Microstuttering masterclass

[qubit]

All you ever wanted to know and then some:



Tom's Hardware

[W1zzard]

added and voted: I'm using a single GPU, so don't care

and what kind of wtf graph is that at thg? a histogram is what is used for data like this

[qubit]

W1zz, I didn't ask for that extra option in there, it's just a copout and spoils my poll - now everyone's voting for that, instead of what I intended.

Can you please remove it?

Look at that diagram in the context of the article and it makes perfect sense.

[Scheich]

I didnt knew 2x CF sucks so hard. I guess i will sell my pile of 5850 now entirely. What a pity

[WhiteLotus]

I have no idea what micro stutter is, hurray for single cards!

[TheMailMan78]

Originally Posted by W1zzard

added and voted: I'm using a single GPU, so don't care

and what kind of wtf graph is that at thg? a histogram is what is used for data like this

Owner OWN3D!

[AthlonX2]

Originally Posted by brandonwh64

he OWNS the site

/END THREAD

+1

[Black Panther]

Please keep the discussion on topic or the thread would have to be closed.

[Spaceman Spiff]

It drove me insane when I was using 2x 4890's until I found a glorious program called D3DOverrider which cured all my CF problems with Microstutter. I've tried to promote it but people won't listen/try it so oh well.

I'll be getting a 2nd 6970 eventually.

[Sinzia]

I've never really noticed it, and I've been running SLi since then 7800 GT days.

[qubit]

Originally Posted by Black Panther

Please keep the discussion on topic or the thread would have to be closed.

Yeah, close it, I'm all for it.

I see that my second to post to w1zz has been disappeared (I have a copy) and none of my concerns have been addressed and I've been roundly ignored by management and had the piss taken out of me by other posters on here - and the posts left up. Great way to piss off your members.

Also, MM's kind second post to me has been disappeared too.

MM, I did appreciate it!

[W1zzard]

i'm sorry for editing your poll, i never intended to mess it up in any way. i honestly thought you didnt think of this option when creating the poll.

i think the added option is important, it doesnt seem to make the other votes invalid, does it ?

[qubit]

Originally Posted by W1zzard

i'm sorry for editing your poll, i never intended to mess it up in any way. i honestly thought you didnt think of this option when creating the poll.

i think the added option is important, it doesnt seem to make the other votes invalid, does it ?

Oh alright then, you're forgiven. Thanks very much w1zz, I really really appreciate your reply. (qubit turns into a big softie, hehe).

Heck, so many people have now voted, it's probably best to leave it there now.

Microstuttering is such a hot topic for mult-GPU users, that I feel this Tom's article would be of great interest to my fellow enthusiasts on here.

EDIT

And I apologize for ranting!

[scaminatrix]

IMO the last option stops people with single card setups (or people who don't know what micro-stutter is) from voting "Doesn't bother me at all "
If everyone with a single GPU setup voted Doesn't bother me at all, that would really mess the results up.
You've got to account for stupidity

[HossHuge]

Originally Posted by Spaceman Spiff

It drove me insane when I was using 2x 4890's until I found a glorious program called D3DOverrider which cured all my CF problems with Microstutter. I've tried to promote it but people won't listen/try it so oh well.

I'll be getting a 2nd 6970 eventually.

Thanks your post lead me to read about triple-buffering. I had never thought about checking that box in CCC before. I'm playing Tom Clancy's Splinter Cell Conviction ATM and it's running much smoother now.

Cheers..

[qubit]

Originally Posted by HossHuge

Thanks your post lead me to read about triple-buffering. I had never thought about checking that box in CCC before. I'm playing Tom Clancy's Splinter Cell Conviction ATM and it's running much smoother now.

Cheers..

Oh yeah, triple buffering is a great function for getting the best out of vsync in terms of smooth animation and reducing input lag. Ars Technica had an in-depth article on it a while back that was posted about on TPU.

[arnoo1]

Only ati/amd has micro stuttering, and i'm never going to use amd so i will never have any problems, and i'm currently runningen single gpu solution(nvidia)

[GotNoRice]

I would love to see the source data for that pie graph, or exactly what sort of "analysis" they performed to come up with that. It looks more like something the author simply pulled out of his ass.

