Microstuttering?

M

mastermike707

Weaksauce
Joined
Jul 25, 2008
Messages
93
I am curious as to the actual impact of microstuttering in real world performance, as it has frankly scared me away from dual card setups. Do you find microstuttering to be an issue or is it just something that happens every once in a while? Keep in mind I am looking at dual 4850s or 4870s when I say this.
 
The issue is over rated. Does it happen, yes, is a big deal and is it something that can usually be addressed with setting options like buffering, yes. Most games are going to run so fast on a 48x0 CF setup that you'll be hard pressed to see it I bet.

SLI and CF are a lot better now than when first introduced. I'm not guaranteeing that you won't experience it, but when you start getting frame rates from the 24th century in most of your games you're probably not going to care anyway.
 
All indications point to the 48xx series basically eradicating the issue. Frames seem to be rendered at very tight intervals on the 4870 X2 and the 4850/4870 in Crossfire, so AMD seems to have this thing pretty well sewn up.
 
phide said:
All indications point to the 48xx series basically eradicating the issue. Frames seem to be rendered at very tight intervals on the 4870 X2 and the 4850/4870 in Crossfire, so AMD seems to have this thing pretty well sewn up.
Click to expand...

Yah, CF is fixed, but I hear SLi still has microstutter issues.
 
to be honest, i'm hard pressed to find 'microstuttering' throughout my day-to-day gaming.

maybe what i'm calling 'slow hard drive' is 'microstuttering', i don't know.

with my 8800gt sli setup, all systems seem nominal to me. i guess i could watch a youtube video about it to learn more, but meh.
 
Some relevant info. Looking at the graphs, the ideal is a straight horizontal line, with no "sawtoothing", indicating large discrepancies between alternately-rendered frames.
 
BigCactus said:
Yah, CF is fixed, but I hear SLi still has microstutter issues.
Click to expand...

It all depends on who you ask. I believe that I saw the issue in CoD4, but made a config change and it went away, but it may have been something else. Setting triple buffering just seemed to smooth things out. Config changes like this for games is not uncommon with AMD or nVidia hardware. It just depends. As far as I'm concerned its a non issue with my setup.

This is an issue that you can ignore IMHO with CF or SLI. Its more myth than man.
 
jimbobb said:
NOT FIXED!
Click to expand...
No data of any kind, just observations.
UT3 is always subjective liquid, we can no serious measure frame outliers. In Crysis improved the condition under DX9 also noticeably, the image distribution is lumpy in places but as with a GPU.
Click to expand...
They're saying UT3 and Crysis are fine, I believe.
Grid Race Driver presented microstuttering if the average frame rate falls below 50 and the wind chill Fps rate is then between 30 and 40. Also NfS Carbon and Half-Life 2 (Ger) show indisputably microstuttering under 40 Fps. The game presented river like the result of a single GPU with 20 Fps.
Click to expand...
I have no idea what that last sentence is supposed to mean.
 
phide said:
No data of any kind, just observations.

They're saying UT3 and Crysis are fine, I believe.

I have no idea what that last sentence is supposed to mean.
Click to expand...

Nope FPS is higher and MR ist not really fixed!
 
Micirostutter is a term used by douchebags who want to find a reason not to use multiple gpus. That being said, i NEVER saw microstutter on my 8800GT sli, my 8800GTS sli, my 9800GTX sli or my current 4870 crossfire.
 
How many of you who posted about microstuttering have actually used a multi GPU setup before?
 
Some notice it. I've not with my old 8800GTXs or current GTX280s.
Many as alg7_munif is implying have not run a multi-gpu setup and toss the term around as if it effects everyone to an extent they notice it.
 
Ive yet to see it on any of the games I play. I have a 9800 GX2. I enable vsync usually most of the time, mostly only if I am getting tearing. I have a 1920x1200 monitor too.

I think it's very overrated and like said, can be avoided by doing a few tweaks to configs or your settings, which is not too much of a deal.

Go forth and run CF 4870/4870x2/GTX 280s! Have fun! :p wish I had a mobo with two x16 pcie lanes :|
 
alg7_munif said:
How many of you who posted about microstuttering have actually used a multi GPU setup before?
Click to expand...

