Microstutter: The Poll

[BD231]

You're obviously not playing with dual cards so of course there's no micro-stutter. Poor frame times ... perhaps.

 

[Hitman928]

New video. This corresponds to the A videos earlier. I changed some things around including enabling mouse acceleration. I could never actually play with it on, but it does smooth out my mouse movements. I am still going to pick another game as this one has a lot of jerkiness in the camera and animations but them me know what you guys think so I can hopefully keep improving the process.

Better? Worse? The same?
https://docs.google.com/open?id=0B11hajGAJWuVUnhRVlBaZVpkVWc

 

[AdamK47]

Not to mention that FRAPS records things in the middle of the graphics chain, not at the end like a high-speed video camera would. So all this frametime talk is useless without a high speed camera to back it up.

More info here about why FRAPS is insufficient:

http://techreport.com/review/21516/inside-the-second-a-new-look-at-game-benchmarking/11

That's what I was thinking.

 

[njdevilsfan87]

http://techreport.com/news/21625/multi-gpu-micro-stuttering-captured-on-video

0:27 - 0:48 is all you need to see.
http://techreport.com/news/21625/multi-gpu-micro-stuttering-captured-on-video

 

[AdamK47]

Also, when you record with FRAPS, isn't it going to place one frame right after the other in the video file regardless of microstutter? I would think none of the microstutter is going to be reproducable during playback.

 

[Hitman928]

Also, when you record with FRAPS, isn't it going to place one frame right after the other in the video file regardless of microstutter? I would think none of the microstutter is going to be reproducable during playback.

This is true, fraps tops out at 60 fps, but then so do most people's monitors. So, if fraps can't catch it, will your monitor ever display it? I need to look into how fraps does recordings a bit more.

 

[lavaheadache]

D was by far the worst. There was definately microstutter in the d's. If you filmed more of just running down a hallway it would have been very easy to see.


A=3's
B=4's
C=2's
D=1's

Final answer Chuck


change it to

A=2
B=4
C=1
D=1

A & C are very similar in graph and video that it could really be either one but I'm dead set on b=4 and d=1

 

[Hitman928]

D was by far the worst. There was definately microstutter in the d's. If you filmed more of just running down a hallway it would have been very easy to see.



d2=1.1

d1=3.1

Just a little clarification, each frame time is in a pair. In other words, 1-1 goes with X1 and 1-2 goes with X2, 2-1 and Y2 are a pair, etc. So, d1 could be 3-1 but d2 couldn't be 1-1. Thanks for the input though, I'm interested to see how this plays out. Should have a different (hopefully better) example tomorrow.

 

[Ferzerp]

AVI at full speed is pretty much useless for this. In fact, it doesn't have the capability at all. A 60 fps AVI can have repeated frames, or null frames, but no variable framerates (though I am not so sure that fraps will even do the tricky encoding required for that).

So, pretty much your options with a 60 fps AVI (*if* encoded to do it) are

16.66666 ms frames, 33.33333ms frames (same frame repeated), 50 ms frames (same frame 3 times) and so on.

The capability just isn't there with the file format you've chosen to do what you're trying to do.

edit: this is also sort of why those other videos were shot via camera at high speed and then slowed down extremely slow. That's really the only way to convey the effect via most video formats (especially 30 fps youtube, but 60 fps avi is really not better)

 

[lavaheadache]

Those videos were fine. I really think I may have nailed them down. A and C were the tough ones because the the graphs were similar and so were the videos. It would have been easier to tell if it was a scene just running down a hallway

 

[Avalon]

They all seemed a little jerky at parts to me, idk...could just be the game. I've never personally played it.

 

[Ferzerp]

They all seemed a little jerky at parts to me, idk...could just be the game. I've never personally played it.

Probably the 33.333 ms frames and/or 50ms frames.

There is no way to convey the frametimes in the graphs he's posted with an AVI unless the AVI is something like 1000 fps (and repeats the same frame 16-17 times for a 16.666 ms frame, etc).

 

[Hitman928]

Probably the 33.333 ms frames and/or 50ms frames.

There is no way to convey the frametimes in the graphs he's posted with an AVI unless the AVI is something like 1000 fps (and repeats the same frame 16-17 times for a 16.666 ms frame, etc).

I'm still looking into how fraps does it, but so far I've been pretty impressed by how well it conveys the actual play experience. For instance, I did this run in farcry 2 earlier:

The one hitch you see in the frame times was definitely a stutter in the game (fps count never showed it though) and the fraps video shows it perfectly clear as well. I'm not going to upload the video because they're just too big and I think it will show in what I will put together tomorrow. Fraps video is definitely catching things though at least up to 60 fps and possibly more, I'm still looking into it.

edit: here's the fps graph for the frame times above:

 

[AnandThenMan]

I think to make this test valid, you should reveal in confidence to someone here which option represents which hardware.

