Multi GPU stuttering captured with 300FPS camera

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
Singel GPU have micro stuttering too...

I believe its about fps not running stable, so the time between pictures shown varies.

This happends with a single gpu as well.
If you lock framerates to say v-sync and keep it constantly there, without diping ect, that should go along way to removeing "micro stutter". (thought it might still be there, like this 300fps camra show)

Also

"Just because you cannot percieve microstutter dosn't mean it isn't there."
Its there, but whats the issue if its "not percievable"?
You dont notice it, then there is no issue with it being there.


Edit:
*** Also why test it at 30 fps? why didnt the do it at 60fps? where its less noticeable?
 
Last edited:

Zanovar

Diamond Member
Jan 21, 2011
3,446
232
106
there is software out there that can measure this,its accurate too.
its easy to replicate,just put a midrange/highend graphics card in a pcie 1.0 x4,your fps will be high 60 constant,but s stuttery mess.
 
Last edited:

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
It happens on single gpu too, it's just not as perceptible because the frame times are more even compared to multi gpu, but they are never perfectly even unless vsync is enabled.
 

Keysplayr

Elite Member
Jan 16, 2003
21,211
50
91
It happens on single gpu too, it's just not as perceptible because the frame times are more even compared to multi gpu, but they are never perfectly even unless vsync is enabled.

I don't know man. The single GPU in slow-mo looked awful steady to me. Like a beating drum.
 

Lonbjerg

Diamond Member
Dec 6, 2009
4,419
0
0
Singel GPU have micro stuttering too...

I believe its about fps not running stable, so the time between pictures shown varies.

This happends with a single gpu as well.
If you lock framerates to say v-sync and keep it constantly there, without diping ect, that should go along way to removeing "micro stutter". (thought it might still be there, like this 300fps camra show)

Also

Its there, but whats the issue if its "not percievable"?
You dont notice it, then there is no issue with it being there.


Edit:
*** Also why test it at 30 fps? why didnt the do it at 60fps? where its less noticeable?


Care to document you claims?
 

tviceman

Diamond Member
Mar 25, 2008
6,734
514
126
www.facebook.com
I think this test needs to be run at 60fps in-game. If multi-GPU solutions are only pumping out 30fps on the game you're playing, it's time to upgrade or drop some visual settings.
 

Aristotelian

Golden Member
Jan 30, 2010
1,246
11
76
It ends up being a give/take situation. Would I have rather spent ~1000 euros on a single gpu solution that was as fast as my 580s in SLI? Yup. Is such a thing available? Nope. Is there a single gpu solution that can play every game at 50fps + at 2560, max settings? Nope. I like the eye candy, though. Microstutter is situation dependent, for example if you see EtheEnthusiast's benchmarks on youtube (even with 4x 3GB 580s in Metro 2033 at higher resolutions), you'll easily notice the stutter. I haven't experienced much myself - some in Crysis 2 during online play, but the truth is if I was as serious about FPS games now, as I was 'back in the day' (Quake 3 tournaments), I'd still be using a trinitron tubed CRT and a single gpu for that ultra-smooth gaming experience.
 

Kenmitch

Diamond Member
Oct 10, 1999
8,505
2,250
136
I don't know man. The single GPU in slow-mo looked awful steady to me. Like a beating drum. __________________

It looked that way to me also :)

But on the other hand I found it strange how the images didn't match as far as coloring depth etc. Makes me beleive it was done on two systems with different video cards....Kinda like the results are what they wanted to show.
 
Nov 26, 2005
15,189
401
126
This should help the people claming that there is no stuttering on multi-GPU setups.

http://www.youtube.com/watch?v=zOtre2f4qZs&feature=player_embedded

Just because you cannot percieve microstutter dosn't mean it isn't there.
People have different thresholds in what we can percieve for sound...to visual stimuli.

Is the refresh rate something to consider in a test like this or is it non-relevant?

EDIT: could be poor video editing too

This test should be done on a CRT anyways.
 
Last edited:
May 13, 2009
12,333
612
126
Wow! I didn't know it was that bad. Even when they showed the first part of the video and it was 60fps you could see the microstuttering. It was like it just stops for a second then goes again. No offense to anyone running sli or crossfire but that is just horrid.

Looks like I was right when I bought my asus direct 2 cu 580. I'd rather play at 60 fps with a single gpu than 120fps with dual cards. I have the ultimate setup for gaming IMO. A 120hz monitor with a highly overclocked single 580.
 

Elfear

Diamond Member
May 30, 2004
7,163
819
126
Wow! I didn't know it was that bad. Even when they showed the first part of the video and it was 60fps you could see the microstuttering. It was like it just stops for a second then goes again. No offense to anyone running sli or crossfire but that is just horrid.

Looks like I was right when I bought my asus direct 2 cu 580. I'd rather play at 60 fps with a single gpu than 120fps with dual cards. I have the ultimate setup for gaming IMO. A 120hz monitor with a highly overclocked single 580.

But would you rather play at 30fps with a single card or 55fps with dual cards? That's where the beauty of dual cards come into play. It makes some games playable where a single card can't keep up. IMO you have a great solution for 1080p or lower with some settings turned down in the games that are resource hogs. I wouldn't call that the ultimate gaming solution however.
 

