very High settings with 4x AA @ 1080p. It's not so much that the game is playing "badly", just that I noticed a 5fps drop going from my 5850 to GTX560, which, if I'm not mistaken, should have been a minor upgrade (and has been so in every other game so far). I was getting 30-35fps, now I'm getting 28-33. When frames are that low each one counts (at least to me) =/

@Blackstone - I am not experiencing the 24p issue, thankfully. The 590 is very powerful, I guess that goes to show crysis really just isn't well optimized. I did run 2x 5850's for awhile and it ran wonderfully, 50fps+ at the same settings.

@ballpoint - Thanks, that's kind of the answer I was looking for, but wasn't able to sum it as nicely as you did =p

thanks for the replies!
 
llmercll said:
I'm playing on Very High settings with 4x AA @ 1080p. It's not so much that the game is playing "badly", just that I noticed a 5fps drop going from my 5850 to GTX560, which, if I'm not mistaken, should have been a minor upgrade (and has been so in every other game so far). I was getting 30-35fps, now I'm getting 28-33. When frames are that low each one counts (at least to me) =/

@Blackstone - I am not experiencing the 24p issue, thankfully. The 590 is very powerful, I guess that goes to show crysis really just isn't well optimized. I did run 2x 5850's for awhile and it ran wonderfully, 50fps+ at the same settings.

@ballpoint - Thanks, that's kind of the answer I was looking for, but wasn't able to sum it as nicely as you did =p

thanks for the replies!
Click to expand...


the crytek engine is about the most unoptimized engine you can possibly find. they tried their best to hide it by making the engine playable at 30fps but in reality it sucks. it doesn't scale worth a damn when you add a much more powerful card thanks to the fact that it has piss poor threading between the crappy physics engine they used and the fact that the games limited to 2 cores. because of this it means the game is highly IPC dependent and why going from say a phenom II 955@4Ghz to a i7 920 @ stock clocks nets you about 15 fps with a gtx 560 and going from a i7 920 to an i7 2600k nets you an additional 10 fps..

if you really want to see the difference between the stock crysis and whats possible even with its limitations, run CCC v2.21 http://www.incrysis.com/forums/viewtopic.php?id=23117
just moving the physics engine thread to its own core makes a world of difference in performance.


danman said:
I'm sure you are an authoritative source. You've obviously seen the code and are a game dev who would know what optomized code is.
Click to expand...

it doesn't take a rocket scientist to know the engine code isn't optimized, anyone thats played the game can see it isn't. crytek cut corners left and right because they didn't want to change the code at the last minute to support anything over 2 cores because quad cores were just coming out at that time. it was just piss poor management in future proofing the engine. but what was even worse is that they didn't even bother to fix it with the crysis wars either.


llmercll said:
Hello everyone,

I recently upgraded to an i5 2500k and "sidegraded" my 5850 to a gtx 560 (gave the 5850 to my brother who needed a good card). Almost all my games are playing smoother with the exception of one, crysis.

Any ideas why that is? Does crysis just run better on ati cards? googling just brings up crysis 2, which seems to run better on nvidia cards. Maybe I'm missing a patch or driver? This is the original crysis btw, not warhead.

the 560gtx is OC'd to 940/1880 and if I'm not mistaken should be approaching ti performance.

Any input would be appreciated, thanks!
Click to expand...


try running CCC(link is above) if anything it should solve the performance problems. there are a lot of issues with the engine on both nvidia and AMD side because both companies gave up trying to fix them after crytek decided they would rather release crysis warhead instead of fixing the first game. so i'm sure most of the fixes Nvidia did with the 8/9/200 series cards was never ported to the 400/500 series cards.

i get the annoying stuttering problem as well but i also play MWLL instead of the actual crysis game which MWLL pretty much pushes the crytek engine to its limits.
 
Last edited:
LuminaryJanitor said:
I remember it was night and day when it released, but last time I took it for a spin (on my new GTX580), DX10 actually came out ~5% in front.
Click to expand...
in Crysis you cant even run very high settings in DX9 though so high is the max you could compare.

EDIT: I just compared high DX9 to high DX10 and those do run about the same with DX10 even being a frame or two higher in some of the same spots. that is certainly a change from the past.

EDIT 2: btw the game started flickering and freaking out after a few minutes in DX9 earlier tonight. temps were fine and its not my card as I fired up Crysis 2 in DX11 which is way more demanding and there are no issues there at all. maybe just a driver issue...
 
