• 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.

6900 CF and microstutter

X

x3sphere

2[H]4U
2FA
Joined
Dec 8, 2007
Messages
2,805
I'm wondering with all the advancements AMD has made in regards to CFX on the 6900 series, if the effects of microstutter have been minimized. So far I haven't seen any reviews investigate this. Any CF owners want to chime in?

Please don't post in this topic and say it doesn't exist. It does. If you want an example of microstutter, here's one of the best vids I've come across

http://www.youtube.com/watch?v=aQa97-ApWvc&feature=player_detailpage#t=140s

You can see, at around the 22 FPS mark, the 480 appears MUCH more smoother than a 5970 even though they are rendering frames at practically the same FPS.
 
Don't hold your breath. You're not likely to get away with the microstutter as long as they're running AFR mode.
 
I thought Micro-stutter was only "really" noticeable when you run at low frames to begin with, frames that you wouldn't play a game at :p?


Also if you watch their next video, which is 5870 vs gtx480, seems like the 5870 exhibits the same behavior as the 5970, is it a driver issue? card choking on heavy tessellation issue? or Micro-stutter issue?
 
Last edited:
Digital Viper-X- said:
I thought Micro-stutter was only "really" noticeable when you run at low frames to begin with, frames that you wouldn't play a game at :p?
Click to expand...

Yeah it's only noticeable at lower frames. Thing is, with a game like Metro at 2560x1600 maxed you are still going to be dropping into the 25-30 FPS range frequently even with 6950 or 6970 CF. While the game would normally be playable under those framerates, microstutter would kill the experience for me..


Digital Viper-X- said:
Also if you watch their next video, which is 5870 vs gtx480, seems like the 5870 exhibits the same behavior as the 5970, is it a driver issue? card choking on heavy tessellation issue? or Micro-stutter issue?
Click to expand...

It COULD be driver related, as there are different rendering modes to handle CFX. AFR, as Raap mentioned, is one. It's probably the most effective in terms of getting high FPS but also the most prone to microstutter. And as far as I know AMD drivers do not let you change the rendering mode.
 
Last edited:
x3sphere said:
Yeah it's only noticeable at lower frames. Thing is, with a game like Metro at 2560x1600 maxed you are still going to be dropping into the 25-30 FPS range frequently even with 6950 or 6970 CF. While the game would normally be playable under those framerates, microstutter would kill the experience for me..
Click to expand...

lame :confused:
 
I have owned the following SLI & Crossfire setups:

2 8800ultras in SLI
1 5970
2 Sapphire Toxic 5870 2gb

Based on my prior experience with SLI & CF, if you are not getting well over 60 fps with multiple cards then it is not worth it. In my opinion, I would rather get 30fps with a one card rig than 50fps with a multi card rig.

The micro stutter (even minor MS) kills all FPS game enjoyment for me when the visuals are not as responsive as my controls/reflexes.
 
x3sphere said:
I'm wondering with all the advancements AMD has made in regards to CFX on the 6900 series, if the effects of microstutter have been minimized. So far I haven't seen any reviews investigate this. Any CF owners want to chime in?

Please don't post in this topic and say it doesn't exist. It does. If you want an example of microstutter, here's one of the best vids I've come across

http://www.youtube.com/watch?v=aQa97-ApWvc&feature=player_detailpage#t=140s

You can see, at around the 22 FPS mark, the 480 appears MUCH more smoother than a 5970 even though they are rendering frames at practically the same FPS.
Click to expand...
Few things:
-This thread title and op implies Microstutter is worse on AMD cards. Nvidia and AMD use the same multi gpu communication rendering.

-Pretty sure that video is not micro stutter, that is just low tessellation performance. Looks like you are using lower erratic fps counter as the measurement for micro stutter which is wrong.

-There are no visible effects other than smoothness. Its more a slow feeling of low fps when you are getting above 40+fps average. You can see it at low fps however. That video is just bad tesseallation optimization.

-It exists but most of forum users mistaken normal stutter caused by other things ie input lag, lag spikes from hdd, bad optimizations
-None of the mainstream reviewers [H] Anand, Guru, Techpowerup ever mention it since the 4870x2 came out what does that tell you? Mountain out of molehill.