Zanovar

Diamond Member
Jan 21, 2011
3,446
232
106
Care to document you claims?
whats your thoughts on this you green weirdo?


Insults are not acceptable.

Please familiarize yourself with the AnandTech Forum Guidelines:
1) No trolling, flaming or personally attacking members. Deftly attacking ideas and backing up arguments with facts is acceptable and encouraged. Attacking other members personally and purposefully causing trouble with no motive other than to upset the crowd is not allowed.

Idontcare
Super Mod
 
Last edited by a moderator:

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
I don't know man. The single GPU in slow-mo looked awful steady to me. Like a beating drum.


It's just more even than dual GPU, but not linear.


Care to document you claims?


You can do it yourself with FRAPS, just tell it to record frame times. Without vsync they will never be even. For 6 frames in 200ms of rendering time at 30 FPS:

With vsync it looks like
33.333
33.333
33.333
33.333
33.333
33.333

Without vsync, single GPU looks like
30.1
35.5
32.4
31.8
34.7
35.5

Without vsync, dual GPU looks like
10.8
56.2
15.4
28.9
66.1
22.6


Dual GPU stutters are more obvious, because your eye is more sensitive to the larger gaps between frames (over 16.6ms apart, 60 FPS drawing time) than two being less than 16.6ms apart. So it feels like your FPS is lower than what the counter says, the counter just counts the total number of frames drawn in one second, it's not an indication of how even frame times are.

PS.
The reason they ran it at 30 fps is because that's all you can see on youtube.
 
Last edited:
May 13, 2009
12,333
612
126
It's just more even than dual GPU, but not linear.





You can do it yourself with FRAPS, just tell it to record frame times. Without vsync they will never be even. For 6 frames in 200ms of rendering time at 30 FPS:

With vsync it looks like
33.333
33.333
33.333
33.333
33.333
33.333

Without vsync, single GPU looks like
30.1
35.5
32.4
31.8
34.7
35.5

Without vsync, dual GPU looks like
10.8
56.2
15.4
28.9
66.1
22.6


Dual GPU stutters are more obvious, because your eye is more sensitive to the larger gaps between frames (over 16.6ms apart, 60 FPS drawing time) than two being less than 16.6ms apart. So it feels like your FPS is lower than what the counter says, the counter just counts the total number of frames drawn in one second, it's not an indication of how even frame times are.

PS.
The reason they ran it at 30 fps is because that's all you can see on youtube.

Even if dual gpu times were even with vsync enabled I'd be able to see that huge stutter. It's like someone doing the robot.
http://www.youtube.com/watch?v=4YJ3BTKMILw
 

JAG87

Diamond Member
Jan 3, 2006
3,921
3
76
Even if dual gpu times were even with vsync enabled I'd be able to see that huge stutter. It's like someone doing the robot.
http://www.youtube.com/watch?v=4YJ3BTKMILw

No you wouldn't. Vsync enables double or triple buffering which buffers frames so that they can be spit out at you evenly, whether it's 16.666 ms apart for 60 Hz, or 8.333 ms apart for 120 Hz.

This is why vsync introduces input lag, because you can't buffer frames and still be immediately responsive to changes.

But there's a trick to avoid that too, and enjoy perfectly even frames without input lag.
 

Madcatatlas

Golden Member
Feb 22, 2010
1,155
0
0
Interesting link ljonberg! cheers, hadnt seen that as an issue and was thinking of adding a second 460.
Any more documentation on microstuttering? i like vids and not endless text documents:(
 

notty22

Diamond Member
Jan 1, 2010
3,375
0
0
There are all sorts of things our eyes might or will never see, unless its manipulated.
Don't know why this popped in my head, but subliminal messages that used to be slipped in to movie frames.

subliminal-message-540x380.jpg
 

Absolution75

Senior member
Dec 3, 2007
983
3
81
Cool video - at first I was skeptical because the fact that the youtube video would likely be 24fps, but since they slowed it down so it seems accurate.
 

alcoholbob

Diamond Member
May 24, 2005
6,380
448
126
That's pretty much how my gaming experience is with dual 6970s. If you pan the screen or run too fast it does start to stutter in routine intervals in a lot of games. However, it improves the gaming experience by keeping the framerate up when a single card would struggle--(would you rather have routine stutter or continuous stutter?)
 
May 13, 2009
12,333
612
126
That's pretty much how my gaming experience is with dual 6970s. If you pan the screen or run too fast it does start to stutter in routine intervals in a lot of games. However, it improves the gaming experience by keeping the framerate up when a single card would struggle--(would you rather have routine stutter or continuous stutter?)

I'd rather have neither. I'm happy with my overclocked GTX 580. I don't get any stuttering and not many games I can't run with very high fps. If I find a game is really demanding I lower AA and I get a nice clean 120fps with no stuttering. There is a reason the 580's cost $500 and haven't seen a price drop even though everyone says "what a bad deal the 580's are since 2 GTX 460's offer equivalent performance for much less money." Yeah enjoy your slideshow gaming.