Recording with FRAPS or Afterburner fixes stutter? PLEASE HELP

[Stutters]

UPDATE: I FOUND OUT WHY TF2 IS STUTTERING!

Its antialiasing.....Even though I get tons of FPS with X4 antialias, if I set it to none then the game is perfectly smooth as if I was recording. Now why in the hell would recording with fraps completely fix stutter that was introduced by antialiasing?!?!?!

4xaa + not recording with FRAPS = Stuttery mess.
4xaa + recording = Perfectly smooth.
0xaa + not recording = Perfectly smooth.

Unfortunately I can't disable the 1 antialiasing that is enabled in 3dmark11 and 3dmark vantage to confirm this fixes those situations too.

Posting this here in hopes of someone being able to help me fix this issue. specs:

i7 920
gigabyte X58-UD3R SLI
6GB DDR3
GTX460

I have been struggling to find a fix for stuttering that I have been experiencing in games and benchmarks. I went through a number of different drivers but none of them helped. Typing "GTX460 stutter" in google brings up a frightening amount of people having stutter problems with their GTX460's so I was hoping to find a fix but I didn't. I wanted to contact MSI to troubleshoot the problem and provide them proof of the stuttering and while recording with fraps it was fixed. It just makes no sense to me.

For an example, 3dmark11's second and third tests both stutter like crazy (the first one is fine for whatever reason), the FPS is very erratic shooting up and then plummeting down but as soon as I hit record with fraps the FPS stabilizes and the stutter is gone. Does anyone know what could be causing this? Google only showed a couple people saying the same thing.

Things that I have tried so far without success:

- Tried an old monitor.
- Disabled all CPU related power saving features in the bios and windows.
- Disabled Hyperthreading.
- Removed sound card and uninstalled its drivers to try using on-board audio.
- Reinstalled windows and along with it chipset/lan/sound card drivers.
- Reinstalled games and benchmarks.
- Tried old and current Nvidia drivers.
- Stress tested the GPU with kombustor and runs of 3dmark11 and vantage.
- Stress tested the CPU/RAM with IBT/OCCT/HCI memtest and prime95 blend at stock optimized defaults and at my 3.5Ghz normal overclock. (all passed)
- Ran DPC latency checker. (nothing over 500 while playing)
- Tried disabling and enabling 64 bit HPET.
- Ran chkdsk and an error check with HDtune on my hard drive.(no errors found with either test)
- Tried testing with Avast! free completely disabled and even tried with MSE instead.
- Replaced SATA cables and tried gigabyte sata ports instead of the normal and better ICH10R sata ports.

Recording with MSI afterburner also stablizes my FPS and completely fixes the stutter just like FRAPS does.

Another example from a couple others having this problem: http://www.tomshardware.com/forum/117551-13-stutter-issues-games

Another example: http://www.swtor.com/community/showthread.php?t=180832

videos showing the second GPU test of 3dmark11:

WITHOUT fraps recording:
http://www.youtube.com/watch?feature=player_embedded&v=APHktVK7Npw

WITH fraps recording:
http://www.youtube.com/watch?feature=player_embedded&v=rrsiMunkAFU

 

[imaheadcase]

Fraps will record at the best frame rate it can achieve when its running. Without it video drivers are using all the power it can use resulting in fps looking like its jumping around. Its perfectly normal.

The only way to get "smoother" results is to limit fps when you are playing the game. Its going to stutter no matter what to a degree, but like anything video recording you are going to be limited to writes to hardrive which will effect gameplay.

Now if you are suggesting that simply having FRAPS running without it doing anything reduces stutter, i'm not sure cause not installed right now, but it might be fraps by default when running limits fps in settings so when you start recording it does not stutter on certain systems.

 

[Stutters]

Fraps will record at the best frame rate it can achieve when its running. Without it video drivers are using all the power it can use resulting in fps looking like its jumping around. Its perfectly normal.

The only way to get "smoother" results is to limit fps when you are playing the game. Its going to stutter no matter what to a degree, but like anything video recording you are going to be limited to writes to hardrive which will effect gameplay.

Now if you are suggesting that simply having FRAPS running without it doing anything reduces stutter, i'm not sure cause not installed right now, but it might be fraps by default when running limits fps in settings so when you start recording it does not stutter on certain systems.

The stuttering that I am seeing is NOT completely normal, it definitely wasn't there before. Fraps is recording at 60FPS and my single GTX460 can only manage 21 or so FPS in the third 3dmark11 test which is a stutter nightmare without fraps recording and perfectly smooth when it is recording.