 

[Hitman928]

I think to make this test valid, you should reveal in confidence to someone here which option represents which hardware.

Sure, any recommendations?

edit: Just because it really doesn't matter for the goal of the experiment, each run was done with the same hardware, just different variables being played around with. Didn't want to start anything from inconclusive findings seeing as how people love to run with things around here. With that said, I can tell someone all the conditions of the test, just to make sure I'm being honest

 

[Ben90]

Obviously IDC, he is the king of the land here.

Redacted my absolutely horrific profanity that caused such a tangent.

 

[Ferzerp]

The one hitch you see in the frame times was definitely a stutter in the game (fps count never showed it though) and the fraps video shows it perfectly clear as well. I'm not going to upload the video because they're just too big and I think it will show in what I will put together tomorrow. Fraps video is definitely catching things though at least up to 60 fps and possibly more, I'm still looking into it. edit: here's the fps graph for the frame times above:

As I consider it more, this is correct. It's the same as your screen does it. Every 16.666ms it tries to display a new frame. It may end up with the same one, or it may get a blend of two, or it may get an all new one. It too is locked in to 16.666 ms, just like a 60 fps avi.

 

[BrightCandle]

This method of capture can work. Its taking the video at the same moment that the frame time would end so if we can see microstutter then it would validate that frame times. But the problem I can see is that you can't just induce microstutter and then fix it by messing with settings, at least not in my experience. You don't seem to have heavy microstutter to begin with in any trace. The focus on the the very slow frames is not the point of this discussion on tech reports reviews' they just happen to be the part of the video everyone can see.

 

[3DVagabond]

I consider having a passage from a religious book for your signature line to be disrespectful. This is an international forum and is not frequented exclusively by Christians. I agree he or she should be more respectful of other peoples cultures and beliefs, so should you. It is hypocritical of someone with a religious message in his/her sig to moan about being disrespectful IMHO.

Please don't say "this is an American site", there is no such thing when it comes to an internet site that is aimed at an international audience.

Enough off topic though.

Good thing he's Christian. All he's done is politely asked for someone not to take the name of Jesus in vain. No one's being blown up for it. What does it being an American site have to do with anything. America's historically the most religiously diverse and tolerant country in the world. Check your own bigotry before you start lecturing others.

 

[ICDP]

Good thing he's Christian. All he's done is politely asked for someone not to take the name of Jesus in vain. No one's being blown up for it. What does it being an American site have to do with anything. America's historically the most religiously diverse and tolerant country in the world. Check your own bigotry before you start lecturing others.

I find taking anyone's God(s) name in vain disrespectful on an international forum. I also find people having a religioius message in their sigs to be disrespectful. I was merely pointing out the irony in demanding people be respectful of their religious views, while not seeing that posting religious messages may be disrespectful to those of other beliefs.

The world is a place full of diverse religious beliefs. IMHO you can't demand people respect your religious beliefs while assuming your own is sancrosanct. At least not on an international site.

 

[videogames101]

I find taking anyone's God(s) name in vain disrespectful on an international forum. I also find people having a religioius message in their sigs to be disrespectful. I was merely pointing out the irony in demanding people be respectful of their religious views, while not seeing that posting religious messages may be disrespectful to those of other beliefs.

The world is a place full of diverse religious beliefs. IMHO you can't demand people respect your religious beliefs while assuming your own is sancrosanct. At least not on an international site.

You can't respect anyone's religion, because I can make one up tomorrow with whatever tenants I want. It's an impossible stance to hold.

But on topic! I couldn't tell any difference on my 60hz laptop, but A (and every other example, but A in paticular) was terrible on my desktop 120hz monitor.

 

[Ben90]

Good thing he's Christian. All he's done is politely asked for someone not to take the name of Jesus in vain. No one's being blown up for it.

Please don't disrespect the Muslim religion like that. Just because they have a tendency to blow things up doesn't mean they are all going to Jihad over the interwebs.

 

[blastingcap]

*ahem* Can we please get back on topic.

 

[Deders]

Unfortunately the general stuttering we see here isn't really microstutter, probably more down to disk access. Don't suppose you have an alternate drive you could record the video onto that doesn't have the game or pagefile on it?

 

[Rikard]

Could you please add units to the vertical axis? One cannot tell how large the micro stuttering and the spikes are otherwise. I think I can see the frame time spikes, but not the micro stuttering tbh, and D looked worst while the others were pretty much the same.