Metro 2033 bench still stutters even on the lowest settings on a gtx680

[Xenphor]

I have an msi gtx 680

All I want to do is have a Metro 2033 benchmark that is 100% smooth (regardless of graphical settings), without any stuttering. To do this, I'm capping my framerate at 60fps and using vsync through the nvidia control panel. However, even with the benchmark set to the lowest settings, I can not maintain a hard vsync'd 60fps. How is this even possible? I've tried doing the same thing with the Heaven Benchmark and was actually able to achieve a vsync'd 60fps.

With the metro 2033 benchmark I can cap the framerate at either 30fps or 60fps and still get stutters. I'm even running at a low resolution (1280x1024) which makes the performance seem even worse. The frame rate counter will show 59-60fps the whole time but the benchmark is still full of stuttering.

 

[Eureka]

What's the rest of your system? You should be able to maintain a hard 60 fps on lowest settings. Sounds like it could be a bottleneck elsewhere.

 

[blackened23]

I have an msi gtx 680

All I want to do is have a Metro 2033 benchmark that is 100% smooth (regardless of graphical settings), without any stuttering. To do this, I'm capping my framerate at 60fps and using vsync through the nvidia control panel. However, even with the benchmark set to the lowest settings, I can not maintain a hard vsync'd 60fps. How is this even possible? I've tried doing the same thing with the Heaven Benchmark and was actually able to achieve a vsync'd 60fps.

With the metro 2033 benchmark I can cap the framerate at either 30fps or 60fps and still get stutters. I'm even running at a low resolution (1280x1024) which makes the performance seem even worse. The frame rate counter will show 59-60fps the whole time but the benchmark is still full of stuttering.

What settings specifically are you using? Are you using phsyx (if so, don't, it's quite a bad implementation) , AA? etc.

Also, don't pay attention to benchmark results. It's not quite indicative of in game performance, that level used in the benchmark scene isn't how well the game plays generally speaking.

 

[Xenphor]

What's the rest of your system? You should be able to maintain a hard 60 fps on lowest settings. Sounds like it could be a bottleneck elsewhere.

i5 3570k stock, 8gb ram, asus p8z77v lk motherboard

settings:
Options: Resolution: 1280 x 1024; DirectX: DirectX 9; Quality: Low; Antialiasing: AAA; Texture filtering: AF 4X; Advanced PhysX: Disabled; Tesselation: Not supported; DOF: Not supported

I did dx9 just to see if there was a difference but no, same as dx11

I'm using the crysis 3 drivers 313.95 but have had the problem with every other version.

 

[frozentundra123456]

I have an msi gtx 680

All I want to do is have a Metro 2033 benchmark that is 100% smooth (regardless of graphical settings), without any stuttering. To do this, I'm capping my framerate at 60fps and using vsync through the nvidia control panel. However, even with the benchmark set to the lowest settings, I can not maintain a hard vsync'd 60fps. How is this even possible? I've tried doing the same thing with the Heaven Benchmark and was actually able to achieve a vsync'd 60fps.

With the metro 2033 benchmark I can cap the framerate at either 30fps or 60fps and still get stutters. I'm even running at a low resolution (1280x1024) which makes the performance seem even worse. The frame rate counter will show 59-60fps the whole time but the benchmark is still full of stuttering.

The Metro 2033 benchmark performs very strangely. I have a HD7770 and an i52320. I cant get above 60fps no matter how low I turn the resolution and graphical settings. Obviously I am gpu limited at higher settings and resolutions, but it looks like even that lowly card can be CPU limited at low resolutions. Any other game I have tested is gpu limited at any resolution. Do you have high cpu physX turned off? If not that could limit you a lot. I did not notice any stuttering to speak of, but my eye for such things is not very critical.

Edit: Strangely even at the lowest settings I got close to 60fps average, but one section still dropped way below 30 for a considerable period of time. In contrast to most benchmarks, it seems the game itself ran quite decent at high settings a 1080p.

 

[rgallant]

you need a Titan


-try msi-ab frame cap
-using it in DS3 seems to work fine.

"I'm even running at a low resolution (1280x1024) which makes the performance seem even worse"
"i5 3570k stock" - try it @ 4.6 at that res. which depends on the cpu.