Turn on Vsync and triple buffering if you feel like there is microstutter in your game. Problem solved. Who even plays without vsync on single player games?
http://www.rage3d.com/reviews/video/ati4870x2cf/index.php?p=2


edit here is what I think is a better video You can compare more apples to apples 5970 to 5870 with real gameplay 1:12 to 1:16 when he goes left around the pole. http://www.youtube.com/watch?v=emG7ZNIsxw8&fmt=22

If you don't notice it don't worry because that is a good thing. That's pretty much microstutter = mountain out of a molehill.
 
Last edited:
FYI, if a game is running below 30fps your going to see "micro-stutter" as you call it. I know the term and effect is real, but below 30fps in a fast moving game your going to see stuttering....its human physiology. You know how they talk about movies running at 24fps/1080p....thats cause the film industry records at 24fps...above that its hard for the human eye to distinguish a individual frames in film. Some people are more acutely aware of it then others and gaming situation dictates, but if you can hit 30fps in a game then its gonna look the same as 120fps.

I've been running CF 5870s for about a year (upgrading to 6970s when they come in the mail) and ive not had an issue that im aware of.
 
Vexerz said:
Few things:
-Micro-Stuttering on SLI doesn't exist? SLI/CF have the same microstutter. This thread title and op implies Microstutter is worse on AMD cards. Nvidia and AMD use the same multi gpu communication rendering do they not?

-Pretty sure that video is not micro stutter, that is just low tessellation performance. Looks like you are using lower erratic fps counter as the measurement for micro stutter which is wrong.

-There are no visible effects other than smoothness. Its more a slow feeling of low fps when you are getting above 40+fps average. You can see it at low fps however. That video is just bad tesseallation optimization.

-It exists but most of forum users mistaken normal stutter caused by other things ie input lag, lag spikes from hdd, bad optimizations
-None of the mainstream reviewers [H] Anand, Guru, Techpowerup ever mention it since the 4870x2 came out what does that tell you? Mountain out of molehill.

Turn on Vsync and triple buffering if you feel like there is microstutter in your game. Problem solved. Who even plays without vsync on single player games?
http://www.rage3d.com/reviews/video/ati4870x2cf/index.php?p=2


edit here is what I think is a better video You can compare more apples to apples 5970 to 5870 with real gameplay 1:12 to 1:16 when he goes left around the pole. http://www.youtube.com/watch?v=emG7ZNIsxw8&fmt=22

If you don't notice it don't worry because that is a good thing. That's pretty much microstutter = mountain out of a molehill.
Click to expand...

I didn't mean to insinuate that microstutter is worse on AMD cards. It's the AMD forum so I wanted to refrain from bringing Nvidia comparisons in here.

I'd agree, 40FPS is the threshold. Anything lower then you can start to notice it. VSync only helps if your FPS is high enough, which doesn't work in the most demanding games like Metro or Crysis (at least at the res I'm running, at 1920 I reckon you'd be okay).

IF you notice it, though, it's hardly a mountain out of a molehill. I regret purchasing a 5970 since I'm pretty much limited to what the 5870 could do, give or take 10-15%. If I stress the card too much, then the effects of microstutter become evident. Although, with that said, I'd probably be able to get more mileage out of a 6900 CF setup thanks to the massively improved scaling.

Also, yeah, that video is probably a better comparison. The effects are a little exaggerated in the other clip, perhaps due to the tessellation load.
 
Last edited:
Are some people more sensitive to micro stutter? Because I have yet to notice it. Running eyefinity and crossfired 5850's I do sometimes see some moments of slowdown at certain times (depending on the game), but generally I try to tune my seetings so I can keep my average frame rate at around 60fps. But as far as a consistant phenomenom, I cannot say that I experience it.
 
x3sphere said:
I didn't mean to insinuate that microstutter is worse on AMD cards. It's the AMD forum so I wanted to refrain from bringing Nvidia comparisons in here.

I'd agree, 40FPS is the threshold. Anything lower then you can start to notice it. VSync only helps if your FPS is high enough, which doesn't work in the most demanding games like Metro or Crysis (at least at the res I'm running, at 1920 I reckon you'd be okay).

IF you notice it, though, it's hardly a mountain out of a molehill. I regret purchasing a 5970 since I'm pretty much limited to what the 5870 could do, give or take 10-15%. If I stress the card too much, then the effects of microstutter become evident. Although, with that said, I'd probably be able to get more mileage out of a 6900 CF setup thanks to the massively improved scaling.

