PlayStation Move input latency analyzed by Digital Foundry

And why isn't this article marked flamebait?...

Why do you seem to be looking for a fight so much recently, Wiizer? Also, your comment has little to do with what Cheesus said, so why reply to him? (Don't answer that second question, we already know why.)

I used to respect you Wiizer, but using my comment as a piggyback to get your comment on top? Have some decency!

OMG, my comment has a lot of relevancy:

"Isn't this like 4-5 months from release?"

I was answering Cheesus's question with another question!

My thoughts as well. Isn't the software that was demoed pre-alpha? They still have a ways to go and I am sure it will be fixed. This is indeed a flamaebait topic.

@WRE

Your comment didn't have anything to do with what Cheesus said either. Just saying.

@ Tigre

And also yours.

Topic: So the only issue is that Sony exaggerated a bit? Surprise!

I love Joystiq. (I also love, Wiizer, let there be no confusion.)

Cheesus ends in 's' so it would just be Cheesus' not Cheesus's
Just saying..

WRE is the BEST! ... And Cheesus's comment is IRRELEVANT IRREGARDLESS.


XD

The original article specifies that GTA IV actually runs with a lag of 160ms with a standard controller...so everyone should think about that. Don't know why that didn't make the Joystiq repost of this, seems like a pretty important distinction.

"This is the generally accepted definition of controller lag (certainly what's been used in our previous pieces on the subject, and that includes the 200ms pre-production Natal measurement)"

So I'm not really understanding what you meant. Are you saying Natal will be better or worse (which would cause either camp to fire flaming "LOL"s in the other direction)?

I think he means because most people (or at least, 360 fans), already have the preconceived notion that Natal is going to blow Move out of the water, so if it came out that Natal had more performance lag than Move....

....sparks would very much fly, from both sides.

Natal actually will have less lag, but the downside is for Natal to eat up more of the CPU than Move will.

The problem with a Natal vs Move battle is that Microsoft has lots of monies to push behind advertising, at least more than Sony. And I can tell you, with the Wii being so big of a hit, the "average consumer" will not notice the lag. Sure us "hardcore" gamers will, but I am fairly certain that neither Sony or Microsoft are planning on marketing this toward us.

Oh wait.

You're both wrong, he means that they'll (DF) carry out this study on Natal and the results will become the start of the great Fanboy wars of 2011.

I believe theyve already carried out this test on natal and it was somewhere between 133-200 (200 being the most quoted seeing as its the higher extremity). Although this was a few months back when they were allowing people to test riqochet.

Molyneux has already hyped the fact that its gotten much better, and even so 165 ms is not bad given the amount of points being recorded and translated by the software (full body i.e. its more than 2).

"Natal will have less lag"

"Most people think Natal will blow Move out of the water".

I doubt either of these are true.

http://www.newscientist.com/article/mg20527426.800-microsofts-bodysensing-buttonbusting-controller.htm

"He says Natal consumes just 10 to 15 per cent of the Xbox's computing resources and it can recognise any pose in just 10 milliseconds. It needs only 160 milliseconds to latch on to the body shape of a new user stepping in front of it."

i believe that comment is referring to how long it takes natal to recognize a new player in its view, not general input lag. an initial scan, if you will.

Yes it looks looks that, though I can't imagine it will be only 10ms normally..
I think PR was simply muddling the water and Natal will need to do the same scan for silhouette(s) in every analysed frame?

That would be in line with 100 or 200ms that was mentioned on other ocassions.

Anyway it is like MOVE in still unfinished stage and there are no concrete information.
Might change also in subsequent library revisons

@anonim:
The article that you linked to poses an interesting effect of that terabyte of motion data. Sony and Apple have movie production software suites and Microsoft could very well enter the movie production suite business through the use of such data on top of using data collected from Natal. If that's the case, that's a pretty ingenious use of Natal's technology right there though it might run into privacy lawsuits because of it.

Thanks for that new scientist article.

It was exactly what I was after..

"The result was a 50-megabyte software package that can recognise 31 different body parts in any video frame."

This 50 Megabyte software package they need to differentiate the body parts where do you think this will run ?

Remember they are trying to keep latency as low as possible, so for Natal to be usable it will need this software or database to be loaded into ram so as you move it will need to grab a snapshot of your movement compare to the data within this db and work out what you are doing..

What does this mean ? Well to simplify it, Natal will need 50MB of ram, no ifs buts maybes, the OS already uses 32MB so any Natal enabled game will have 82MB less ram to play with in total, which in turn will mean lower res textures, Less chance of proper AA rendering more than likely the screen res for most Natal games will be sub-hd.. Thats just the start..

The 15% CPU usage they mention is probably not including the OS usage and as far as I remember the xbox360 OS reserves a thread for itself..

The cpu has 3 cpu cores and 6 hardware threads (hyperthreading IBM style).. each thread works out to ~ 16.6% total processing power ..
They reckon it uses about 15% cpu time , along with the OS thats 2 threads reserved which leaves 4 threads for the actual game .. So we have the 360 actually running at a 1/3 of its original processing power..

Not to paint a grim picture but all I can see for the future of Natal this generation is a new way to operate the nxe, which will get very boring very quickly.. Also I see a tonne of shovelware.. Stuff like dancing games etc etc which would never tax the 360 in terms of memory or cpu.. To put it bluntly EyeToy games in HD

I could be completely wrong and the Natal really will usher in a new era of motion controlled games. I doubt it not this generation anyway, next generation quite possibly but by next generation Sony will only have to release a new eye that is capable of everything the Natal claims to do.. Infact if you check here:

http://www.newscientist.com/article/dn18115-sony-demos-game-controller-to-track-motion-and-emotion.html

It already has.. Its created a stereoscopic camera system which basically does everything Natal does.

http://www.atracsys.com/_products/icu.php -- This is the product Sony have created in collaboration with atracsys..