 

[Xenphor]

I do have msi ab capping the framerate at 60. I'm not sure I want to overclock right now if it's a gpu problem.

This what does with everything on:
Options: Resolution: 1280 x 1024; DirectX: DirectX 11; Quality: Very High; Antialiasing: MSAA 4X; Texture filtering: AF 16X; Advanced PhysX: Enabled; Tesselation: Enabled; DOF: Enabled

Total Frames: 2409, Total Time: 59.85336 sec
Average Framerate: 40.35
Max. Framerate: 76.58 (Frame: 820)
Min. Framerate: 3.41 (Frame: 1008)

Honestly I think it was smoother than dx9 with vsync but probably because the framerate was in the 20s most of the time so stutter wasn't as noticeable (also vysnc was off).

One of the main things I notice in the benchmark is that exploding barrel that occurs in the middle as it always gets a huge stutter.

Also as the camera pans upwards in the beginning there will be multiple stutters, especially on the first run.

 

[Xenphor]

Ok I looked at the graph it gives and think it illustrates the problem:

See all those spikes where it drops down?? Is that not normal?

 

[Rhezuss]

Don't blame the card, blame the bad optimization and lack of PC recognition from some devs.

 

[BallaTheFeared]

Yes it's normal for the benchmark with Metro 2033.

http://www.overclock.net/t/817064/metro-2033-benchmark-thread-using-official-tool

How does the actual game feel for you?

If you want a frame time chart, use FRAPS with FrafsBenchViewer.

 

[Xenphor]

Hmm what's up with those thick black areas on the graph? Your graph doesn't seem to have as many drops or at least less severe.. I did another run with some higher level settings and get tons of drops, especially the beginning.

Options: Resolution: 1280 x 1024; DirectX: DirectX 11; Quality: Very High; Antialiasing: MSAA 4X; Texture filtering: AF 16X; Advanced PhysX: Disabled; Tesselation: Enabled; DOF: Disabled

The actual game feels a bit sluggish. I have to force triple buffering and vsync through d3doverrider to get rid of the tearing that plagues the game, and then that subsequently gives me input lag. Really any time I try to pan the camera around using my 360 joystick I will run into stutters.

 

[lavaheadache]

Yes it's normal for the benchmark with Metro 2033.

http://www.overclock.net/t/817064/metro-2033-benchmark-thread-using-official-tool

How does the actual game feel for you?

If you want a frame time chart, use FRAPS with FrafsBenchViewer.

Doesn't do that with my 7970.

I have run that bench so many times too and haven't seen that. The only slow down I get is my the camera passes the tank and you are looking at that super bright spot light. I think I drop down to about 35-40 fps @ 1080 right there. If I drop the settings to the second to highest the fps go way up. Lighting absolutely kills framerates in that game.

 

[BrightCandle]

I would really like to run and confirm your problem but alas having owned it and never actually tried to play it until today I found that the aspect ratio is totally broken and its unplayable. I wouldn't trust any benchmark figures that it kicks out, and it appears unfortunately that I can't fix the aspect issues I have with it. How rubbish.

 

[BallaTheFeared]

I linked to a huge pool with cards from both sides spanning several generations...


7970

I see OP did try the game, you should have a pretty smooth experience with the game.

What type of drive is the game installed on? What ram speed /timings are you running?

Also is the game fully patched? Try removing your forced 3rd party settings and using simply Adaptive Vsync as well.

 

[blackened23]

It should not be jerky. Especially not at your resolution.

Trust me, it should be smooth. While metro 2033 is pretty stupid in that half the graphics settings don't actually improve the graphics, with low or mid settings it should be smooth without stuttering. I haven't actually played it in some time, but while the benchmark can be all over the place - playing through the game should not be an issue at 1080p. I played metro 2033 @ 1080p on a GTX 580 ages ago, single card, and it was absolutely always smooth (MOST of the game is small indoor areas, so MOST of it is fast).

If its stuttering it's not how it's supposed to be, there's an issue. Maybe a driver one, not sure. Do you have the latest version of physx installed? Sometimes that can cause issues since the game uses it for collision detection even if you're not using GPU physx.

 

[Xenphor]

It should not be jerky. Especially not at your resolution.