Also, yeah, that video is probably a better comparison. The effects are a little exaggerated in the other clip, perhaps due to the tessellation load.
Click to expand...

This goes back to what I was saying, and Ill expand on it. Its a bigger issue with forum users because 9/10 they usually mistaken any kind of normal stutter (
bad optimizations, drivers, hdd hitching) to microstutter. Multi gpu's always have buggier drivers

Seems to be quite a misunderstanding between FPS and framerates. You can already see the replies in this thread as proof.
 
Is this video a single 480? If it is, my guess is that the 5970 was running in single GPU mode. Did anyone notice that the 5970 power consumption numbers were significantly lower? I find it hard to believe that a dual chip based 5970 would be using that much less power than even a single power hungry 480.
 
Un4given said:
Is this video a single 480? If it is, my guess is that the 5970 was running in single GPU mode. Did anyone notice that the 5970 power consumption numbers were significantly lower? I find it hard to believe that a dual chip based 5970 would be using that much less power than even a single power hungry 480.
Click to expand...

no the power figures are correct :p the 480 might be OC'd a bit too.
 
jhatfie said:
Are some people more sensitive to micro stutter? Because I have yet to notice it. Running eyefinity and crossfired 5850's I do sometimes see some moments of slowdown at certain times (depending on the game), but generally I try to tune my seetings so I can keep my average frame rate at around 60fps. But as far as a consistant phenomenom, I cannot say that I experience it.
Click to expand...

Some people (Hell a lot of people) don't even notice the insanely annoying Gamebyro stutter in Oblivion, Fallout 3, New Vegas, and Divinity 2. That one is blatantly obvious to me and a lot of people claim that they have never seen it.

On the other hand there are people out there who will blame every little stutter or irregularity on "microstutter". So as always you kind of need to take opinions on stuff like this with a grain of salt.

Vexerz said:
This goes back to what I was saying, and Ill expand on it. Its a bigger issue with forum users because 9/10 they usually mistaken any kind of normal stutter (
bad optimizations, drivers, hdd hitching) to microstutter. Multi gpu's always have buggier drivers

Seems to be quite a misunderstanding between FPS and framerates. You can already see the replies in this thread as proof.
Click to expand...

QFT
 
Last edited:
To add also if you are getting below 30fps consistently, you have bigger concerns than microstutter. What game is enjoyable below that? It makes the whole argument for microstutter below that fps moot really.
 
I noticed microstutter under heavy graphics situations in cod etc when all hell breaks loose even at high fram rates with both a nvidia and amd :( enough so I look for single card solutions.
 
p3sty said:
I noticed microstutter under heavy graphics situations in cod etc when all hell breaks loose even at high fram rates with both a nvidia and amd :( enough so I look for single card solutions.
Click to expand...

turn on vsync/framebuffering
 
l88bastard said:
ehhmm, vsync tends to add input lag which also sucks while playing first person shooters
Click to expand...

No perfect way to do it. Leave vsync off and enjoy your tearing (alot worse than microstutter). Even with single gpu I still turn on vsync (its 2010) unless its multiplayer.
If its multiplayer online do you care more about quality or performance. If you are competitive Im hoping its the latter.

also make sure triple buffering is on.
 
Last edited:
you are going to get plain stutter regardless of if you have two cards or not. It is not microstutter it is just you are running very low framerate. If microstutter was such a big issue you would see all kinds of articles on it. I think it is there but some people just think of it too much.
 
NKD said:
you are going to get plain stutter regardless of if you have two cards or not. It is not microstutter it is just you are running very low framerate. If microstutter was such a big issue you would see all kinds of articles on it. I think it is there but some people just think of it too much.
Click to expand...

The problem is the naming itself "micro stutter". Its become a buzzword which doesn't imply everything in its naming. Alot of confusion on these so called 'hard' forums.

Should be called 'framerate afr asynchronization' or something similar to stop the confusion.
 
Last edited:
Vexerz said:
The problem is the naming itself "micro stutter". Its become a buzzword which doesn't imply everything in its naming. Alot of confusion on these so called 'hard' forums.

Should be called 'framerate afr asynchronization' or something similar to stop the confusion.
Click to expand...

I like the term 'micro-stutter' as a way of naming AFR-odd/even problems
(which I think is what the problem is with SLI/CF setups)

- general low frame-rate judder is just that - 'judder'

I'm assuming SLI/CF micro-stutter is caused by the difficulty of ensuring the frames from each card are displayed for the same amount of time
- which must be difficult to engineer as the time to generate each frame will vary from frame to frame.
 