Last edited:
Thanks a lot for that info sirmonkey, I'm running a solid 40fps+ now, and think it looks better =)

CCC is a must have!
 
cannondale06 said:
in Crysis you cant even run very high settings in DX9 though so high is the max you could compare.
Click to expand...
There are a bunch of files which define the quality presets. Copy-paste the "Very High" section into the "High" section and you've got all the supposedly DX10-only effects in DX9.

Skakruk said:
EDIT: Actually, I can remember another BIG reason why I ditched DX10: Triple Buffering doesn't work. And D3D Overrider won't force it in DX10 Crysis neither.
Click to expand...
There's a trick to it. Disable then re-enable vsync once you're in-game and D3DOverrider should work. Bit of a pain having to do it every time, though.
 
Skakruk said:
In the Crysis Bin64 there is a 'Crysis.exe' and a 'Crysis64.exe', which one did you tie D3DOverrider to? And what detection level did you set?

Also, did you have Vsync and Triple Buffer enabled in D3DOverrider, or just the latter?
Click to expand...
Just had it running globally, triple buffering only, detection level medium.
 
Skakruk said:
I've just been screwing around on an old PC and I got it working like you said. Also, I think I inadvertently discovered another 'trick'. I Alt+Tabbed in DX10 mode once, and from then on the damned game always starts in windowed mode so I've been Alt+Entering every time. However, when starting the game this way, Triple Buffer seems to be always active without needing to turn V-Sync off and on in-game.
Click to expand...
Yeah, I think changing resolution does the same thing. Maybe changing AA modes too. Basically, anything which restarts the renderer when you apply it.
 
needmorecarnitine said:
I thought it might be that but wasn't sure. Thanks




Oh gee, only 1.9 million results. At least I learned about google.com. Never heard of that before. :D
Click to expand...
well since it was such a common issue, I knew there would be plenty of hits for it. lol, you could have had your answer quicker than it took you to ask what it was. :p

Crysis is a pretty odd game. different drivers can give a vastly different experience. and sometimes a high end card can feel worse than a slower card even though the framerate is higher. Crysis is also the only game I have ever played that Afterburner will not display info for.
 
My Crysis results. Single HD 6970 Lightning
<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

9/26/2011 9:56:23 PM - Vista 64

Run #1- DX10 1900x1200 AA=16xQ, 32 bit test, Quality: VeryHigh ~~ Overall Average FPS: 36.43

AFTER:

10/6/2011 2:58:33 PM - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX10 1900x1200, AA=16xQ, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
============================================================ ==
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 51.93s, Average FPS: 38.51
Min FPS: 30.66 at frame 156, Max FPS: 45.81 at frame 980
Average Tri/Sec: -14701829, Tri/Frame: -381750
Recorded/Played Tris ratio: -2.40
!TimeDemo Run 1 Finished.
Play Time: 48.84s, Average FPS: 40.95
Min FPS: 30.66 at frame 156, Max FPS: 46.63 at frame 876
Average Tri/Sec: -15059727, Tri/Frame: -367770
Recorded/Played Tris ratio: -2.49
!TimeDemo Run 2 Finished.
Play Time: 48.80s, Average FPS: 40.98
Min FPS: 30.66 at frame 156, Max FPS: 46.88 at frame 998
Average Tri/Sec: -15060262, Tri/Frame: -367496
Recorded/Played Tris ratio: -2.49
TimeDemo Play Ended, (3 Runs Performed)
============================================================ ==

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

10/6/2011 2:58:33 PM - Vista 64

Run #1- DX10 1900x1200 AA=16xQ, 32 bit test, Quality: VeryHigh ~~ Overall Average FPS: 40.965

Crossfire 6970 Lightnings

10/31/2011 8:03:12 PM - Vista 64
Beginning Run #1 on Map-island, Demo-benchmark_gpu
DX10 1900x1200, AA=16xQ, Vsync=Disabled, 32 bit test, FullScreen
Demo Loops=3, Time Of Day= 9
Global Game Quality: VeryHigh
============================================================ ==
TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 31.98s, Average FPS: 62.55
Min FPS: 46.98 at frame 177, Max FPS: 84.96 at frame 853
Average Tri/Sec: -21277352, Tri/Frame: -340173
Recorded/Played Tris ratio: -2.69
!TimeDemo Run 1 Finished.
Play Time: 28.46s, Average FPS: 70.26
Min FPS: 46.98 at frame 177, Max FPS: 86.56 at frame 924
Average Tri/Sec: -23081420, Tri/Frame: -328504
Recorded/Played Tris ratio: -2.79
!TimeDemo Run 2 Finished.
Play Time: 28.42s, Average FPS: 70.36
Min FPS: 46.98 at frame 177, Max FPS: 87.50 at frame 890
Average Tri/Sec: -23099502, Tri/Frame: -328286
Recorded/Played Tris ratio: -2.79
TimeDemo Play Ended, (3 Runs Performed)
============================================================ ==