Had 6800GT, 7800512, 7950GX2, 3870X2 SLI/CF and I noticed the issue. That coupled with the other issues with multi-gpu setups have turned me off until I see something revolutionary regarding SLI or CF. Some love it, others don't. It's something you have to research, and try out yourself basically to see if advantages for what you are using it for out weight the disadvantages. When I see something revolutionary regarding multi-gpu I'm sure I will be back, but that hasn't occured and will not with the 4870X2 despite all the rhetoric.







 
alg7_munif said:
How many of you who posted about microstuttering have actually used a multi GPU setup before?
Click to expand...

Probably 1 out of 10, and of those I'm wondering how many can present proof that it's from their gpus and not from lag, latency, shitty config.
 
Lag and latency implies network, which is not the case here with stand alone benchmarks. Shitty config? Have you seen the benchmarks and notice where they are coming from? Professional review sites, including this one. It's so made up that ati redesigned their xfire bridge and drivers. Boy, we really fooled them. Prove microstutter doesn't exists. There is empirical data on microstutter existing with graphs as well as actual real world results. Now you can sit down.
 
micro stuttering only occours when you're playing at sub-50 fps. I have not noticed it when i went past 60fps.
 
I've had both 8800GT and GTX 280, both single and in dual SLI configurations for each. Both SLI setups microstutter to some degree. However, unless I'm trying to play Crysis at 8xAA 16xAF 1080p Very High, the GTX 280 SLI setup has more power and usually maintains better "worst case" framerates even with microstuttering, so it's tolerable for me, whereas SLI 8800GT would become extremely choppy.
 
I haven't run dual cards yet, but all the microstuttering talk convinced me to avoid SLI. That and the lack of multi-monitor support. For that kind of money its almost an insult to both remove features and introduce show-stopping bugs.
 
veedubfreak said:
Micirostutter is a term used by douchebags who want to find a reason not to use multiple gpus.
Click to expand...

Hey. I'm not a douchebag! I just can't rationalize buying two cards at a time when I will want a new model six months to a year from now anyway. I may be a cheap bastage (read smart) but I am no douchebag.
 
With no clear answer, I think I will err on the side of caution and stick with a single 4870 for now, thanks everyone for your input as I appreciate your opinions and information.
 
Mega666 said:
Lag and latency implies network, which is not the case here with stand alone benchmarks. Shitty config? Have you seen the benchmarks and notice where they are coming from? Professional review sites, including this one. It's so made up that ati redesigned their xfire bridge and drivers. Boy, we really fooled them. Prove microstutter doesn't exists. There is empirical data on microstutter existing with graphs as well as actual real world results. Now you can sit down.
Click to expand...

What are you 3? 'Now you can wah wah'

POST PROOF MORON.

If there's empirical data, like I said before, LET US SEE IT.

The only empirical data I'm seeing is someone's inability to read ENGLISH.
 
veedubfreak said:
Micirostutter is a term used by douchebags who want to find a reason not to use multiple gpus. That being said, i NEVER saw microstutter on my 8800GT sli, my 8800GTS sli, my 9800GTX sli or my current 4870 crossfire.
Click to expand...

QFT!
although 1 exception: the 3870x2 actually had microstuttering cause it had a crappy interconnect between the gpu's. and that one design flaw spawned paranoia through the minds of everyone contemplating building a rig with more than one graphics card.
 
veritas7 said:
Ive yet to see it on any of the games I play. I have a 9800 GX2. I enable vsync usually most of the time, mostly only if I am getting tearing. I have a 1920x1200 monitor too.

I think it's very overrated and like said, can be avoided by doing a few tweaks to configs or your settings, which is not too much of a deal.

Go forth and run CF 4870/4870x2/GTX 280s! Have fun! :p wish I had a mobo with two x16 pcie lanes :|
Click to expand...

http://www.youtube.com/watch?v=DYnXxI1UjxE

Looks pretty bad on a GX2 in comparison to a GTX280 or 4870 even though the GX2 is generating higher frames
 
I n-n-n-never her-her-heard of my-my-my-microstuttering before :)
 
Apallohadas said:
What are you 3? 'Now you can wah wah'

POST PROOF MORON.

If there's empirical data, like I said before, LET US SEE IT.