Last edited:
I bought two 5850 for crossfire and tested multiple games at 5700x1200. Some games experienced "microstutter", especially BFBC2. The bottom line was that the experience was more playable with crossfire disabled, so I sold one of the cards back. Call it what you will--I tried plenty of video options to try to fix it and few online had the problem. The theory I had heard was that there was a limitation on the crossfire connector bandwidth.
 
Low framerate isn't much of a culprit as is framerate variation. What we're noticing is how the framerate changes. If it was a constant 25fps, no more no less, then playback wouldn't be jerky.
 
Digital Viper-X- said:
I thought Micro-stutter was only "really" noticeable when you run at low frames to begin with, frames that you wouldn't play a game at :p?


Also if you watch their next video, which is 5870 vs gtx480, seems like the 5870 exhibits the same behavior as the 5970, is it a driver issue? card choking on heavy tessellation issue? or Micro-stutter issue?
Click to expand...

Not true. I had my games are 40-50 and still noticed it. Game was very playable and smooth, but I'm so picky.
 
l88bastard said:
I have owned the following SLI & Crossfire setups:

2 8800ultras in SLI
1 5970
2 Sapphire Toxic 5870 2gb

Based on my prior experience with SLI & CF, if you are not getting well over 60 fps with multiple cards then it is not worth it. In my opinion, I would rather get 30fps with a one card rig than 50fps with a multi card rig.

The micro stutter (even minor MS) kills all FPS game enjoyment for me when the visuals are not as responsive as my controls/reflexes.
Click to expand...

This is exactly my sentiment. My next card will be a single card hands down. Microstutter is very noticeable in eyefinity since you can no longer scream at 100+ fps with two cards. I often find myself resorting to 1 card even if it's 35 fps vs. 50 fps. The single card at 35 fps honestly feels smoother.

The problem is the naming itself "micro stutter". Its become a buzzword which doesn't imply everything in its naming. Alot of confusion on these so called 'hard' forums.

Should be called 'framerate afr asynchronization' or something similar to stop the confusion.
Click to expand...
While I don't like the proposed name, 'micro sutter' is certainly a misnomer since the phenomenon isn't really 'stutter' in the typical sense. There's really no way to explain it until you've experienced it. Put it simply, it makes 50 fps feel like 25 fps.
 
Last edited:
I'm just bumping this thread since I started playing Metro 2033 recently with my 6950 crossfire setup. For the most part the game runs really, really smoothly. There are a few parts though where the frame rate dips down a bit. Now I've played this game with a single gpu and I remember dipping under 40fps and not being too bothered by it but when I get down to around the 40fps mark or lower the game doesn't run a smoothly as it should. Then it hit me, this must be the microstutter that so many people mention. I then ran a fraps bench in one of the areas that was bothering me and used that "microstutter program" to see if that is what in fact I was experiencing and sure enough thats what it was.


Now we all have to run Crysis when you buy a new piece of hardware. When I upgraded from a GTX280 to a single 6950 I gave Crysis a shot and I remember being pretty impressed. It ran really smoothly. When I ran it again with the crossfire setup I was a bit disappointed since for some reason that I didn't know at the time it just didn't feel as smooth as it should have when the framerate dipped. This actually happened to me as well with a 4870x2 as well. Even though the x2's framerate was higher I couldn't really tell the difference between it and the 280. It's one of the reasons that I sold the card. I was actually one of those people who said that they never noticed microstutter. The fact is that I had and it had annoyed me but I just didn't know what it was.

Thankfully with Metro it isn't constant and I wouldn't be playing the game maxed out with any single gpu on the market so I guess that I can't really complain there. The same could be said for a lot of games and since scaling is so good I'm not too upset. I do have to admit that Crysis is a bit disappointing, hopefully there aren't too many other games that perform the same thanks to microstutter.
 
USMCGrunt said:
Some people are more acutely aware of it then others and gaming situation dictates, but if you can hit 30fps in a game then its gonna look the same as 120fps.
Click to expand...


Please stop, that is a lie that people caught up due to wrong correlation.

24fps is used for a different reason, and it feels kinda regular because cameras take pictures of moving objects like blurs instead of perfectly defined objects.