Does this begin to paint a picture for you ?

@scooge:

There is no reason Natal can't have onboard memory and some co-processing capability to take load off of the 360.

They may or may not do this. We won't know for a while.

And sony's stereo camera system is only the hardware. To do everything Natal does it will have to see people in low light, recognize faces, recognize gestures from tall people, short people, adults, children, the skinny and the fat and everything in between, all with different limb ratios and everything.

Sony is no where near that.

Statistics: The art of making numbers say what you want them to say. Nice job on changing 15% to 66%, but I dont really think anyone is buying.

Actually, scratch that, im sure tige/enrique/el serp are on their way here to downvote me.

I suppose I should explain.

For starters, you are assuming that a virtual thread contains just as much processing power as a physical thread. This is incorrect. Theortically speaking I suppose you would be correct, but in practice a "virtual" CPU (which is what hyperthreading creates) lends about a 50% performance increase. So, if I had a 4ghz HT'd CPU, it would be about as fast as a 6ghz non HT'd one (ignoring the obvious instability of such a high clock). So your math about the threads is completly wrong.

Im sure that the "original processing power" should include the fact that the Xbox 360 OS has always been there and isnt suddenly, magically taking up a thread now that Natal is out, so thats out too. So now we are looking at a total of one (1) extra thread being used by Natal.

So now we are lookign at the 3 physical cores, and the 3 virtual cores. The power split should be about 75/25 respectively. If Natal needs about 15%, the thing to do is dedicate probably 2 virtual threads to it, which equates to 16 and 2/3's % of the 360s total power. Which is a lot different than your equation of 66% of its total power.

Even then we are operating under the assumption that they will need to dedicate specific threads to Natal. Im a tech, not a programmer, so I dont know how it would work, but I would say thats a pretty large assumption to make.

Regardless of shovelware or not, I would totally enjoy the ability to play Parapara Paradise on either console.

@Ordeith .. If only, they removed the co-processor from the Natal to save money, if the Natal did all of the processing itself, all would be good. This library they talk about has to run from memory if it runs from another source it has to then be transferred to actual system ram before being sent to cpu cache to process. (Adds a lot more latency)

@R , Sorry. I did it again, it was a typo what I meant to say is:

"So we have the 360 actually running at 2/3 of its original processing power"

or alternately:

"So we have the 360 actually running at 1/3 less than its original processing power"

I am changing 15% to equal a 1/3 of the processing ie 33% (including the system os), or 1 total cpu if the 360 os actually reserves a cpu thread for itself, which I read previously but have nothing to back up my statement, I might be completely wrong here.

I really dont like statistics, I had to use percentage because Microsoft quote in percent rather than actual cpu time.

IBM hyperthreading is actually a lot better than Intels. IBM has been using threading for years and Intel have only recently started using it. Also the cpus are in-order which should make the threading even better as the cpus will probably be blocked a lot.

As far as the OS is concerned it sees 6 virtual cpus , the actual threading is taken care of in hardware , so the OS will have no control over it.

Using your calculations the results become a lot worse.. You are right there is an uneven balance between actual physical core to virtual core though, I agree.

I am not disagreeing or agreeing with the statement "the 360 OS takes up 1 thread" What I am saying is that it always takes up a thread. By your calculations, the 360's max power is not 100%, its 84.3%, of which you minus yet another thread to get your 66% figure.

Either way you put it, its the kind of math you would see at a used car dealership located on Shady Cr. and Scrooge Ave.

@R
Im trying to figure out what will be available to the game / game developer, so the OS has to be taken into account because the OS is already using resources which is why I add the two together, its far from perfect its about as flawed as the digital foundry analysis of the ps move latency, to be blunt.

Your own calculation has Natal using 2 threads anyway :)

Do the eight frames even matter? The only games where it would be an issue are fighters and shooters, and if you're playing either of them with Move, then God help you.

@Trey

Don't even act like you wouldn't get a quarter-chub from doing an actual hadouken.

@Churchbus - I know I would... maybe even 1/3

Hopefully it's not 1 to 1... or my tiger uppercut is gonna suck.

Or my Dhalsim stretch punch sweeps are going to go nowhere

it's both!.. at the same time.

@solarshadows074:

How?

If anything Natal can be both, as it can be used with and without a controller. But the PS Move cannot function without a glowing ball to look at. So no, it can never be Natal.

@Ordeith
you see, the camera(eyetoy) itself is LIKE natal. hell, natal is based on the original eyetoy. so technically, it would function without the glowing ball.

I doubt any games would use both at the same time.

I doubt it too, but it just seems really tedious. Maybe a game in the near future uses the up button as sprint, and you have to use the analog stick to move. I doubt a game will actually do that, but you get my point.

It's been mentioned that the Sub-controller will be optional, and that one can simply hold a DualShock 3 in their left hand to complement the Move in their right.

@Dr. Blight, SOCOM 4 was demoed using the Sub-controller.

Tesseract, I'm talking about using the D-Pad AND Analog Stick at the same time. Not the two controllers.

Developers aren't stupid enough to require you to push anatomically impossible button combos. The left analogue stick is on the same side as the D-pad on both the Dualshock controller and the 360 pad and it's never been an issue.

Exactly. The video is being captured, analyzed, then double output to the monitor. Some of that lag is due to the merging of the actual video feed with the overlaid graphic. This should be optimized on the software end before release. A more accurate measurement would be to demo a game, assuming the Move's firmware is already up to par. Of course, there is nothing to test on at this time. That's the only way a poor analysis like this actually makes news.