Again, this is not normal by any means. The stutter was not present in games or benchmarks, I am trying to find out what is causing it. Capping FPS does not help the stuttering at all (I tried it with afterburner frame limiter), only recording will completely fix it. Just having fraps open doing nothing but displaying FPS doesn't help anything.

 

[imaheadcase]

I'm not sure why you are concerned with the 3dmark11 test. I have a Evga GTX460 in other system and confirmed fraps works fine. But you will get stuttering when recording at higher fps than card will get.

If that test is smooth when its recording, like i mentioned fraps is going by video card, if its a stuttering nightmare without it, its because fps is low (confirmed by video you posted) and simply card performance, when fraps is recording and its perfectly smooth is because fraps is targeting a certain fps (based on whatever setting you have in it) and will maintain that no matter what.

Default FPS for movies is 30fps.. you are well below that. That confirms GTX460 is trying to get what it can in fps during test. Running FRAPS will limit FPS and smooth it because of overhead between card and recording.

 

[Stutters]

Okay, ill just make things really simple. These videos are not mine but they perfectly represent what is happening. Using the third test for an example:

THIS stuttery mess is exactly what I see on my computer when I am NOT recording with fraps or after burner: http://www.youtube.com/watch?feature=player_detailpage&v=3K7PC-bZnwM#t=86s

THIS is how the same test looks when I record with fraps (My rig gets almost the same FPS, maybe a little more. Low FPS while still being smooth):http://www.youtube.com/watch?feature=player_detailpage&v=BfU7InXwuQc#t=96s

This problem is not just with 3dmark 11, it also happens in other games. 3dmark 11 is just a quick and easy way to reproduce the problem.

 

[Cerb]

Is the GPU running at max clock all the time when FRAPS isn't running?

Also, have you disabled software power-saving for the CPU in Windows?
Power->change->advanced->Processor power management->set min and max to 100.

 

[imaheadcase]

Like i said stutters, with it recording it maintains a normal fps to smooth out video.

Video card running it will try to get max fps no matter what with will result in hitching you see.

 

[kylebisme]

The videos show lower minimum framerates when not recording, that's not normal behavior.

 

[Cerb]

Also, try disabling threaded optimization, if currently forced on or set to auto, globally. Kind of a long shot, but it won't hurt to try.

 

[Childs]

This is actually the opposite behavior most people see when they are recording vs not recording. What is your CPU and GPU utilization when you are not recording?

 

[Stutters]

Like i said stutters, with it recording it maintains a normal fps to smooth out video.

Video card running it will try to get max fps no matter what with will result in hitching you see.

This is not normal. The stutter wasn't always there.

Is the GPU running at max clock all the time when FRAPS isn't running?

Also, have you disabled software power-saving for the CPU in Windows?
Power->change->advanced->Processor power management->set min and max to 100.

Yes I have tried this but since it didn't help I turned the power saving features back on. The GPU stays at 781 core clock and 99%usage during the tests as reported by GPU-z and afterburner.

Also, try disabling threaded optimization, if currently forced on or set to auto, globally. Kind of a long shot, but it won't hurt to try.

Thanks for the suggestion but after giving it a try it didn't help.

This is actually the opposite behavior most people see when they are recording vs not recording. What is your CPU and GPU utilization when you are not recording?

This is without recording after running the first three GPU tests of 3dmark11 and then closing the program and taking a screen shot. GPU usage is always 99% whether I'm recording or not.



Uploaded with ImageShack.us

 

[Rakewell]

Prolly a dumb question,

But does the stutter coincide with an audio event?

Are you using onboard sound?

You might need to update your audio drivers...

EDIT:

Nevermind. Saw the video of stutter, it's clearly not audio interference. please disregard my comment.

 

[Childs]

This is without recording after running the first three GPU tests of 3dmark11 and then closing the program and taking a screen shot. GPU usage is always 99% whether I'm recording or not.

Use Afterburner, add CPU and GPU monitoring, and run your game windowed and look at the graphs. Hopefully you'll see something in the monitors.

 

[Stutters]

Use Afterburner, add CPU and GPU monitoring, and run your game windowed and look at the graphs. Hopefully you'll see something in the monitors.