I don't want to go out and fish all the explanation, but honestly, don't repeat that lie, it isn't about some people being more sensible than others, the 30fps thing is just a myth.

edit to add:
http://www.cookieofdoom.com/2009/07/01/busting-framerate-myths/
Also to add a bit more, if you look around you can find out that some film makers are pushing films to go all the way to 48fps instead of the 24 used currently, because quite honestly it is better.
 
Last edited:
Revdarian said:
Please stop, that is a lie that people caught up due to wrong correlation.

24fps is used for a different reason, and it feels kinda regular because cameras take pictures of moving objects like blurs instead of perfectly defined objects.

I don't want to go out and fish all the explanation, but honestly, don't repeat that lie, it isn't about some people being more sensible than others, the 30fps thing is just a myth.

edit to add:
http://www.cookieofdoom.com/2009/07/01/busting-framerate-myths/
Also to add a bit more, if you look around you can find out that some film makers are pushing films to go all the way to 48fps instead of the 24 used currently, because quite honestly it is better.
Click to expand...

I will retract that statement, since posting that ive learned a few things about framerate. Also to add to what your saying, James Cameron has been pushing to move films to 60fps...I read something about him purchasing a bunch of Red Epic cameras presumably for Avatar 2 (http://www.brightsideofnews.com/new...tar-2-announce-the-new-era-of-filmmaking.aspx)
 
i just got my cf 6950 last week and have been playing around with them all week and the first thing i noticed was that it didn't fell smooth kind of like the way a game fells when it drops to the mid to low 20 fps when i was running at 40ish fps, so for a test i disabled cf and used 1 card and fired up Metro maxed out and was getting like 20fps and it relay didn't fell much slower and when i turned down the settings and was tuning at about 30 fps it felt way smother then my cf running at 40ish, so i put cf back on and lowered some settings a bit to keep it over 50fps and it plays really well as long as the fps stay at the 45+ range
 
thax said:
i just got my cf 6950 last week and have been playing around with them all week and the first thing i noticed was that it didn't fell smooth kind of like the way a game fells when it drops to the mid to low 20 fps when i was running at 40ish fps, so for a test i disabled cf and used 1 card and fired up Metro maxed out and was getting like 20fps and it relay didn't fell much slower and when i turned down the settings and was tuning at about 30 fps it felt way smother then my cf running at 40ish, so i put cf back on and lowered some settings a bit to keep it over 50fps and it plays really well as long as the fps stay at the 45+ range
Click to expand...

This is the reason why I give up on Multi-GPU setup....
 
Just a random bit of info to toss in here that may affect some people... For people seeing excessive stuttering or uneven frames with AMD setups, try disabling Afterburner. There have been some issues creeping up lately where Afterburner keeps trying to poll/adjust the fan speeds way too often and it can cause performance problems. I think it was Guru3D or Rage3D where I saw the threads on it, but a bunch of people were like "OMG turning that off fixed everything" once they tried it :p
 
l88bastard said:
I have owned the following SLI & Crossfire setups:

2 8800ultras in SLI
1 5970
2 Sapphire Toxic 5870 2gb

Based on my prior experience with SLI & CF, if you are not getting well over 60 fps with multiple cards then it is not worth it. In my opinion, I would rather get 30fps with a one card rig than 50fps with a multi card rig.

The micro stutter (even minor MS) kills all FPS game enjoyment for me when the visuals are not as responsive as my controls/reflexes.
Click to expand...

I agree with this assessment based on my experience with crossfire. I've tried 2 x 5870s, a 5970, and more recently 2 x 6950s. If the frames get below 60 you might as well be on a single card with 30 to 40 fps as it feels more responsive despite the lower reported frames. This particularly applies to eyefinity configurations where frames below 60 are more common. Crossfire in single screen does perform very well.
 
Eyefinity is where my problem started. A single 6970 was enough for me on a single monitor to manhandle most games running 1920x1080. When I started running at 5760X1080 I felt the need for some extra horsepower to continue playing the same games with the same settings. In alot of games I see the same chuggy frames that others are talking about. I ran multiple FPS programs to check what I was seeing and never did my frames go below 50-60. The games I tried this on were New Vegas, Team Fortress 2, Half Life 2 w/ MOD10, Bound in Blood, and many others. I've been thinking about ditching the multi screen / multi-GPU setup and getting a 30' monitor and a single card like a GTX580.

I love my eyefinity setup but I also love eye candy.
 
You must log in or register to reply here.
Back
Top