<><><><><><><><><><><><><>>--SUMMARY--<<><><><><><><><><><><><><>

10/31/2011 8:03:12 PM - Vista 64

Run #1- DX10 1900x1200 AA=16xQ, 32 bit test, Quality: VeryHigh ~~ Overall Average FPS: 70.31


With 2560x1600 and same quality settings I get about 45fps on average with the lowest being 31fps. So similar to 1920x1200 with a single card but slightly faster. For comparison I get about double that with BF3, so Crysis isn't a great way to determine performance.
 
N1GHTRA1N said:
With 2560x1600 and same quality settings I get about 45fps on average with the lowest being 31fps. So similar to 1920x1200 with a single card but slightly faster. For comparison I get about double that with BF3, so Crysis isn't a great way to determine performance.
Click to expand...

It seems that Crysis in your system is CPU bottlenecked.
 
N1GHTRA1N said:
It doesn't seem that way to me. Why do you think so?

I ran the same video cards in a i5 2500k @ 4.5GHz and only got about a 5% increase in FPS.
Click to expand...

Well, I think so because your frame did not increase after upgrading graphics card.

But given that you tested the game on an i5 2500K then it seems that Crysis has some scalability problems.
 
24P is when movies are locked into 24 frames per second. In the case of Crysis, many different monitors (and almost all TV's that support the 24P format) see the game as being a 24P movie. As a result, no matter how fast your PC is, you're going to get 24 frames per second and some crazy mouse lag.

In terms of the game itself, I'm absolutely one of those people that hates to go under 60FPS, so I lower the details until I can get that experience (with vsync). To this day I tend to set Crysis up to run at a mix of medium/high/very high as a result. Don't get stuck in that mode of having to max everything out just because someone says you should. Personally, I'd rather have something running as smooth as silk and looking really good than something that looks stunning, but runs like ass or has image tearing. Lots of the people talking about maxing things out either have crazy rigs, no vsync, or framerates in the 30's.

No matter the reason for it, don't be afraid to lower the details in Crysis to get a smoother experience. It looks better than 95% of the games out there with the details set fairly low.

If your monitor keeps locking you down in 1080p/24, alt+tab repeatedly and re-load the same game a couple times. Once it works, you'll notice instantly.
 
Filiprino said:
Well, I think so because your frame did not increase after upgrading graphics card.

But given that you tested the game on an i5 2500K then it seems that Crysis has some scalability problems.
Click to expand...

My fps didn't increase? Did you look at my results? Single card 41FPS, Crossfire 71FPS? That is a pretty big increase. I even included the bottom line whcih was 36fps, that was with crossfired 4870's. So cfx 4870's to single 6970 was 5fps increase, then another 30fps when adding a second 6970. I said my results at 2560x1600 with crossfire was similar to 1920x1200 with a single card. That is also a nice result. After about 3GHz CPU with quad core and crysis the results become GPU limited. The difference between a i5 2500k at 3.3 and 4.5 is only 1-2 fps.
 
Thats odd, I saw a decent increase when moving from an i5 760 at a higher clock speed to an i5 2500k at 4.6ghz with 6950 crossfire in Crysis. About a 10% difference in average framerate and a bit more than that in minimum framerate.
 
Skakruk said:
Why do you keep mentioning V-Sync like it should have an impact on performance. It only does if you disable Triple Buffering. Enable Triple Buffering, and problem solved.
Click to expand...

Interesting! I didn't know that!
Thanks kindly for the info
 
if you really want to see the difference between the stock crysis and whats possible even with its limitations said:
http://www.incrysis.com/forums/viewtopic.php?id=23117[/url]
just moving the physics engine thread to its own core makes a world of difference in performance.
Click to expand...

I remember using that program in the past but it never seemed to help me. (Might have been using it wrong though.) However, I do not recall ever seeing the ability to move the physics processing to another core. How do you do that?
 
You must log in or register to reply here.
Back
Top