I can't find an option anywhere in Afterburner to monitor the CPU. Here is a screen of the graph after running the same three 3dmark11 tests like I did above: (the drops are caused by the loading screens between each test)



Uploaded with ImageShack.us

 

[Childs]

The framerate looks relatively steady in the graphs, but 20 seems low. What does it look like when you record?

 

[Stutters]

The framerate looks relatively steady in the graphs, but 20 seems low. What does it look like when you record?

Afterburner doesn't refresh the FPS counter fast enough for it to show up in the graph. That's why I added those videos in the OP where I recorded the built in FPS meter with and without fraps recording.

 

[Stutters]

I noticed something while messing with TF2 in a created server all alone. If I turn vsync on and stand still the built in net graph shows a 100% solid 59 FPS. If I start moving or spinning the camera the FPS starts to spike up and down from what I can see is 55-62. Its very hard to see because it refreshes so fast in the net graph. When I start a recording with fraps the FPS stays on 59-60 and only those two values, it doesn't bounce around as much and the stutter is gone.

EDIT: I apologize for the double post.

 

[Stutters]

I FOUND OUT WHY TF2 IS STUTTERING!

Its antialiasing.....Even though I get tons of FPS with X4 antialias, if I set it to none then the game is perfectly smooth as if I was recording. Now why in the hell would recording with fraps completely fix stutter that was introduced by antialiasing?!?!?!

4xaa + not recording with FRAPS = Stuttery mess.
4xaa + recording = Perfectly smooth.
0xaa + not recording = Perfectly smooth.

Unfortunately I can't disable the 1 antialiasing that is enabled in 3dmark11 and 3dmark vantage to confirm this fixes those situations too.

 

[imaheadcase]

sigh. I'm not going to explain it again.

 

[Cerb]

sigh. I'm not going to explain it again.

No one is asking you to. Everyone understood what you said. It's just not the problem. Variable FPS and stuttering are not the same thing. Whether or not it's exactly what happens, it feels like the frames are rendering, waiting a few refreshes, then cycling through rendered frames at the fresh rate, then repeating that cycle, rather than swapping out buffers at the next refresh.

I never turn AA off, so the OP's problem could very well be my own, as well, with most Forcewares (280.26 w/o HDMI works fine for me, yet no other version since 257.21 does).

--

For DX9 games, you might try an FXAA injector, and light transparency SSAA (I use 2xTR-SSAA w/ FXAA), and see if that is OK (w/o transparency AA, close up things are fuzzy, far away pixelated).

 

[imaheadcase]

For the problem he was having, what i said, is exactly what he was having. Video card with AA on - stuttering since it could not handle it. Forget it, he figured it out anyways.

 

[Childs]

For the problem he was having, what i said, is exactly what he was having. Video card with AA on - stuttering since it could not handle it. Forget it, he figured it out anyways.

AA makes sense for general stuttering, but what doesnt is how Fraps would smooth out his experience. Normally recording video adds additional stuttering because of writes to the HD. Then take into account his data points about setting Fraps to record at 60fps, and in 3Dmark he only gets 20fps, everything should be below the Fraps max.

Now if its just TF2 where normally he is above 60fps and then it spikes from like 80 to 60, which would be noticeable, then recording would be smoother because of the limit.

 

[Cerb]

For the problem he was having, what i said, is exactly what he was having. Video card with AA on - stuttering since it could not handle it. Forget it, he figured it out anyways.

4x generally has a <25% performance hit, that degrades performance smoothly, without variable input lag. Lower FPS != stuttering. I still use a Core 2 Duo, so I know all about low FPS . Some DX11 games are exceptions, but TF2 wouldn't qualify, and the 3Dmark video is not normal, and is about what the 270 and 275 drivers were like, for me. This a distinct problem from low FPS, and does not go away when FPS is high. It seems to mostly occur with Fermi on Windows 7 64-bit, though I've seen some Kepler reports of it, too.

 

[Stutters]

sigh. I'm not going to explain it again.

For the problem he was having, what i said, is exactly what he was having. Video card with AA on - stuttering since it could not handle it. Forget it, he figured it out anyways.

4xAA without recording = 60fps with Stutter @ 19% GPU usage.
4xAA WITH recording = 60fps Perfectly smooth with 22% GPU usage.
4xAA without vsync = 270 FPS and 90% usage.

Yeah, the card really cant handle TF2??? You're just posting bullshit without suggesting possible solutions. If the card couldn't handle the game then the stuttering would be there whether I was recording or not.

Please don't explain.