The only empirical data I'm seeing is someone's inability to read ENGLISH.
Click to expand...

This is probably the data he is talking about (there may be more elsewhere).

http://www.pcgameshardware.de/aid,6...enu=browser&image_id=864544&article_id=653475

http://www.pcgameshardware.de/aid,6...enu=browser&image_id=864545&article_id=653475

Now, I'd like to actually see similar graphs of single GPU options to compare it to. But this is the whole thing, it would be better to show time-per-frame, like these graphs, rather than frames-per-time which is what everyone talks in.

Look at those graphs, the first one, getting 45fps, which implies 22ms (milliseconds) between each frame. But what's ACTUALLY happening is one frame gets displayed in 13ms (eqv. 77fps), then the next frame is displayed in 37ms (eqv. 27fps). Because framerate is taken as an average over a given time, TECHNICALLY the game is still running at 45fps, however each individual frame alternates between 77fps and 27fps.

Now I'm not saying this is proof the X2 microstutters badly, because to see that you'd need to compare it directly to a single GPU card. However it does show the card is "microstuttering" (we just dont know if a single GPU would have the same amount of microstutter or more or less).

This is why I dont like "framerate" as a measure of how well a game runs. Example, Crysis on my PC at the settings I run plays at 20-30fps... however it "feels" smooth. Gears of War runs at 60+fps on my PC, however it is no where near smooth, even with things turned down there is still noticeable stutter, even though its still running at 60+fps. Its not about framerate, its about the delay from one frame to another. Same with COD4, it runs pretty smooth on my PC except when walking through smoke, then it drops to around 40fps, however it FEELS much lower than 40fps. If I didn't have fraps open to see it, I would have said it felt more like 10-20fps.
 
I miss voodoo.....best SLI ever imho. Sorry, but sad technology from 98 is still better implemented than today in 08. 3dfx's SLI was damn smmoooth and didn't micro stutter. Totally gay. $600-1000 on SLI and it stutters. This has put me off on it as well. Not saying it always does this or you can't work around it, but it's a pita.
 
SLIZone FAQ
How does this technology differ from 3dfx's SLI?
NVIDIA SLI differs in many ways. First, 3dfx SLI was implemented on a shared bus using PCI. The PCI bus delivered ~100MB/sec. of bus throughput, while PCI Express is a point-to-point interface that can deliver ~60x the total bandwidth of PCI. Second, 3dfx SLI performed interleaving of scan lines, and combined in the analog domain, which could result in image quality issues due to DAC differences and other factors. 3dfx Voodoo technology also only performed triangle setup, leaving the geometry workload for the CPU, hence 3dfx SLI only scaled simple texture fill rate, and then used inter-frame scalability. NVIDIA SLI technology is PCI Express based, uses a completely digital frame combining method that has no impact on image quality, can scale geometry performance, and supports a variety of scalability algorithms to best match the scalability method with application demands.
Click to expand...
 
Mega666 said:
http://www.hardforum.com/showthread.php?t=1317582

There are many more links but I have better things to do than to research for you.
Click to expand...



Those graphs show how deceptive the framerate values can be. Because FRAPS will take an average over a given time, it wont tell you how long its taking between each frame.

You can say "wow! that dual GPU card has 60fps while the single card is only 40fps!" When in reality the dual GPU card is EFFECTIVELY going at 40fps anyway.

Though again I'm not sure how much I trust that source :p It doesn't actually say what cards are being tested, what settings, etc.
 
RadXge said:
IMO, the microstuttering issue is over rated.
Same story with input lag.
Click to expand...



I agree its over rated as well, however I think people buying a card should be aware that the framerate values are an average over a given time, not an instantaneous thing. So with stuttering of any description (whether it be microstuttering or other types of stuttering) 60fps can easily "feel" like 30fps. You're buying an expensive card to get a higher framerate when in reality its not effectively going much faster than a single GPU option.

It doesn't make much difference at higher framerates, but when you end up eventually playing games with 20-30fps when in reality it feels like 10-15fps, that's when you really start to notice it.

At the end of the day the framerate shown by programs like FRAPS is secondry to having a smooth playing game. Regardless of whether you have a single or multi GPU system.
 
You must log in or register to reply here.
Back
Top