I don't really notice any microstutter on my 2x 4870x2 setup and it's just as smooth as any single-GPU system I've ever seen. Then again their results don't show much microstutter on triple or quad GPU setups either.

[W1zzard]

Originally Posted by GotNoRice

I would love to see the source data for that pie graph, or exactly what sort of "analysis" they performed to come up with that. It looks more like something the author simply pulled out of his ass.

without reading the article i would guess that the author used fraps to log the frametimes during rendering (that's how i would do it).

[cadaveca]

Originally Posted by W1zzard

fraps to log the frametimes during rendering

Does using FRAPS mean measuring the time it takes a frame to end up in the final buffer, or how the frames are output from said buffer? I see a potential fault in this method of testing...


I used to be a big proponent lauding the existence of Microstutter, when many were still denying it exists. I'm very happy to say that I haven't seen much of this particular problem in some time, but that could be due to me properly chosing system configs to avoid the problem?

[erocker]

Where's the poll option for those who use CrossFire and don't get microstuttering? Or, at least an I don't care because I find this poll unbalanced option?

[qubit]

Originally Posted by GotNoRice

I would love to see the source data for that pie graph, or exactly what sort of "analysis" they performed to come up with that. It looks more like something the author simply pulled out of his ass.

Brilliant, I love the way you phrased that. I've only read a bit of the article myself at the moment, so can't answer the question right now.

Originally Posted by GotNoRice

I don't really notice any microstutter on my 2x 4870x2 setup and it's just as smooth as any single-GPU system I've ever seen. Then again their results don't show much microstutter on triple or quad GPU setups either.

You won't notice microstutter if your rig can render without any dropped frames and vsync is on, regardless of the refresh rate of your monitor.

From what I can tell, you get microstuttering with single cards too, when the system can't maintain the framerate. Basically, you get an uneven, jerky mess, as each frame takes a different time to render as the scene changes. Perhaps the exact way it looks on screen is subtly different, but I'm not sure.

Originally Posted by erocker

Where's the poll option for those who use CrossFire and don't get microstuttering? Or, at least an I don't care because I find this poll unbalanced option?

The article is graphics card brand agnostic, hence the poll doesn't refer to particular brands. Anyway, the microstutter problem is only there when the system can't keep up with the monitor refresh, as I explained above.

[erocker]

Originally Posted by qubit

The article is graphics card brand agnostic, hence the poll doesn't refer to particular brands. Anyway, the microstutter problem is only there when the system can't keep up with the monitor refresh, as I explained above.

Brands? I didn't mention brands. If a system can't keep up with the monitor refresh, there would be "microstutter" with any GPU configuration. The real shame is that this is based off of a Tom's Hardware article. I try to avoid Tom's and it's a shame that I have to find it here.

[crazyeyesreaper]

i dont care it dosent bother me and heres why i dont experience it

back in the days of the DX9 cards nvidia 7000 series AMD x1000 series yea it was a bitch

it was also a pain in the ass with my single 4870x2 that i had, but, with my 5850s and 6970s ive had 0 microstutter and no issues to speak of in that regard im entirely happy with the performance 2 gpus give and microstutter if there is completely undetectable to me, and considering even a varience in PING in a multiplayer game pisses me off and makes me rage thats saying alot. i can feel the difference frame rate makes and the difference ping makes to the point 20-30 drop on my PING in BC2 pushed my avg k/d up from 1.4 to 1.6 per round average since the Intel NIC was better, now considering i can feel and benefit from that tiny difference and ive seen micro stuttering before, i would honestly have to say Toms hardware is fudging the hell out of those tests.

as i have the games there testing and i notice no issues at all, its smooth as can be, with no frame drop.

[qubit]

Originally Posted by erocker

Brands? I didn't mention brands. If a system can't keep up with the monitor refresh, there would be "microstutter" with any GPU configuration. The real shame is that this is based off of a Tom's Hardware article. I try to avoid Tom's and it's a shame that I have to find it here.

You said:

Originally Posted by erocker

Where's the poll option for those who use CrossFire and don't get microstuttering?

hence the brand association.