Trust me, it should be smooth. While metro 2033 is pretty stupid in that half the graphics settings don't actually improve the graphics, with low or mid settings it should be smooth without stuttering. I haven't actually played it in some time, but while the benchmark can be all over the place - playing through the game should not be an issue at 1080p. I played metro 2033 @ 1080p on a GTX 580 ages ago, single card, and it was absolutely always smooth (MOST of the game is small indoor areas, so MOST of it is fast).

If its stuttering it's not how it's supposed to be, there's an issue. Maybe a driver one, not sure. Do you have the latest version of physx installed? Sometimes that can cause issues since the game uses it for collision detection even if you're not using GPU physx.

Well I have whatever physx came with the 313.95 drivers. I'm running on a clean install of Windows with those drivers.

Metro 2033 has always run poorly regardless of the driver. I've also run it on a 550ti, 6850, and a 670 with poor results.

As far as the game goes, I can give specific details on which areas caused very bad stutter:
Many places in the intro will cause stutter, like when you're just coming out of the tunnels, there will be heavy use of lighting coming from the outside that kills the framerate. Once outside however, the game is smooth. After that the big fight at the end is bad.

Opening of the game with the dad and kid

Opening of the game when looking at the target practice area after you get supplies

Opening of the game, when you're just outside the hospital, there's a guy that comes through the outside door, and when looking at the lighting behind him, there is huge performance hit

Another example of the lighting killing performance is when you get outside for the first time and look up at the ceiling where the light is coming through. Basically any section of the game that uses that lighting technique will tank hard. Roaming outside can be mostly smooth as long as that lighting technique is not present

edit: I'm assuming then that you don't get the spikes on the graph like in mine? I don't suppose you could do a run and post a graph?

 

[BallaTheFeared]

Xenphor, what HDD are you using for the game?

 

[Xenphor]

Xenphor, what HDD are you using for the game?

I'm using a WD Blue drive.

 

[BallaTheFeared]

Do you have a pagefile, or did you disable it?

Just to be clear, is this the only instance where you can recreate the issue?

 

[Xenphor]

I have a pagefile although I reduced it from 8gb to 4gb. It's set at 4096mb min and max. I'm assuming I only need it on one of my hard drives correct?

I've been able to detect stuttering on a number of games but never as bad as Metro 2033. If I cap the frame rate to 60fps and use a combination of vsync and triple buffering, I can usually minimize the stutter in other games. My goal would be to have every game running as well as Unreal Tournament 3, which I've found to be the best performing game I have. I know it's older but at least I can maintain a locked 60fps at all times.

 

[felang]

Apparently a portion of the benchmark has low fps no matter what the processor, as stated in this article by THG.

http://www.tomshardware.com/reviews/gaming-processor-frame-rate-performance,3427-4.html

 

[Saffron]

I have a pagefile although I reduced it from 8gb to 4gb. It's set at 4096mb min and max. I'm assuming I only need it on one of my hard drives correct?

I've been able to detect stuttering on a number of games but never as bad as Metro 2033. If I cap the frame rate to 60fps and use a combination of vsync and triple buffering, I can usually minimize the stutter in other games. My goal would be to have every game running as well as Unreal Tournament 3, which I've found to be the best performing game I have. I know it's older but at least I can maintain a locked 60fps at all times.

I agree with bad Metro 2033 performance in general regardless of hardware, but you do state you need to use vsync and triple buffering to minimize stutter in other games. You shouldn't get stutter in any game with these setting enabled or not with your current computer. This points more towards hardware problems, whether it be the drivers/settings or the hardware itself. The poor performance in Metro just amplifies it I think.

There are a lot of older and newer games that I can play with vsync off and not get any type of screen tearing, surprisingly enough. That or it's so unnoticeably minimal I don't need it on. Stutter is non-existent for me with my computer (very similar to yours, see below).

 

[Xenphor]

Apparently a portion of the benchmark has low fps no matter what the processor, as stated in this article by THG.

http://www.tomshardware.com/reviews/gaming-processor-frame-rate-performance,3427-4.html

I suppose that one dip to 20fps looks to be consistent, but I don't really see the same sharp spikes downwards that were at the beginning of my benchmark.